UvA-DARE (Digital Academic Repository) European ... · of paediatric RRT programs in the 1960s, ESRD in children was a death sentence. Since then, substantial advances in renal medicine

UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)

UvA-DARE (Digital Academic Repository)

European disparities in the incidence and outcomes of children with end-stage renal disease

Chesnaye, N.C.

Link to publication

Creative Commons License (see https://creativecommons.org/use-remix/cc-licenses):Other

Citation for published version (APA):Chesnaye, N. C. (2017). European disparities in the incidence and outcomes of children with end-stage renaldisease.

General rightsIt is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulationsIf you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, statingyour reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Askthe Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,The Netherlands. You will be contacted as soon as possible.

Download date: 27 Nov 2020

https://dare.uva.nl/personal/pure/en/publications/european-disparities-in-the-incidence-and-outcomes-of-children-with-endstage-renal-disease(1f827658-e308-4d0c-a5db-33224d330948).html

EUROPEAN DISPARITIES IN THE INCIDENCE AND OUTCOMES OF CHILDREN WITH END STAGE RENAL DISEASE

—Nicholas Chesnaye

European disparities in the incidence

and outcomes of children with

end-stage renal disease

Nicholas Christopher Chesnaye

Nicholas Christopher Chesnaye, Amsterdam, The Netherlands Printed by PrintQuest Cover design by Deimion van der Sloot All rights reserved. No part of this thesis may be reproduced, stored or transmitted in any form or by any means without prior written permission of the author. Financial support by the Dutch Kidney Foundation for the publication of this thesis is gratefully acknowledged.

EUROPEAN DISPARITIES IN THE INCIDENCE

AND OUTCOMES OF CHILDREN WITH

END-STAGE RENAL DISEASE

ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad van doctor

aan de Universiteit van Amsterdam

op gezag van de Rector Magnificus

prof. dr. ir. K.I.J. Maex

ten overstaan van een door het College voor Promoties ingestelde commissie,

in het openbaar te verdedigen in de Aula der Universiteit

op vrijdag 1 december 2017, te 13:00 uur

door Nicholas Christopher Chesnaye

geboren te Singapore

Promotiecommissie

Promotor: Prof. dr. K.J. Jager, AMC-UvA

Copromotor: Dr. K.J van Stralen, Spaarne Gasthuis

Overige leden: Prof. dr. J.B. van Goudoever, AMC-UvA

Prof. dr. A.E. Kunst, AMC-UvA

Prof. dr. E. Levtchenko, Katholieke Universiteit Leuven

Prof. dr. F.W. Dekker, Universiteit Leiden

Dr. V.S. Stel, AMC-UvA

Dr. K. Cransberg, Erasmus MC

Faculteit der Geneeskunde

CONTENTS

Chapter 1 General Introduction 7

Chapter 2 Demographics of Paediatric Renal Replacement Therapy in Europe: a Report of the ESPN/ERA–EDTA Registry

13

Chapter 3 Disparities in Treatment Rates of Paediatric End-Stage Renal Disease across Europe: Insights from the ESPN/ERA-EDTA Registry

25

Chapter 4 Mortality Risk Disparities in Children with End-Stage Renal Disease across Europe - An ESPN-ERA/EDTA Registry Analysis

51

Chapter 5 Survival in Children Requiring Chronic Renal Replacement Therapy

75

Chapter 6 Mortality Risk in European Children with End-Stage Renal Disease on Dialysis

91

Chapter 7 Comparison of Clinical Outcomes in Infants on Chronic Peritoneal Dialysis and Hemodialysis

113

Chapter 8 The Association of Donor and Recipient Age with Graft Survival in Pediatric Renal Transplant Recipients - an ESPN/ERA-EDTA Registry Study

129

Chapter 9 Discussion 149

Summary 167

Acknowledgements 173

Cirriculum Vitae & Portfolio 179

References 185

1

General Introduction

Introduction

8

END-STAGE RENAL DISEASE

Our kidneys are tasked with the vital responsibilities of filtering waste products from the

blood, producing urine, regulating blood pressure, erythrocyte production, and controlling

calcium, phosphate, and magnesium metabolism. Chronic kidney disease (CKD) is a general

term for a diversity of disorders affecting the kidney and is characterized by a deterioration of

kidney function over time. The final stage of CKD is end-stage renal disease (ESRD), a

devastating condition that is associated with considerable morbidity, mortality, and a poor

quality of life [1, 2]. In adults, ESRD is a leading cause of morbidity and mortality worldwide,

with an estimated prevalence of 2.6 million patients receiving treatment in 2010, and a

projected doubling of this number by 2030 [3]. In children, ESRD is considered a rare and

complex condition caused by a variety of aetiologies, constituting approximately 1-2% of the

total ESRD population. Compared to adults, where diabetes and hypertension form the leading

causes of ESRD, the majority of paediatric patients suffer from a multiplicity of congenital

anomalies, hereditary nephropathies, and glomerular defects [4, 5].

RENAL REPLACEMENT THERAPY IN CHILDREN

ESRD necessitates chronic renal replacement therapy (RRT) to sustain life. Prior to the start

of paediatric RRT programs in the 1960s, ESRD in children was a death sentence. Since then,

substantial advances in renal medicine have been achieved. The treatment modality choices for

RRT consist of peritoneal dialysis (PD), haemodialysis (HD), and renal transplantation (Tx).

Although the latter is considered the optimal modality choice with regard to patient survival,

cognitive development, quality of life, and growth [6–10], approximately three-quarters of

patients will initiate RRT on dialysis to bridge the preparation time needed for transplantation

or will require dialysis after graft loss [11]. The provision of RRT to children is more

expensive compared to adults, as children are ideally treated by an extensive and specialized

multidisciplinary team, and often require expensive medications such as growth hormone. A

Swiss study estimated the annual costs of approximately $200,000 for paediatric dialysis,

$300,000 for the total cost of all care during the first year of a paediatric kidney transplant,

and about $75,000 for each following year [12, 13].

Introduction

9

THE ESPN/ERA-EDTA REGISTRY

Up-to-date, accurate, and detailed epidemiologic data regarding the paediatric ESRD

population is vital for evidence-based policymaking and for informing patients, physicians, and

healthcare providers. Accurate data on the demographics, such as on the number of existing

(prevalence) and new cases (incidence) of children on RRT in Europe, are scarce. In addition,

national or single-centre studies are often unable to provide sufficient statistical power to

accurately assess treatment outcomes. Fortunately, over the past decades, various

(inter)national registries have collected sufficient data to power statistically valid and clinically

meaningful studies, and have been instrumental in advancing epidemiologic research and

expanding the growing evidence base regarding treatment outcomes in this population. In

2007, the European Society for Pediatric Nephrology / European Renal Association –

European Dialysis and Transplant Association (ESPN/ERA-EDTA) Registry was established to

consolidate data collected by population-based European national renal registries on children

with ESRD treated with RRT. Currently, the registry collects data annually from 36 European

countries and holds records on over 10,000 paediatric patients. Given the large number of

data, we are now able to accurately assess the recent epidemiology of paediatric RRT in

Europe, which we will present in chapter 2.

EUROPEAN DISPARITIES REGARDING TREATMENT AND MORTALITY

RATES IN THE PAEDIATRIC RRT POPULATION

One of the main focus points of the European health policy framework, Health 2020, is to

significantly reduce health inequalities and ensure universal, equitable, and high quality health

services across Europe [14]. Although all European Member States have made commitments

towards this goal, considerable international disparities in treatment- and mortality rates have

been described in the adult RRT population. Disparities in treatment rates have been

attributed to the percentage of elderly, the prevalence of diabetes and hypertension in the

general population [15–17], and factors affecting access to care [18, 19], whereas disparities in

mortality rates have been explained by differences in country macroeconomics, general

population mortality rates, patient demographics, the distribution of cause of renal disease, the

quality of renal care, access to treatment, and attitudes regarding acceptance to and

withdrawal from treatment [20–22].

Introduction

10

It is difficult to extrapolate findings from the adult to the paediatric RRT population due to

differing primary renal diseases, the rarity of paediatric ESRD, and the high costs involved

treating children in mostly academic settings by a multidisciplinary team. Apart from a single

non-population-based study which studied the influence of gross national income on paediatric

PD prevalence and mortality rates [23], little is known about the extent of disparities in the

European paediatric RRT population, and their underlying factors remain unknown.

As nearly all cases of paediatric ESRD consist of rare disorders of at least some genetic origin,

we postulate that differences in treatment rates across Europe could partly be explained by

geographical differences in genetic background [24]. On the other hand, non-medical country-

level factors, such as macroeconomics, are likely to affect access to care and may also play a

role in explaining the variation in paediatric RRT rates [23, 25]. In chapter 3, we therefore

assess the factors that determine the incidence of RRT by exploring how much of the variance

can be attributed to genetic factors, and how much can be attributed to disparities in access to

renal care.

Similarly, European disparities regarding mortality rates in the paediatric RRT population may

be explained by both patient-level factors, such as renal disease distribution, and by country-

level factors, such as the number of paediatric treatment centres. In chapter 4, we aim to

determine the magnitude of variation in country mortality rates, and disentangle both patient-

and country-level factors to understand which mechanisms may be responsible for these

geographical disparities.

SURVIVAL IN THE PAEDIATRIC RRT POPULATION

Although other patient-related outcomes such as growth and quality-of-life are crucial,

prolongation of patient survival may be arguably the most relevant clinical goal. Mortality in the

paediatric RRT population is multifactorial, owing to the complex nature and diversity of

ESRD. In chapter 5, we touch on several factors which have been shown to affect the

mortality risk in the paediatric RRT population, including age at RRT initiation, time on RRT,

primary renal disease (PRD), the presence of comorbidities, and initial treatment modality

[26].

Introduction

11

Transplantation is considered the optimal modality choice with regard to patient survival,

however, most patients will initiate RRT on dialysis to bridge the preparation time needed for

transplantation [11]. In the adult dialysis population, a large number of observational studies

have investigated survival differences between patients starting on HD and PD. Although

comparisons between studies are hampered by differences in case-mix adjustments and

analysis techniques, in Western countries there seems to be a consistent trend showing a

survival advantage for patients initiating dialysis on PD during the initial years on dialysis, and in

younger, healthier, and non-diabetic patients [27–32].

In children, the few studies that have explored the effect of dialysis modality on mortality risk

show conflicting results [8, 33–35]. In Europe, no such study has previously been undertaken

on an international scale, and the rarity of paediatric ESRD has limited exploration of the

heterogeneity of treatment effects across patient subgroups and time-dependent treatment

effects, as have been demonstrated in the adult population. Therefore, in chapter 6, we

describe the mortality risk in the paediatric dialysis population, and compare the mortality risk

between patients starting RRT on haemodialysis and peritoneal dialysis. Furthermore, as it is

generally believed that in infants HD should only be reserved for cases where PD is not

feasible, we will answer the same questions focusing specifically on the infant dialysis

population in chapter 7.

GRAFT FAILURE RISK IN DIFFERENT DONOR AND

RECIPIENT AGE COMBINATIONS

It has been well established that renal transplantation offers better patient outcomes

compared to dialysis [6–9]. Nonetheless, 10 years after transplantation, approximately 40% of

paediatric transplant recipients will have lost their graft [10]. Moreover, returning to dialysis

after graft failure has been associated with a 4.4-fold increase in mortality risk [36]. Recipient

and donor age are amongst the many factors that influence graft survival. In most European

countries, a deceased donor-recipient ‘young-for-young’ matching policy has been

implemented, where young donor grafts are preferentially allocated to children [37–44]. These

donor-allocation policies aim to reduce waiting times and provide high-quality grafts to the

best-matched recipients in order to improve graft survival. However, earlier reports have

shown a higher risk of graft loss in recipients of the youngest donors due to surgical

Introduction

12

complications, high rates of graft thrombosis, early rejection, and hyperfiltration injury [39, 40,

45–47]. Furthermore, although it is known that living donation offers better long-term graft

survival compared with deceased donation [11, 48, 49], it remains unknown whether utilizing

kidneys from elderly living donors, should be preferred over kidneys from age-matched

deceased donors. As it remains unclear which organs should be ideally allocated to children,

donor-allocation policies continue to differ between countries, hampering equal access to

renal transplantation for children across Europe [42]. Consequently, to help optimize kidney

donor allocation policies, in chapter 8 we examine how the relationship between donor age

and recipient age affects graft survival in paediatric kidney transplant recipients.

AIM

This thesis aims to reveal health inequalities and improve outcomes in the European paediatric

RRT population by determining the epidemiology of the paediatric RRT population across

Europe, exposing international disparities in treatment rates and mortality risk in this

population, and investigating factors that may explain these differences. Lastly, this thesis aims

to help define the optimal donor kidney allocation policy by examining the relationship

between donor age and graft loss.

Demographics

2

Demographics of Paediatric

Renal Replacement Therapy

in Europe: a Report of the

ESPN/ERA-EDTA Registry

Nicholas C Chesnaye, Marjolein Bonthuis, Franz Schaefer, Jaap

W Groothoff, Enrico Verrina, James G Heaf, Augustina

Jankauskiene, Viktorija Lukosiene, Elena A Molchanova, Conceicao

Mota, Amira Peco-Antić, Ilse-Maria Ratsch, Anna Bjerre, Dimitar

L Roussinov, Alexander Sukalo, Rezan Topaloglu, Koen van

Hoeck, Ilona Zagozdzon, Kitty J Jager, Karlijn J van Stralen

Pediatr Nephrol 2014 Dec; 29(12): 2403–2410

ABSTRACT

Background: The ESPN/ERA-EDTA Registry provides data on European children with end-

stage renal disease receiving renal replacement therapy (RRT). This paper provides the results

of the demographic data collected from 2009 to 2011.

Methods: Data on paediatric RRT patients were extracted from the ESPN/ERA-EDTA

registry for 37 European countries regarding primary renal disease, incidence, prevalence, 4-

year survival, transplantation rate, and causes of death.

Results: The incidence of paediatric RRT in Europe was 5.5 cases per million age related

population (pmarp) in patients aged 0-14 years, and varied markedly between countries (IQR

3.4 – 7.0 pmarp). RRT prevalence was 27.9 pmarp and increased with age, with 67% of

prevalent patients living with a functioning graft. The probability of receiving a transplant within

4 years was 76.9%, and was lowest in patients aged 0-4 years (68.9%). Mortality in paediatric

RRT patients was 55 times higher than that of the general EU paediatric population. Overall

survival at 4 years was 93.7%, with the poorest survival in patients aged 0-4 years and in

patients starting on dialysis. Infections (19.9%) were the primary cause of death in European

paediatric RRT patients.

Conclusion: Considerable variation exists in the current demographics of children treated

with RRT across Europe.

European demographics

15

INTRODUCTION

End-stage renal disease (ESRD) is a very rare condition in children [5]. Information on the

number of new cases (incidence), the size of the problem (prevalence), and the outcomes with

respect to survival are important to patients, physicians, and health care providers.

International collaboration is required to empower statistically valid and clinically meaningful

studies. Therefore, in 2007, the European Society for Paediatric Nephrology (ESPN) and the

European Renal Association and European Dialysis and Transplantation Association (ERA-

EDTA) initiated the ESPN/ERA-EDTA Registry [50]. This registry provides data on all

European children with ESRD receiving renal replacement therapy (RRT). Currently, 37

European countries contribute data annually, covering a total paediatric population of 176

million children and adolescents aged 0 to 19 years. As up-to-date, accurate, and detailed

demographics on the state of paediatric RRT in Europe are necessary for purposes of policy

making and information provision to patients, physicians, and health care providers, we aim to

describe the demographics of paediatric RRT over the period 2009-2011.

METHODS

Individual patient data on date of birth, gender, start date of RRT, treatment modality, changes

in treatment, and events such as graft failure, death, and transfer out of the registry, were

extracted from the ESPN/ERA-EDTA registry database for 37 European countries. Most

countries provided data for the years 2009-2011. Moldova, Turkey, and Bosnia & Herzegovina

provided data for 2011 only. Albania, Germany, Malta, and Ukraine provided data from 2010

onwards, and Montenegro provided data for the year 2009 only. Germany only reported

transplant patients, and the coverage was 12 out of 20 transplant centres for 2010, and 15 out

of 20 centres for 2011. The incidence and prevalence calculations were adjusted accordingly.

(Pre-emptive) transplant patients were not reported by Italy. Overall prevalence and incidence

rates were therefore provided with and without Germany and Italy, as including these

countries will underestimate the incidence and prevalence. Detailed information on the

ESPN/ERA-EDTA registry can be found elsewhere [50, 51]. As most countries report

information collected from paediatric centres only, older children may be treated in adult

centres and therefore missed by the registry. As the incidence and prevalence in the 15-19 age

group may therefore be underestimated, demographics are presented for the 0-14 age group


16

for all 37 countries, as well as for the 0-19 age group for 9 countries reporting from both

paediatric and adult centres (listed in table 1).

For each country, the 3-year average RRT incidence was calculated. The 3-year average was

chosen to minimize the impact of random variation caused by the rarity of paediatric ESRD.

Incidence was defined as the number of new paediatric patients starting RRT per year,

between 2009-2011, per million age-related population (pmarp). The point prevalence was

given by the total number of paediatric patients on RRT on December 31 2011, expressed as

pmarp [52]. The prevalence in Montenegro was calculated for the 31st December of 2009.

Separate incidence rates and prevalence were calculated for gender, age groups (0-4, 5-9, 10-

14, 15-19) and treatment modalities (peritoneal dialysis [PD], haemodialysis [HD], and (pre-

emptive) transplants [Tx]). They were calculated using age- sex- and year- specific census data

obtained from the Eurostat database [53]. The treatment modality at 30 days after starting

RRT was taken for the calculation of modality-specific incidence. Primary renal diseases (PRD)

were classified following the ERA-EDTA grouping of PRD codes for children [54]. The

estimated glomerular filtration rate (eGFR) at the start of RRT was calculated using the

revised bedside Schwartz formula [55]. The median eGFR was calculated, and following the

KDOQI guidelines for initiation of dialysis in children, patients were categorized as starting

RRT with an eGFR of 8 mL/min/1.73m2 or less, an eGFR between 9 -14 mL/min/1.73m2, or an

eGFR of 15 mL/min/1.73m2 or above[56].

The overall survival probability at 4 years after starting RRT was calculated per age group and

by treatment modality at start, for incident patients starting RRT between 2007 and 2011,

using the Kaplan-Meier method. In addition, the probability of receiving a transplant within 4

years after starting RRT was calculated using the cumulative incidence competing risk (CICR)

method. A competing risk has been defined as “an event that either hinders the observation of

the event of interest or modifies the chance that this event occurs” [57]. As we are interested

in the event of receiving a transplant, we considered death as a competing risk, as these

patients can no longer receive a transplant. The analyses were restricted to 4 years because of

the limited amount of follow-up time [58]. Differences between groups were analysed using

Cox regression hazard ratios (HR) and 4-year survival probabilities, using the oldest age group

and the pre-emptive Tx group as reference groups. The causes of death were provided using


17

the ERA-EDTA coding system, whereas ‘cardiac failure’, ‘cardiac arrest/sudden death other

causes’, and ‘myocardial ischemia and infarction’ were combined to ‘cardiovascular mortality’

[59]. All analyses were performed using SAS version 9.3.

RESULTS

Incidence

A total of 1697 patients aged 0-14 started RRT between 2009 and 2011 in 37 European

countries. The average overall incidence rate of paediatric RRT was 5.5 pmarp, and 5.2 pmarp

including Germany and Italy. The incidence was 5.1 pmarp in children aged 0-4 years, 4.1

pmarp in children aged 5-9 years, and 6.3 pmarp in children aged 10-14 years. In 9 countries

reporting from both paediatric and adult centres, the incidence rate was 8.3 pmarp in the 0-19

age group, and 13.3 pmarp in children aged 15-19 years.

The following results are presented for the 0-14 age group in 37 European countries. The

incidence varied markedly between countries, following a Poisson distribution, with an inter-

quartile range (IQR) of 3.4 – 7.0 pmarp and a variance of 6.6, as illustrated by figure 1. Males

(5.8 pmarp) had a 29% higher incidence as compared to females (4.5 pmarp). Almost half of

the patients started with PD (47.1%, 2.4 pmarp, IQR between countries 0.9 – 3.3), followed by

HD (33.3%, 1.7 pmarp, IQR 1.0 – 2.4), while 317 children received a pre-emptive Tx (19.6%,

1.0 pmarp, IQR 0.0 – 1.6). Congenital anomalies of the kidney and urinary tract (CAKUT)

were the most common cause of renal disease, accounting for 41.3% (2.2 pmarp) of all

incident RRT patients. Glomerulonephritis ranked second with an incidence of 0.8 pmarp and

cystic kidney disease third with an incidence of 0.5 pmarp. The frequency distribution and

incidence of primary renal diseases are displayed in figure 2. The median eGFR at start of RRT

was 9 mL/min/1.73 m2. 49.8% of the patients started RRT with an eGFR of < 9, 34.4% with an

eGFR between 9 -14 mL/min/1.73m2, and 15.8% with an eGFR of 15 mL/min/1.73 m2 or above.


18

Figure 1. Paediatric RRT incidence for patients ages 0-14 per country for the period 2009-2011. Data from

Germany are based on transplantation patients only. Transplantation patients are not included in the patients

from Italy. Therefore, the numbers are an underestimation of the true incidence.

Prevalence

On the 31st of December 2011 there were 3595 prevalent patients between the age of 0-14

years in 37 European countries, resulting in a point prevalence of paediatric RRT in Europe of

27.9 pmarp, whereas it was 28.1 pmarp excluding Germany and Italy. Prevalence was 13.5

pmarp in children aged 0-4 years, 26.4 pmarp in children aged 5-9 years, and 44.4 pmarp in

children aged 10-14 years. In 9 countries reporting from both paediatric and adult centres, the

prevalence was 58.0 pmarp in the 0-19 age group, and 109.0 pmarp in children aged 15-19

years. The following results are presented for the 0-14 age group in 37 European countries.

Similar to the incidence, the prevalence varied markedly between countries with an IQR of

21.8 – 43.9 pmarp. 74.6% of the prevalent patients were living with a renal allograft (19.4

pmarp), 23.1% on PD (6.0 pmarp), and 13.8% on HD (3.6 pmarp). Prevalence is presented by

age group, gender, and treatment modality, for each country in table 1.


19

Table 1. Prevalent paediatric patients on RRT on the 31st of December 2011, per million age related

population, by age, gender, and modality. * Countries reporting data collected from both adult and paediatric

centres. The total prevalence in the 15-19 age group is based on these countries. a Based on transplant patients

only. b Transplant patients are not included. c Prevalence calculated on the 31st December 2009.

0-14y 0-4y 5-9y 10-14y 15-19y Males Females HD PD Tx

N pmarp pmarp pmarp pmarp pmarp pmarp pmarp pmarp pmarp pmarp

Albania 3 5.0 0.0 5.2 8.4 3.6 6.4 3.5 5.0 0.0 0.0 Austria * 52 42.3 25.4 34.5 65.1 99.5 57.1 26.7 3.3 0.8 38.2 Belarus 31 21.8 9.5 18.1 39.8 53.0 17.8 26.1 2.1 7.0 12.7 Belgium 85 45.6 14.0 39.4 85.0 116.9 48.3 42.8 8.0 7.0 30.6 Bosnia and Herzegovina 10 16.5 24.9 4.5 22.3 44.8 16.1 16.8 9.9 1.6 4.9 Bulgaria 10 10.2 0.0 9.4 22.3 35.2 17.9 2.1 5.1 2.0 3.1 Croatia 21 31.8 28.0 24.3 41.7 102.6 35.4 28.0 4.5 9.1 16.7 Czech Republic 41 27.4 6.8 40.9 37.6 42.9 28.0 25.5 3.9 9.1 13.7 Denmark * 44 43.9 24.5 24.1 81.3 115.9 62.3 24.5 2.0 4.0 37.9 Estonia 2 9.7 0.0 29.6 0.0 27.8 9.4 10.0 0.0 0.0 9.7 FYR of Macedonia 8 22.4 17.4 8.7 39.4 13.4 27.2 17.4 5.6 14.0 2.8 Finland * 75 84.4 52.9 82.4 118.7 139.4 88.1 80.6 2.3 6.8 75.4 France * 417 35.0 14.7 35.3 54.7 112.1 41.6 28.0 6.0 3.8 18.9 Germany a 190 22.0 12.2 26.3 26.6 38.9 27.5 16.2 N.A. N.A. 19.6 Greece * 50 30.8 21.0 28.1 44.3 41.2 32.2 27.9 3.1 12.9 14.8 Hungary 41 28.3 14.7 26.9 42.9 96.8 32.3 24.1 2.1 9.0 17.2 Iceland * 4 60.2 42.5 47.0 92.5 129.2 58.9 61.5 15.1 15.1 30.1 Italy b 258 30.3 16.6 30.9 43.4 52.7 36.3 23.9 2.7 6.6 N.A. Lithuania 10 22.2 13.3 14.7 36.5 62.3 21.6 22.8 6.6 6.6 8.9 Malta 4 64.6 50.4 51.4 88.4 0.0 62.8 66.4 0.0 0.0 64.6 Moldova 3 5.2 0.0 0.0 14.8 21.9 6.7 3.5 3.4 1.7 0.0 Montenegro c 4 32.8 76.3 0.0 23.5 22.1 47.3 17.0 0.0 16.4 16.4 Netherlands * 133 45.8 28.2 39.9 67.7 116.7 55.9 35.3 4.1 5.9 35.8 Norway * 41 44.4 29.1 56.8 47.9 123.3 50.8 37.8 3.3 5.4 35.8 Poland 227 38.9 17.9 38.6 61.3 51.1 45.8 31.3 2.9 7.9 27.8 Portugal 77 49.1 31.1 47.7 65.8 87.1 57.4 40.5 2.6 15.9 30.6 Republic of Serbia 30 27.6 11.7 23.7 46.4 67.2 28.6 26.5 4.6 4.6 18.4 Romania 40 12.4 5.6 10.4 20.9 56.3 11.5 13.4 5.3 5.3 1.9 Russia 289 13.6 6.0 12.0 24.2 26.8 15.1 12.1 2.4 4.2 6.8 Slovakia 21 25.2 3.5 11.4 60.3 73.4 32.8 17.3 3.6 14.4 7.2 Slovenia 12 41.0 9.3 21.8 97.0 49.3 53.2 28.2 6.8 10.3 23.9 Spain 295 41.9 19.5 39.5 69.5 61.0 49.7 33.6 3.7 4.8 33.2 Sweden * 81 51.4 37.2 45.9 74.0 117.3 65.5 36.6 5.1 5.1 41.3 Switzerland 52 43.5 12.6 44.1 73.0 71.4 48.9 37.9 1.7 5.0 35.2 Turkey 298 15.8 7.4 10.8 28.2 24.9 17.2 14.3 2.5 8.3 4.9 Ukraine 38 5.8 1.6 2.9 13.9 18.4 6.0 5.7 3.1 1.7 1.1 United Kingdom 598 54.5 24.4 55.1 87.9 105.5 64.9 43.7 4.9 7.5 42.1

Total 3595 27.9 13.4 26.4 44.4 56.6 32.5 23.0 3.6 6.0 19.4


20

Transplantation

The probability of receiving a first Tx within 4 years after initiating RRT was 76.9% for patients

aged 0-19; 68.9% for the youngest patients, 81.6% for 5-9 year olds, 79.3% for the 10-14 year

olds, and 75.8% for the 15-19 year olds, meaning that the probability of receiving a Tx within 4

years for the 0-4 years old was 35% lower as compared to the 5-9 year olds (HR = 0.65, 95%

CI; 0.57 – 0.74), 39% lower as compared to the 10-14 year olds (HR = 0.61, 95% CI; 0.54 –

0.68), and 36% lower as compared to the 15-19 year olds (HR = 0.64, 95% CI; 0.56 – 0.72).

This probability did not differ significantly between the oldest three age groups (10-14 as

compared to 5-9, HR = 0.93, 95% CI; 0.82 – 1.04; 15-19 as compared to 5-9, HR = 0.95, 95%

CI; 0.86 – 1.06). The probability of receiving a Tx within 4 years after initiating RRT on HD

and PD was 70.9% and 71.3%, respectively. HD patients seemed to receive a Tx earlier (mean

1.06 years) as compared to PD patients (1.29 years, p<0.0001). However, this relationship was

fully explained by the difference in age, as HD patients are generally older than PD patients.

The probabilities of receiving a Tx within 4 years after initiating RRT and corresponding hazard

ratios are presented in table 2 by modality at start and age group.

Figure 2. PRD frequency distribution and incidence for paediatric patients aged 0-14 years starting RRT

between 2009 - 2011.


21

Survival and cause of death

The median follow-up was 2.17 years for paediatric patients aged 0-19 starting RRT between

2007 and 2011. The overall survival probability at 4 years was 93.7%. The crude mortality rate

was 20 deaths per 1000 patient years in the 0-19 age group, and 23 deaths per 1000 patient

years in the 0-14 age group, which was 55 times higher compared to the general paediatric

population (0.42 deaths per 1000 children aged 0-14 years in the European Union in

2011)[60]. The youngest age group had the poorest 4-year survival with a 4.4 fold increased

risk of death when compared to the oldest age group (HR = 4.4, 95% CI; 2.8 – 7.0). The 4-

year survival for the 5-9 year olds did not differ significantly from the oldest age group (HR =

1.4, 95% CI; 0.8 – 2.4), nor did the 10-14 year olds compared to the oldest age group (HR =

1.1, 95% CI; 0.6 – 1.9).

Figure 3. Causes of death for paediatric patients aged 0-19 years starting RRT between 2007-2011, by

modality.


22

Patients initiating RRT with dialysis had a 6.6 fold increased risk of death as compared to

patients receiving a pre-emptive Tx (HR = 6.6, 95% CI; 2.9 – 14.8). There was no significant

difference in survival between patients treated initially with HD or PD (reference group is PD,

HR = 1.0, 95% CI; 0.7 – 1.4). The survival probabilities at 4 years and corresponding hazard

ratios are presented by age group and treatment modality at start in table 2. The main known

overall cause of death in patients was infections (19.9%), followed by cardiovascular causes

(13.1%), and cerebrovascular accidents (6.8%). Infections were the leading cause of death in

those on PD and those with a functioning graft, whereas cardiovascular causes of death

predominated in those on HD. The frequency distribution for all causes of death is displayed

by modality in figure 3.

Table 2. Probabilities and hazard ratios of death at 4 years, and receiving a transplant (Tx) within 4 years, by

age group and treatment modality at start, for paediatric patients starting RRT between 2007-2011. * Death

was considered a competing risk for receiving a Tx.

4-year survival Received Tx within 4 years*

% (95% CI) HR (95% CI) % (95% CI) HR (95% CI)

Overall 93.7 (92.8 - 94.7) - 76.9 (74.9 - 78.8) -

0-4 87.1 (84.5 - 89.7) 4.4 (2.8 - 7.0) 68.9 (64.8 - 73.0) 0.64 (0.56 - 0.72)

5-9 95.3 (93.3 - 97.3) 1.4 (0.8 - 2.4) 81.6 (77.6 - 85.6) 0.98 (0.86 - 1.11)

10-14 96.2 (94.9 - 97.6) 1.1 (0.6 - 1.9) 79.3 (76.3 - 82.3) 1.05 (0.94 - 1.17)

15-19 96.3 (94.2 - 98.4) 1 75.8 (70.5 - 81.0) 1

PD 92.5 (90.9 - 94.1) 6.5 (2.9 - 14.9) 71.3 (68.1 - 74.5) 0.90 (0.82 - 1.00)

HD 92.3 (90.4 - 94.2) 6.6 (2.9 - 15.2) 70.9 (67.3 - 74.4) 1

Tx 99.1 (98.5 - 99.8) 1 100 (by definition) -


23

DISCUSSION

In this paper, we present an epidemiological picture of the incidence, prevalence, survival, and

other paediatric RRT demographics in 37 European countries over the period 2009-2011. The

incidence, prevalence, and initial treatment modality of paediatric RRT varied greatly amongst

European countries. The average European paediatric RRT incidence was 5.5 cases pmarp in

children aged 0-14 years and 8.3 pmarp in children aged 0-19, approximately 20 times lower

compared to adults (165.7 pmarp, aged 20+, for 2011) [54]. As most countries reported

information collected from paediatric centres only, the incidence will be underestimated in the

15-19 age group. We will therefore refer to the incidence rate for children aged 0-14 for

comparisons with other published paediatric RRT information. The incidence in Europe was

similar as in Malaysia (5.7 pmarp, ages 0-14, for the period 2008-2012) [61] and Canada (5.9

pmarp, ages 0-14, for 2010) [62], lower compared to Australia and New Zealand (8.3 and 6.7

pmarp, ages 0-14, for the period 2007-2011) [63], and approximately half that observed in the

US (11.6 pmarp, ages 0-14, for 2011) [64]. In the US and Canada, HD is the most common

therapy for paediatric patients [62]. In contrast, PD is the first treatment of choice in Europe,

as is also the case in the Malaysian and Australia & New Zealand registries [61, 63]. The

international variation in RRT rates may be explained, to some extent, by random variation

caused by the rarity of paediatric ESRD, by variation in the occurrence of different causes of

renal failure in each country (e.g. the relatively high incidence of Finnish-type nephropathy in

Finland), and by economic disparities between countries, as has been shown for PD prevalence

[23], and transplantation rate [42]. Furthermore, the higher incidence in the US has been

attributed to an earlier RRT start [5, 65]. Indeed, our data show that almost half of the

European children start RRT in the lowest eGFR category. However, the majority of variation

between countries remains unexplained, warranting further investigation to establish the

underlying factors.

The overall survival at 4 years was 93.7% in European RRT children. Mortality in paediatric

RRT patients was 55 times higher than that of the general paediatric population [60]. The

European RRT mortality rate in 0-14 year old patients of 23 deaths per 1000 patient years was

approximately half that of the US (44 deaths per 1000 patient years in patients aged 0-14,

period 2005-2009). Part of this difference might be explained by the higher number of patients

aged 0-4 starting RRT in the US [62]. The European mortality rate was similar as observed in


24

the Canadian [66], Australia & New Zealand [8], and Taiwanese registries [35] (17.9, 21.0, and

23.4 deaths per 1000 patient years, ages 0-19, periods 1993-2002, 1992-2007, and 1995-2004,

respectively). Age as a risk factor for mortality in European RRT patients is consistent with

other registries, showing superior survival in patients with a functioning graft compared to

patients on dialysis, and the poorest survival in the youngest patients.

In contrast to reports from other international registries where cardiovascular disease is the

leading cause of death in paediatric RRT patients [34, 64, 67], the primary overall cause of

death in Europe is infection, although followed closely by cardiovascular disease. Compared to

a previous European report covering the period 1980-2000, the distribution of causes of death

has remained very similar, with infections predominating in those on PD and those with a

functioning graft, whereas cardiovascular mortality was the leading cause of death in those on

HD [68]. In a long-term follow-up cohort of formerly paediatric patients starting RRT <15

years of age, a shift towards infection as the leading cause of late mortality was found over the

past decade (in the 4th and 5th decade of life) due to a decreased risk of cardiovascular

mortality, possibly attributable to an increased awareness and improved treatment of

cardiovascular disease [69, 70]. A similar decline in cardiovascular mortality was seen in US

children starting RRT on dialysis, and in adult dialysis patients in Australia & New Zealand,

although cardiovascular disease remains the primary cause of death in these registries [64, 67].

Unfortunately, several countries were unable to provide information on the cause of death,

resulting in the high percentage of causes reported as unknown in our data which may

complicate comparisons, although the ranking of causes remained the same when excluding

these countries.

In conclusion, the current paper provides demographics on the incidence, prevalence, survival,

causes of death, and eGFR at treatment start for paediatric patients treated with RRT for

ESRD in 37 European countries over the period 2009-2011, using ESPN/ERA-EDTA registry

data. Considerable variation exists in the demographics of RRT within Europe, for largely

unknown reasons. Future investigation is necessary to ascertain the factors underlying this

variation.

Demographics

3

Disparities in Treatment

Rates of Paediatric End-Stage

Renal Disease across Europe:

Insights from the


Nicholas C Chesnaye, Franz Schaefer, Jaap W Groothoff,

Fergus J Caskey, James G Heaf, Stella Kushnirenko, Malcolm

Lewis, Reiner Mauel, Elisabeth Maurer, Jussi Merenmies, Diamant

Shtiza, Rezan Topaloglu, Natalia Zaicova, Argyroula Zampetoglou,

Kitty J Jager, Karlijn J van Stralen

Nephrol Dial Transplant 2015 Aug; 30(8): 1377–1385

ABSTRACT

Background: Considerable disparities exist in the provision of paediatric renal replacement

therapy (RRT) across Europe. This study aims to determine whether these disparities arise

from geographical differences in the occurrence of renal disease, or whether country-level

access-to-care factors may be responsible.

Methods: Incidence was defined as the number of new patients aged 0-14 years starting RRT

per year, between 2007-2011, per million children (pmc), and was extracted from the

ESPN/ERA-EDTA registry database for 35 European countries. Country level indicators on

macroeconomics, perinatal care, and physical access to treatment were collected through an

online survey and from the World Bank database. The estimated effect is presented per 1

standard deviation (SD) increase for each indicator.

Results: The incidence of paediatric RRT in Europe was 5.4 cases pmc. Incidence decreased

from Western to Eastern Europe (-1.91 pmc/1321km, p<.0001), and increased from Southern

to Northern Europe (0.93 pmc/838km, p=.002). Regional differences in the occurrence of

specific renal diseases were marginal. Higher RRT treatment rates were found in wealthier

countries (2.47 pmc/€10378 GDP per capita, p<.0001), among those that tend to spend more

on health care (1.45 pmc/1.7% public health expenditure, p<.0001), and among countries

where patients pay less out-of-pocket for health care (-1.29 pmc/11.7% out-of-pocket health

expenditure, p<.0001). Country neonatal mortality was inversely related with incidence in the

youngest patients (ages 0-4, -1.1 pmc/2.1 deaths per 1000 births, p=.10). Countries with a

higher incidence had a lower average age at RRT start, which was fully explained by country

GDP per capita.

Conclusion: Inequalities exist in the provision of paediatric RRT throughout Europe, most of

which are explained by differences in country macroeconomics, which limit the provision of

treatment particularly in the youngest patients. This poses a challenge for health care policy

makers in their aim to ensure universal and equal access to high-quality healthcare services

across Europe.

European disparities in RRT incidence

27

INTRODUCTION

End-stage renal disease (ESRD) in children is a rare and life-threatening disorder, which

requires complex and expensive renal replacement therapy (RRT), i.e. dialysis or renal

transplantation, to sustain life. All European Union Member States have made commitments

towards universal access to high-quality health services, however, inequalities persist in the

provision of paediatric RRT, with considerable differences in treatment rates between

countries [51]. These disparities may exemplify inequalities in the provision of specialized care

in Europe for other rare disorders that are complex and costly to treat [71].

In adults, the geographic variation in RRT rates has been explained, to some extent, by the

percentage of elderly and the prevalence of diabetes and hypertension in the general

population [15–17, 22]. Factors influencing access to treatment, such as the adequacy of renal

service supply, travel times, the number of private for profit centers, macroeconomics, and

access to RRT for older and more comorbid patients, also play a role in explaining the

variation of RRT incidence in adults [18, 19, 25].

It is difficult to extrapolate previous findings from the adult to the paediatric RRT population,

as ESRD is much rarer in children forming only 1% of the total RRT population. In addition,

children are generally treated in public (mostly academic) facilities, and often have a higher

priority to care over adults, for example via the paediatric prioritization of donor kidneys [42].

Conversely, as nearly all cases of paediatric ESRD have at least some genetic origin, differences

in treatment rates across Europe could partly be explained by geographical differences in

genetic background.

As little is known about the causes underlying international variation in paediatric RRT rates,

in the current paper we aim to describe the geographic variation in paediatric RRT incidence

across Europe, and to determine whether this variation arises from geographical differences in

the genetic susceptibility to certain renal diseases (e.g. congenital nephrotic syndrome of the

Finnish-type (CNF) in Finland), and to what extent non-medical country-level factors affect

access to care.


28

METHODS

Paediatric RRT incidence

The European Society for Paediatric Nephrology / European Renal Association-European

Dialysis and Transplantation Association (ESPN/ERA-EDTA) Registry collects data on

paediatric RRT. This population-based registry covers a general population of almost 130

million children from 37 European countries [51]. Most countries report information collected

from paediatric treatment centres only. As older children may be treated in treatment centres

for adult patients, we limited ourselves to children aged 0–14 years. All countries provided

data for the years 2007 to 2011, except for Moldova and Bosnia & Herzegovina (2011 only),

Albania, Germany, Malta, Ukraine (2010 onwards), and Montenegro (2009 only). Germany

reported only on transplant patients, and pre-emptive transplant patients were not initially

reported by Italy. As this will have led to an underestimation of the RRT incidence in Germany

and Italy, these countries were excluded from all analyses. Turkey was excluded as the

national treatment coverage is unknown. Our outcome measure, paediatric RRT incidence,

was defined as the number of new paediatric patients starting RRT per year, between 2007-

2011 (or otherwise available), per million children, as obtained from the Eurostat database for

each corresponding year. RRT incidence rates were standardized for age using the EU-27 for

the year 2010 as reference population [53].

Geographical distribution of RRT incidence

To describe the geographic variation in RRT incidence across Europe, we explored

geographical gradients in RRT incidence by modelling country RRT incidence against the

country’s longitude (from West to East) and latitude (from South to North). Longitude and

latitude were determined by calculating the spatial centroid of each country using ArcGIS

software [72]. Longitude was corrected for latitude and vice-versa to isolate the effect of each

direction. As the geographical location of the centre of Russia could highly affect the

geographic gradient results due to the extreme Eastern position, combined with a higher

population density in the West compared to the East of Russia, coordinates were based on

the position of Moscow.


29

Geographical distribution of renal disease

To determine whether the variation in RRT incidence arises from geographical differences in

the genetic susceptibility to certain renal diseases, we examined differences in the occurrence

of causes of renal failure by European region. To achieve the statistical power necessary to

detect geographical differences, genetically similar countries were aggregated to the regions

East, West, South, and North following the regional division of Europe as described by Ralph

et. al. [73] (as listed in appendix 1). In their paper, the authors describe the genetic geography

of Europe, creating regional divisions of Europe based on the geographic location and

correlations in the pattern of genome-wide data from 2,257 Europeans. As the countries

Belarus (East), Moldova (East), Lithuania (East), Malta (South), and Iceland (North) were not

included in their study, we allocated them based on geographic location alone. Causes of renal

failure were classified into 10 primary renal disease (PRD) groups according to the ERA-EDTA

coding system for children [54]. We compared observed and expected percentages by PRD

group and region using the chi-square test. The chi-square contribution of each table cell was

used to determine, per region, which PRD groups occurred disproportionately more or less

frequently compared to the rest of Europe, including correction for multiple testing using false

discovery rate adjusted (FDR) p-values [74].

Country access-to-care indicators

We constructed a conceptual framework to illustrate how potential country-level factors may

influence access to paediatric RRT in Europe (appendix 2), based on a review of the literature

and consultation with paediatric nephrologists from the 37 European countries. Country

indicators were extracted from the World Bank Database for each country and averaged for

the corresponding years that RRT incidence data were collected [75]. Information regarding

the number of paediatric RRT centres and reimbursement policies were collected through an

online survey to a paediatric nephrologist in each of the 37 European countries between

November and December 2013 (appendix 3). Reminders were sent in case of non-response.

The response rate was 70%. Information on the proportion of diagnosed congenital anomalies

of the kidney and urinary tract (CAKUT) cases resulting in termination of pregnancy was

obtained from the EUROSCAN project for congenital malformations registries in 10 European

countries (appendix IV) [76].


30

Statistical analyses

Linear regression models were used to explore the associations between country indicators

and paediatric RRT incidence. Multivariate models were adjusted for macroeconomic factors.

All regression analyses were weighted with the inverse standard error of the RRT incidence

rate, which gives larger countries with a larger number of cases more influence on the

regression slope. To allow for comparison of effect size across continuous indicators, we

presented the estimated effect for a 1 standard deviation (SD) increase for each indicator [77].

Joinpoint regression was applied to identify any significant changes in the linear regression

slope, allowing us to identify potential non-linearity (including floor and ceiling effects) in the

relationship between continuous indicators and RRT incidence [78].

Country Indicator Description

GDP per capita1 Gross domestic product (GDP) per capita, expressed as purchasing power parity (PPP) in Euros, is a measure for country wealth. The PPP method allows for the international comparison of economies.

Public health expenditure1 Public health expenditure consists is expressed as the percentage of national GDP that a government spends on health care.

Out-of-pocket health expenditure1

Out-of-pocket health expenditure is the proportion of private health expenditure borne directly by the patient where insurance does not cover the full cost of the health care.

Neonatal mortality rate1 An indicator of the quality of paediatric health care systems, expressed as the number of neonates dying before reaching 28 days of age, per 1000 live births.

CAKUT termination rate2 The proportion of CAKUT cases terminated during pregnancy.

Prenatal screening3 Availability of a prenatal ultrasound screening (at 18-22 weeks) to examine the foetus for presence of abnormalities.

Density of centres providing paediatric RRT3

The number of centres providing paediatric RRT, expressed per million children (ages 0-14).

Paediatric population density1 The number of children (ages 0-14) per km2.

Rural population1 The percentage of rural population.

Paved roads1 The percentage of paved roads.

Capability to treat neonates3 The capability to provide RRT to neonates (<28 days).

Reimbursement of treatment3 Reimbursement of paediatric RRT treatment, defined as >90% reimbursement of treatment costs for HD, PD, and Tx.

Reimbursement of travel costs3

Reimbursement of travel costs, defined as >90% reimbursement of travel costs.

Sources: 1World Bank Database,

2Wiesel et al,

3Survey.


31

RESULTS

Geographic variation of RRT incidence

The overall age-adjusted incidence rate of paediatric RRT between 2007 and 2011 for 34

countries in Europe was 5.4 pmc and varied markedly between countries, as illustrated by

figure 1. RRT incidence was lowest in Eastern Europe (3.6 pmc), followed by Southern (7.2

pmc), Western (7.8 pmc), and Northern Europe (8.1 pmc). The geographical variation in RRT

incidence was demonstrated by a strong gradient decreasing from West to East (-1.91 pmc

per SD increase in longitude, 1321km, p < .0001), and increasing from South to North (0.93

pmc per SD increase in latitude, 838km, p=.002).

Figure 1. Paediatric RRT incidence per country for the period 2007-2011. Data from Germany are based on

transplantation patients only and transplantation patients are not included from Italy. Therefore, the numbers

are an underestimation of the true incidence and were not mapped.


32

Geographic variation of renal disease

Using RRT incidence as a proxy for disease occurrence, we explored whether regional

differences in the distribution of renal diseases could explain the variation in RRT incidence

(figures 2a and 2b). Relative to the total number of cases, in Eastern Europe there was a

significantly lower incidence of RRT for hereditary nephropathies (-1.7%, p=.0001, FDR p=.03)

and haemolytic-uraemic syndrome (0.4%, p=.003, FDR p=.03), both of which remained

statistically significant after adjustment for multiple testing, and a higher incidence of RRT for

cystic kidney disease (2.8%, p=.03, FDR p=.14). Furthermore, there was a relatively high

proportion of patients with hereditary nephropathies, mainly CNF, in Northern, (8.6%, p=.01,

FDR p=.07) and Western Europe (1.9%, p=.05, FDR p=.19). Finally, there was a relatively high

incidence of patients with haemolytic-uremic syndrome in Southern Europe (3.0%, p=.01, FDR

p=.07). There were no further significant differences in PRD group distribution across

European regions (table 2).

Macroeconomics

RRT incidence increased with 2.47 pmc for every SD increase in GDP per capita (95%CI 1.68

– 3.26, p<.0001; figure 3c). In countries spending less than 7.5% of national GDP on public

health, RRT incidence increased with 2.48 pmc for every SD increase of public health

expenditure (95%CI 2.19 – 2.77, p<.0001), but this effect was absent in countries spending

more than 7.5% of GDP on public health (p=.26, figure 3d). There was an inverse association

between out-of-pocket health expenditure and RRT incidence, with incidence decreasing by

1.83 pmc for every SD increase in out-of-pocket health expenditure (95%CI -2.27 to -1.38, p

<.0001). All factors remained significant after adjustment for GDP.


33

Table 2. Summary of incidence, macroeconomic, perinatal care, access to treatment, and reimbursement

characteristics of countries. *Because of the skewed distribution, the log-transformation of paediatric

population density was used in all analyses.

Mean (std.)

No. of countries (%)

Incidence (pmc)

All 5.4 (2.6) 34 (100)

Ages 0-4 5.2 (5.9) 34 (100)

Ages 5-9 4.2 (3.3) 34 (100)

Ages 10-14 6.8 (3.9) 34 (100)

Western Europe 7.8 (1.5) 5 (15)

Eastern Europe 3.6 (2.4) 21 (62)

Southern Europe 7.2 (1.2) 3 (9)

Northern Europe 8.1 (0.9) 5 (15)

Age at start RRT (years) 7.8 (5.0) 34 (100)

Macroeconomics

GDP per capita (per €10 000) 2.5 (1.3) 34 (100)

Public health expenditure (% of GDP) 6.1 (1.7) 34 (100)

Out-of-pocket expenditure (% of total health expenditure) 25 (11.9) 34 (100)

Perinatal care

Prenatal screening (% yes) 96 26 (76)

Proportion of CAKUT cases terminated during pregnancy (%)

20.8 (18.7) 10 (29)

Neonatal mortality (per 1000 births) 4.0 (2.1) 34 (100)

Access to treatment Paediatric population density (children per km2) * 22.1 (35.2) 34 (100)

Rural population (%) 29.8 (13.8) 34 (100)

Centres providing paediatric RRT (pmc) 2.4 (6.2) 26 (76)

Capability to treat neonates (% yes) 92 25 (74)

Paved roads (%) 76.5 (24.3) 29 (85)

Quality of trade and transport-related infrastructure (scale 1-5) 2.9 (0.6) 34 (100)

HD reimbursement (% yes) 92 26 (76)

PD reimbursement (% yes) 92 26 (76)

Tx reimbursement (% yes) 92 26 (76)

Travel cost reimbursement (% yes) 69 29 (85)


34

Figure 2a. The geographical distribution of renal disease divided across European regions. The cumulative

incidence is presented for the largest PRD groups. GN = glomerulonephritis, CAKUT = congenital anomalies

of the kidney and urinary tract, HN = hereditary nephropathies, CKD = cystic kidney disease.

Figure 2b. The geographical distribution of renal disease across Europe. GN = glomerulonephritis, CAKUT =

congenital anomalies of the kidney and urinary tract, CKD = cystic kidney disease, HN = hereditary

nephropathies, IS = ischaemic renal failure, HUS = haemolytic-uraemic syndrome, MD = metabolic disorders,

VAS = vasculitis. * P-value < 0.05. ** Significant after FDR adjustment for multiple testing.

1.1 1.0 1.20.6

2.5 3.0 3.0

1.4

0.80.7 0.7

0.5

1.0 0.7 0.5

0.2

2.7 2.41.9

0.9

0

1

2

3

4

5

6

7

8

North West South East

Cumulative RRT inciden

ce

Other

HN

CKD

CAKUT

GN

2.2

‐0.2

2.8*

‐1.7**

1.5

‐0.4** ‐0.5 ‐0.2

‐1.3

‐4.9

‐0.8

8.6*

‐1.1

‐0.4 ‐0.7

1.70.9

4.5

‐0.7

0.0

‐0.7

3.0*

‐0.3

0.3

‐1.6‐0.6

‐1.3

1.9*

‐0.1 0.7

1.1

‐0.1

‐10

‐8

‐6

‐4

‐2

0

2

4

6

8

10

GN CAKUT CKD HN IS HUS MD VAS

Explained

regional variation in

RRT inciden

ce (%)

East

North

South

West


35

Perinatal care

Country neonatal mortality, as a proxy for the access to and quality of paediatric care, was

negatively correlated with RRT incidence (-1.82 pmc for each SD, 95%CI -2.46 to -1.28,

p<.0001, figure 3e). This effect was explained by country macroeconomics, as after adjustment

for GDP per capita and public health expenditure the estimate was reduced to -0.43 pmc

(95%CI -1.13 to 0.27, p=.22). In countries with a neonatal mortality of less than 6.7 per 1000

births, RRT incidence declined by 2.2 pmc per SD increase (95%CI -1.64 to -0.47, p=.001),

while there was a floor effect in countries with a neonatal mortality rate over 6.7 per 1000

births (p=.26).

All countries, except for Montenegro and Moldavia, reported it was possible to provide RRT

to neonates. According to the questionnaire, ultrasound screening in the second trimester to

examine the foetus for the presence of structural abnormalities was available to all women in

all but one country. We identified an inverse relationship between RRT incidence and the

proportion of pregnancies diagnosed with CAKUT that were terminated, with incidence

declining by 1.79 pmc for every SD increase in termination rate (95%CI -3.34 to -0.24, p=.03,

figure 3f). However, this association was lost after excluding the Ukraine as an influential

outlier (0.15 pmc, 95%CI -2.12 to 1.80, p=.86).

Physical access to treatment

In total, there were 153 centres providing one or more modes of paediatric RRT spread

across 29 countries where this information was available, representing an overall density of 2.4

centres pmc. All countries were able to provide both haemodialysis (HD) and peritoneal

dialysis (PD). There was no association between the number of (modality-specific) centres

pmc and (modality-specific) RRT incidence, even after correcting for country GDP per capita

and public health expenditure.

As country indicators for the average distance to a treatment centre and patient travel time,

we explored the relationships between RRT incidence, the general paediatric population

density, the percentage of rural population, the percentage of paved roads, and the quality of

trade and transport-related infrastructure. Countries with a denser paediatric population had

higher RRT incidence rates (4.08 pmc for every SD increase in density, 95%CI 2.74 to 6.10,


36

p<.0001), also after correcting for country GDP per capita. RRT incidence tended to be lower

in countries with a higher percentage of rural population (-1.02 pmc per SD increase in rural

population, 95%CI -2.22 to 0.19, p=.10), but this was reversed after correcting for country

GDP per capita (1.17 pmc per SD increase in rural population, 95%CI 0.07 to 2.28, p=.04).

The quality of trade and transport-related infrastructure was associated with RRT incidence,

increasing by 1.94 pmc for every SD increase in infrastructure score (95%CI 1.63 to 2.25,

p<.0001), however, we were unable to adjust for GDP per capita due to collinearity (r=0.89).

The percentage of paved roads was not associated with RRT incidence (p=.68 after adjustment

for GDP per capita).

RRT treatment costs were reimbursed for all modalities in 26 responding countries, except

for Bulgaria and Moldova, where none were reimbursed. Patient costs for travelling to RRT

centres were reimbursed in 20 of the 29 responding countries, and were not associated with

RRT incidence (p=.49).

Sensitivity analyses by patient age

We performed sensitivity analyses to establish whether the associations between country-level

indicators and RRT incidence were explained by country differences in patient age distribution.

RRT incidence varied more between countries in patient ages 0-4 (IQR 2.9–8.8) compared to

ages 5-9 (IQR 3.1–5.6) and 10-14 (IQR 4.1–7.8) years. Countries with a higher RRT incidence

were treating children at a younger age (IQR for all countries 7.1–8.8), with RRT starting age

decreasing by 0.66 years for every SD increase in incidence (1 SD = 2.43 pmc, p=.0002). This

effect was largely explained by country GDP per capita (p=.26 after adjustment). Regarding

macroeconomics, national GDP had a stronger effect on the RRT incidence in the youngest

patients (ages 0-4, 3.33 pmc per SD, p<.0001), and to a lesser extent in the older patients

(ages 5-14, 1.91 pmc per SD, p<.0001). Similar increases in effect were observed for public

health expenditure and out-of-pocket health expenditure for the youngest patient group. In

addition, there was a trend in the association between neonatal mortality and RRT incidence

in the youngest patient group after adjustment for GDP per capita (-1.84 pmc per SD, 95%CI -

2.8 to -0.9, p=.0003) and public health expenditure (-1.1 pmc per SD, 95%CI -2.4 to 0.25,

p=.10), but was no longer present in the older age groups.


37

Table 3. Associations between paediatric RRT incidence and country level indicators. The effect estimate is

expressed as incidence per million children per standard deviation increase. aAdjusting for GDP per capita.

bAdjusting for GDP per capita and public health expenditure. cExcluding Ukraine as an outlier. dModality specific

incidence was used as the dependent variable.

Univariate Adjusted for macroeconomics

Effect estimate (95%CI) P-value Effect estimate (95%CI) P-value

Macroeconomics GDP per capita in € (per SD = €10378 increase) 2.47 (1.68 to 3.26) < .0001 - -

Public health expenditure as percentage of GDP (per SD = 1.7% increase) 1.71 (1.34 to 2.08) < .0001 1.45 (0.83 to 2.07) a < .0001

Out-of-pocket expenditure as % of total health expenditure (per SD = 12% increase)

-1.83 (-2.27 to -1.38) < .0001 -1.29 (-1.91 to -0.67) a < .0001

Perinatal care Neonatal mortality (per SD = 2.1 deaths in 1000 births increase)

-1.82 (-2.46 to -1.28) < .0001 -0.43 (-1.13 to 0.27) b .22

Proportion of CAKUT cases terminated during pregnancy (per SD = per 19% increase)

-1.79 (-3.34 to -0.24) 0.03 -0.15 (-2.12 to 1.80) c 0.86

Access to treatment Population density in children per km2 (per SD = 35.2 children per km2 increase) 4.08 (2.74 to 6.10) < .0001 2.73 (1.99 to 3.77) a < .0001

Percentage of rural population (per SD = 14% increase) -1.02 (-2.22 to 0.19) .11 1.17 (0.07 to 2.28) a 0.04

Percentage of paved roads (per SD = 24% increase) 0.28 (-0.68 to 1.23) 0.56 -0.13 (-0.75 to 0.49) b .68

Reimbursement of travel costs (yes versus no) -1.16 (-4.61 to 2.28) .49 -0.66 (-2.32 to 0.99) b .78

Centres providing paediatric RRT (per SD = 6.2 centres pmc increase) -0.76 (-1.92 to 0.39) .14 -0.46 (-1.42 to 0.48) b .32

Centres providing paediatric HD (per SD = 6.1 centres pmc increase) d -0.52 (-1.08 to 0.05) .06 -0.43 (-1.02 to 0.16) b .11

Centres providing paediatric PD (per SD = 4.2 centres pmc increase) d 0.24 (-0.32 to 0.81) 0.38 0.13 (-0.45 to 0.72) b 0.64


38

DISCUSSION

In Europe, there are considerable disparities in paediatric RRT treatment rates between

countries, demonstrated by a decreasing gradient in RRT incidence from West to East and

from North to South. These disparities were largely explained by country differences in

macroeconomics, mainly affecting access to care in the youngest patients.

Geographical differences in the occurrence of specific renal diseases played a marginal role in

explaining the variation in treatment rates. We found that the incidence of RRT for hereditary

nephropathies, including CNF, which constitutes approximately 7% of the total European

paediatric patient population [79], varied most between European regions. Regional variation

in other disease groups was less pronounced, showing only negligible differences in disease

incidence, suggesting that international variation in treatment rates is predominantly

determined by other factors than geographical differences in disease occurrence.

In adults, disparities in country macroeconomics have also been shown to explain a large

portion of the international variation in RRT rates [19, 25]. In children, a non-population-based

study has shown the influence of country wealth on paediatric PD prevalence [23], and

Harambat et al. have shown how the percentage of children living with a functioning graft is

affected by country wealth [42]. Here we show for the first time that disparities in country

wealth and health care financing strongly affect RRT rates across the full spectrum of European

paediatric RRT patients, finding higher treatment rates in wealthier countries, which tend to

spend more on health care, and where patients bear less out-of-pocket health expenditures.

This relationship is understandable given the complexity and cost involved in the provision of

renal care to children by a multi-professional paediatric team, and suggests that the need for

paediatric RRT is not being met by governments burdened with financial constraints [80–82].

Moreover, the effect of country macroeconomics was strongest in the youngest patients,

suggesting that financial constraints particularly hamper access to treatment in infants, who are

the most challenging and resource intensive to treat [12]. Encouragingly, we identified a ceiling

effect in countries spending more than 7.5% of GDP on health care, suggesting that RRT for all

children with ESRD is attainable with health care spending around this margin.


39

Neonatal mortality has previously been used as a marker of health and care during pregnancy

and delivery, and may reflect the effectiveness of health systems in the very young, as well as

various socio-economic factors [83–85]. We found an inverse trend between neonatal

mortality and RRT incidence in the youngest patients, after adjustment for GDP and health

expenditure, suggesting that the effectiveness of a country’s paediatric health-care system is

affecting access to RRT. In addition, it may also reflect country differences in obstetric policies

and the physicians’ willingness to treat very young children with severe comorbidities. As

neonatal dialysis constitutes less than 5% of the European RRT population, we do not consider

neonatal mortality as a direct competing risk for RRT [86].

Congenital urinary tract abnormalities account for 15-20% of all birth defects and are

associated with a high perinatal mortality rate of around 15-30%, mostly due to the

termination of pregnancies and pulmonary hypoplasia [87, 88]. Increases in pregnancy

terminations have previously been associated with declines in the prevalence of congenital

anomalies among live-born infants [89]. Here we show that countries with a high proportion

of CAKUT cases terminated during pregnancy tend to have lower RRT incidence rates,

indicating the degree to which terminations of pregnancy may affect live-birth CAKUT

occurrence [90], although this association should be interpreted with caution given the small

sample size of 10 countries and the loss of significance and effect after excluding the Ukraine.

We found conflicting results regarding physical access to paediatric RRT. On the one hand, we

found higher RRT rates in countries with a high paediatric population density, and lower

incidence rates in more rural countries, suggesting that health services are physically more

accessible in densely populated and urbanized countries, as patients may expect to face lower

travel times and costs compared to more dispersed and rural countries [91]. On the other

hand, we found no association between RRT incidence and the density of RRT centres, nor

with the percentage of paved roads. This may be due to several reasons: 1) the availability of

paediatric renal care throughout Europe has improved substantially over the past decades,

especially for the youngest patients [68, 81], which is supported by our survey results which

show that currently all European countries are able to provide at least one modality of

paediatric dialysis; 2) most children are treated initially with PD, which takes place at home


40

[92]; and 3) parents are likely willing to travel relatively long distances to bridge a short pre-

dialysis period before transplantation.

The current study has several limitations. Inherent to the observational nature of the study,

there is likely to be residual confounding in these associations, which limits us in our ability to

infer causality. The explanatory factors collected via the survey reflect the situation in each

country as of 2013, whilst the country incidence rate was calculated over the period 2007-

2011. Some of the collected indicators may vary over time and this may therefore have

influenced the accuracy of our results. In addition, we do not have full coverage of the entire

study period for all countries which may impact the reliability of our incidence estimates,

although we have attempted to correct for this by weighting our analyses. Aggregation of

countries to regions was necessary to provide the statistical power needed to detect

differences in the geographical distribution of renal disease. However, although this may reflect

the genetic background of these regions, differences between individual countries may be lost.

Furthermore, due to underestimated RRT incidence rates in Italy and Germany, and unknown

national coverage in Turkey, it was necessary to exclude the three large countries from all

analyses, shrinking the relatively small sample size of European countries, which may have

influenced the width of our confidence intervals.

In conclusion, significant inequalities persist in the provision of paediatric RRT throughout

Europe, most of which are explained by differences in country macroeconomics, that appear

to limit the provision of paediatric RRT particularly in the youngest patients, which are notably

challenging to treat. Considering the austerity-driven cuts in healthcare budgets experienced

by most European countries over the past few years, this poses a challenge for health care

policy makers in their aim to ensure universal and equal access to high-quality healthcare

across Europe.


41

Figure 3A. Bubble plot displaying the univariate association between country RRT incidence and longitude,

including the corresponding weighted (joinpoint) regression lines. The bubble sizes reflect the size of the

countries paediatric (patient) population and the countries influence on the regression line. Country

abbreviations were expressed by ISO2 country codes.


42

Figure 3B. Bubble plot displaying the univariate association between country RRT incidence and latitude,

including the corresponding weighted (joinpoint) regression lines. The bubble sizes reflect the size of the




43

Figure 3C. Bubble plot displaying the univariate association between country RRT incidence and GDP per

capita, including the corresponding weighted (joinpoint) regression lines. The bubble sizes reflect the size of the




44

Figure 3D. Bubble plot displaying the univariate association between country RRT incidence and public health

expenditure, including the corresponding weighted (joinpoint) regression lines. The bubble sizes reflect the size

of the countries paediatric (patient) population and the countries influence on the regression line. Country



45

Figure 3E. Bubble plot displaying the univariate association between country RRT incidence neonatal

mortality, including the corresponding weighted (joinpoint) regression lines. The bubble sizes reflect the size of

the countries paediatric (patient) population and the countries influence on the regression line. Country



46

Figure 3F. Bubble plot displaying the univariate association between country RRT incidence and the

proportion of CAKUT cases terminated during pregnancy, including the corresponding weighted (joinpoint)

regression lines. The bubble sizes reflect the size of the countries paediatric (patient) population and the

countries influence on the regression line. Country abbreviations were expressed by ISO2 country codes.


47

Appendix 1. European regions as defined by Ralph et al [73] and population data. *These countries were not

included in the original paper and were allocated based on geographic location alone.

Paediatric population per 100 000 2007 2008 2009 2010 2011 East 402.8 417.5 425.5 494.6 506.9 Albania - - - 6.3 6.0 Austria 13.0 13.7 12.6 12.4 12.3 Belarus* - 14.2 14.2 14.2 14.2 Bosnia and Herzegovina - - - - 6.1 Bulgaria - - 10.2 10.3 9.8 Croatia 6.9 6.8 6.8 6.7 6.6 Czech Republic 14.8 14.8 14.9 15.1 15.3 Estonia 2.0 2.0 2.0 2.0 2.1 FYR of Macedonia 3.8 3.7 3.7 3.6 3.6 Greece 16.0 16.1 16.2 16.3 16.3 Hungary 13.1 15.0 14.8 14.7 14.5 Lithuania* 5.3 5.1 5.0 4.9 4.5 Moldova* - - - - 5.8 Montenegro 1.2 1.2 1.2 - - Poland 59.6 58.7 58.1 57.7 58.4 Romania 31.8 31.2 31.0 30.9 31.9 Russia 212.5 212.5 212.5 212.5 212.5 Serbia 11.4 11.3 11.2 11.0 10.9 Slovakia 8.6 8.4 8.3 8.3 8.3 Slovenia 2.8 2.8 2.9 2.9 2.9 Ukraine - - - 64.9 65.1

North 44.4 44.3 44.3 44.4 44.6 Denmark 10.3 10.2 10.2 10.1 10.0 Finland 9.0 8.9 8.9 8.9 8.9 Iceland* 0.7 0.7 0.7 0.7 0.7 Norway 9.1 9.1 9.1 9.2 9.2 Sweden 15.5 15.4 15.5 15.6 15.7

West 284.3 285.1 286.5 287.1 288.5 Belgium 18.0 18.1 18.2 18.4 18.6 France 117.7 118.3 119.2 118.9 119.3 Switzerland 11.9 11.9 11.9 11.9 11.9 The Netherlands 29.5 29.3 29.2 29.1 29.0 United Kingdom 107.2 107.5 108.0 108.7 109.6

South 81.7 83.6 85.0 86.5 86.7 Malta* - - - 0.6 0.6 Portugal 16.3 16.3 16.2 16.1 15.7 Spain 65.4 67.4 68.8 69.8 70.4


48

Appendix 2. Conceptual framework hypothesizing how potential country-level factors may influence the

incidence of paediatric RRT in Europe. Macroeconomics: GDP per capita, public health expenditure; general

population characteristics: age- and gender structure, PRD distribution; Access to renal health services: number

of facilities providing RRT to children, capability to treat neonates, reimbursement rates; Perinatal care:

neonatal mortality rate, prenatal screening, abortion rate.


49

Appendix 3. Country survey.

Please identify the country you currently work in.

Please identify the centre you are currently working for.

In your opinion, does the framework contain the all the important country-level factors that may impact paediatric

RRT incidence in Europe? [If no] In your opinion, which important factor(s) are missing from the framework?

In your country, are all pregnant women offered ultrasound screening (around the 18-22 weeks) to examine the

foetus for presence of abnormalities?

In your country, are paediatric patients reimbursed at least 90% of costs for RRT treatment [HD/PD/Tx] (apart from

travel costs and medication)?

In your country, are paediatric patients reimbursed at least 90% of costs for the following medications [growth

hormone, antihypertensives, erythropoiesis stimulating agents, eculizumab, azathioprine, mycophenolate mofetil,

cyclosporine, tacrolimus, sirolimus, daclizumab, rituximab]?

In your country, are paediatric patients reimbursed at least 90% of costs for RRT related travel costs?

In your country, is it possible to provide RRT to neonates (<28 days old)? [If no] In your country, is it possible to

provide chronic RRT to infants (<1 year old)? [If no] In your country, what is the minimum age at which chronic RRT

can be started?

In your country, has it occurred in the past 3 years, that a paediatric patient did not start RRT because of the long

travel time to the clinic?

In your opinion, approximately which percentage of paediatric ESRD patients do not start RRT because of long travel

time?

Please list the centre(s) that can provide RRT to children under the age of 15.


50

Appendix 4. Aggregated country counts and rates extracted from Wiesel et al for the proportion of CAKUT

cases resulting in termination of pregnancy (TOP).

Country Births Cases Per 1000 TOP (n) TOP (%)

Austria 29026 56 1.93 5 8.9

Switzerland 18907 88 4.65 12 13.6

Denmark 8788 11 1.25 0 0.0

Spain 57523 172 2.99 36 20.9

France 60705 220 3.62 89 40.5

Croatia 10718 11 1.03 1 9.1

Lithuania 95469 84 0.88 13 15.5

The Netherlands 81980 55 0.67 4 7.3

Ukraine 44761 21 0.47 13 61.9

United Kingdom 78695 148 1.88 45 30.4

4

Mortality risk disparities in

children receiving chronic

renal replacement therapy for

the treatment of end-stage

renal disease across Europe:

An ESPN-ERA/EDTA Registry

analysis

Nicholas C Chesnaye, Franz Schaefer, Marjolein Bonthuis,

Rebecca Holman, Sergey Baiko, Esra Baskın, Anna Bjerre, Sylvie

Cloarec, Elisabeth AM Cornelissen, Laura Espinosa, James Heaf,

Rosário Stone, Diamant Shtiza, Ilona Zagozdzon, Jérôme

Harambat, Kitty J Jager, Jaap W Groothoff, Karlijn J van Stralen

The Lancet 2017 May; 389(10084): 2128-2137

ABSTRACT

Background: We explored the variation in country mortality rates in the paediatric

population receiving renal replacement therapy (RRT) across Europe, and estimated how

much of this variation could be explained by patient-level and country-level factors.

Methods: In this registry analysis, we extracted patient data from the ESPN/ERA-EDTA

Registry for 32 European countries. We included incident patients younger than 19 years

receiving RRT. Adjusted hazard ratios (aHR) and the explained variation were modelled for

patient-level and country-level factors with multilevel Cox regression. The primary outcome

studied was all-cause mortality while on renal replacement therapy.

Results: Between Jan 1, 2000, and Dec 31, 2013, the overall 5 year RRT mortality rate was

15.8 deaths per 1000 patient-years (IQR 6.4–16.4). France had a mortality rate (9.2) of more

than 3 SDs better, and Russia (35.2), Poland (39.9), Romania (47.4), and Bulgaria (68.6) had

mortality rates more than 3 SDs worse than the European average. Public health expenditure

was inversely associated with mortality risk (per SD increase, aHR 0.69, 95% CI 0.52–0.91) and

explained 67% of the variation in RRT mortality rates between countries. Child mortality rates

showed a significant association with RRT mortality, albeit mediated by macroeconomics (eg,

neonatal mortality reduced from 1.31 [95% CI 1.13–1.53], p=0.0005, to 1.21 [0.97–1.51],

p=0.10). After accounting for country distributions of patient age, the variation in RRT

mortality rates between countries increased by 21%.

Interpretation: Substantial international variation exists in paediatric RRT mortality rates

across Europe, most of which was explained by disparities in public health expenditure, which

seems to limit the availability and quality of paediatric renal care. Differences between

countries in their ability to accept and treat the youngest patients, who are the most complex

and costly to treat, form an important source of disparity within this population. Our findings

can be used by policy makers and health-care providers to explore potential strategies to help

reduce these health disparities.

Mortality risk disparities

53

RESEARCH IN CONTEXT

Evidence before this study

We searched Pubmed with the terms “paediatric”, “renal replacement therapy”, “end-stage

renal disease”, “variation”, “mortality”, “economic”, “disparities”. We set no limit for language

of publication and searched for articles published up to 22 March 2016. All relevant

publications were reviewed. International variation in mortality rates in the adult renal

replacement therapy population has previously been attributed to both country and patient

level determinants. In the paediatric population, a single study demonstrated that country

mortality rates in patients treated with peritoneal dialysis were strongly affected by country

gross national income.

Added value of this study

We describe considerable international disparities in mortality risk in the paediatric renal

replacement therapy population across 32 European countries. These disparities were largely

explained by differences in public health expenditure, which seems to limit the availability and

quality of paediatric renal care. We also demonstrate that country differences in their ability to

accept and successfully treat the youngest patients, who are the most complex and costly to

treat, formed an important source of disparity within our population.

Implications of all the available evidence

By exploring the magnitude of health-care inequalities, and by identifying both patient- and

country-level determinants, we hope to increase the awareness amongst policy makers and in

the paediatric nephrology community, and explore potential strategies to help reduce these

disparities. Considering the austerity-driven cuts in healthcare budgets experienced by most

European countries over the past few years, our results pose a challenge for health care policy

makers in their aim to ensure universal access to high-quality healthcare across Europe.


54

INTRODUCTION

Considerable international and regional variation in mortality rates has been observed in the

adult renal replacement therapy (RRT) population. Both patient and country-level factors may

explain this variation. Differences in country macroeconomics, general population mortality,

patient demographics, the distribution of cause of renal disease, the quality of renal care,

access to treatment, and attitudes regarding acceptance to and withdrawal from treatment

have been described as country-level factors explaining differences in mortality in the adult

RRT populations [20–22].

In contrast to the adult patient population, European variation in paediatric RRT mortality

rates has not previously been described due to the rarity of end-stage renal disease (ESRD) in

children and the high survival rates, which makes it difficult to provide statistically robust

estimates. Extrapolation from the adult to the paediatric RRT population is hampered by the

facts that children suffer from different underlying causes of renal disease with a

preponderance of genetic and other congenital causes, and that RRT provision to children is

resource-intensive as they are generally treated in specialized (academic) paediatric facilities by

extensive multidisciplinary teams [12].

To our knowledge, comparing mortality rates in children treated for a chronic disease has not

previously been studied on a European scale. By exploring the magnitude of health-care

inequality, and by identifying explanatory factors, we hope to increase the awareness amongst

policy makers and in the paediatric nephrology community, and explore potential strategies to

help reduce these disparities. The current paper therefore aims to 1) describe mortality rates

in the paediatric RRT population across European countries, 2) study the relationships of, and

potential interactions between, patient- and country-level factors with mortality rates, and 3)

quantify how much of the variation in mortality rates is explained by these factors.


55

METHODS

Study population

The European Society for Paediatric Nephrology / European Renal Association – European

Dialysis and Transplant Association Registry (ESPN/ERA-EDTA) was established to consolidate

data collected by European population-based national renal registries on children with end-

stage renal disease treated with renal replacement therapy. Data is collected annually in a

standardized manner on various patient- and treatment characteristics and is subject to regular

data quality checks both on the national as well as the Registry level [51]. We included

incident patients under the age of 19, starting RRT between January 1, 2000, and December

31, 2013 for 32 European countries. . Austria (AT), Belgium (BE), Croatia (HR), Denmark

(DK), Finland (FI), Greece (GR), Iceland (IS), the Netherlands (NL), Norway (NO), Spain (ES),

Sweden (SE), Switzerland (CH), and The United Kingdom (UK) provided data from January 1,

2000, to December 31, 2013. France (FR) from 2004, Czech Republic (CZ), Hungary (HU),

Lithuania (LT), Macedonia (MK), Portugal (PT), Romania (RO), Russia (RU), Serbia (RS),

Slovakia (SK), and Slovenia (SI) from 2007, Belarus (BY), Bulgaria (BG), Estonia (EE),

Montenegro (ME), and Poland (PL) from 2008, Albania (AL) and Ukraine (UA) from 2010, and

Bosnia & Herzegovina (BA) from 2011. As Germany reported only on transplant patients and

Italy reported only on patients starting on dialysis, these countries were excluded from the

analyses.

Patient and country-level determinants

The primary outcome studied was all-cause mortality on RRT. Follow-up time was censored

when renal function recovered, patients were lost to follow-up, reached the end of study, or

after 5 years of follow-up. We developed a conceptual framework (Appendix 1) describing the

hypothesized causal pathways between various country- and patient-level factors and country

RRT mortality rates (adapted with permission from Kramer et. al.[20]). Potential variables

explaining the variation in mortality between countries were defined a priori and collected on

the individual patient level as well as on the country level. On the patient level, data were

extracted from the ESPN/ERA-EDTA Registry database on patient age at RRT initiation,

gender, primary renal disease (PRD), time under treatment of a nephrologist prior to RRT,

and initial treatment modality. We assumed a non-linear association between age and mortality

based on previous reports [8, 79], an assumption which was confirmed by the analysis of our


56

own data. We therefore chose to use clinically relevant age groups, defined as starting RRT

between the ages 0-1, 2-5, 6-12, and 13-18, instead of continuous age. PRDs were classified

following the ERA-EDTA grouping of PRD codes for children[54]. Country-level data were

extracted from the World Bank Database for each country and averaged for the

corresponding years that patient data were collected (Appendix 2). Country-level data on RRT

incidence, transplantation rate, and the proportion of pre-emptive transplants was extracted

from the ESPN/ERA-EDTA Registry, whereas the number of centres providing paediatric RRT

(available for 29 countries) and reimbursement rates (available for 20 countries) were

collected through a previously conducted online survey, which is described in detail elsewhere

[93]. The number of paediatric nephrologists per million children was extracted for 29

countries from a paper by Ehrich et al [81], where a paediatric nephrologist was defined as a

paediatrician working full time in general paediatric nephrology, dialysis and paediatric renal

transplantation. There were no missing values for patient-related variables, except for the

variable “time under treatment of a nephrologist prior to RRT start” (available for 20

countries, N=2928). Analyses were restricted to complete cases only.

Statistical analysis

Crude country RRT mortality rates were calculated by dividing the number of deaths by the

number of patient follow-up years and are displayed using a funnel plot. Funnel plots allow an

objective comparison of institutional performance [94]. We compared individual countries’

mortality rates using control limits to indicate the expected limits of random deviation from

the overall European mortality rate[94]. The control limits were calculated by assuming that

the number of deaths in a country followed a Poisson distribution with parameter equal to the

overall European mortality rate multiplied by the observed number of years of follow-up in

that country. Country names are abbreviated using 2 digit ISO codes.

As country variation in mortality rates may be attributed to both country-level factors and to

country differences in the effect of patient characteristics, we adopted a multi-level approach

using a Cox regression random effects model (shared frailty model). In this model, the baseline

hazard of a country is modelled as the random effect (or frailty), and the effect of patient level

covariates are allowed to vary by country. The random effect for each country represents the

degree of deviation in mortality risk from the overall (European) mortality risk. The


57

heterogeneity in mortality risk between countries is reflected by the variance estimate of the

random effect. The variance estimate in the empty model without covariates therefore

represents the variation in country mortality risk. Adding an explanatory factor to the model,

on either the country or the patient level, allows the variance estimate to be obtained adjusted

for this factor. The proportional change in variance (PCV) after addition of an explanatory

factor to the empty model therefore allows examination of its effect on the variation in

mortality risk between countries. The PCV is calculated by simply subtracting the adjusted

variance from the baseline variance and dividing by the baseline variance[95–98].

Using the same model, we were able to estimate both the crude and adjusted hazard ratio of

each factor on country mortality. The proportional hazards assumption was graphically

checked for all variables and accepted as not violated. The regression coefficients were

standardized, removing the unit of measurement, thus improving the comparability of the

hazard ratios over multiple measures. All variables were graphically checked for violation of

the normal distribution and were not found to be severely skewed. Following an aetiological

approach, we decided on which confounders to include in the multivariate models based on

the conceptual framework (Appendix 1) and the criteria for confounding [99]. All analyses

were performed with SAS version 9.3 .

Mortality rate estimates may be imprecise in smaller countries as increased variability is

expected from smaller populations with limited follow-up time and events. As this may

introduce noise into the results, we performed a sensitivity analysis excluding countries with

less than 100 patient follow-up years; 22 out of 32 countries remained in the sensitivity

analysis dataset (Austria, Belgium, Belarus, Croatia, Czech Republic, Denmark, Finland, France,

Greece, Hungary, the Netherlands, Norway, Poland, Romania, Portugal, Russia, Serbia,

Slovakia, Spain, Sweden, Switzerland, and The United Kingdom). In addition, as most countries

report information collected from paediatric centres, older children treated in adult centres

may be missed by the registry. To avoid potential selection bias caused by age differences

between countries, we performed a second sensitivity analysis including only patients up to 14

years of age. Thirdly, as the youngest children intrinsically have the shortest time prior to

RRT, we repeated this analysis excluding children under 2 years. Lastly, as some countries had

incomplete coverage of the study period, we performed a sensitivity analysis including only


58

data from 2007 onwards and adjusted for calendar year. Results from all sensitivity analyses

yielded similar variance estimates and hazard ratio profiles and were therefore not described

in the results.

RESULTS Country mortality

Between 2000 and 2013, 365 deaths were registered in 32 European countries during 23,078

years of patient follow-up in a total of 7108 patients, the equivalent of a crude 5-year mortality

rate of 15·8 deaths per 1000 patient years. Country mortality rates (MR) ranged from 0·0 to

81·9 deaths per 1000 patient years (IQR 6·4 – 16·4 ) and are presented using a funnel plot

(figure 1). Compared to the European average mortality rate, France (MR 9·2) performed

more than 3 SDs better and The Netherlands (MR 9·4 ) performed more than 2 SDs better.

Russia (MR 35·2), Poland (MR 39·9), Romania (MR 47·4), and Bulgaria (MR 68·6) performed

more than 3 SDs worse compared to the European average, and Sweden (MR 26·0), Czech

Republic (MR 38·6), and Bosnia and Herzegovina (MR 81·9) performed more than 2 SDs

worse compared to the European average. The remaining countries did not differ from the

European average any more than explained by random variation (figure 2). The number of

deaths, follow-up years, and crude 5-year mortality rates are presented in Appendix 3.

As mortality is highest during the first year on RRT, and 5 years of follow-up was not available

for all countries, we also studied 1-year mortality rates to avoid potential bias (whilst trading

in statistical power). The crude 1-year European mortality rate was 32·6 deaths per 1000

patient years (215 deaths over 6600 years of patient follow-up). Compared to the European

average mortality rate, France (MR 20·7) performed more than 2 SDs better, Russia (MR

72·8) and Romania (MR 84·1) performed more than 3 SDs worse, and Sweden (MR 60·8)

performed more than 2 SDs worse compared to the European average (Appendix 4).

Patient-level determinants

We explored how country differences in the effect and composition of patient level

determinants would affect the variation in country mortality rates. The variation in country

mortality rates increased after adjustment for patient age at RRT initiation (21%), the time


59

under treatment of a nephrologist prior to RRT start (available for 20 countries, N=2928,

29%), and PRD (8%). Conversely, country differences in the effect of initial RRT treatment

modality reduced the variation in country mortality rates by 13%, whereas patient gender had

no effect on the variation in country mortality rates (table 1). To illustrate using the example

of patient age, Finland has a higher proportion of younger patients starting RRT (ages 0-2,

42·6%) compared to the European average (ages 0-2, 13·8%), and has a lower mortality risk in

the youngest patients relative to the European average. As a consequence, the age-adjusted

mortality risk will shift away from the European average, thus contributing to an increase in

variation of RRT mortality risk between countries. The effect of patient age on the variation in

country RRT mortality risk is visualized in figure 3.

Figure 1. Funnel plot displaying crude 5-year country mortality rates and aggregated regional mortality rates of

paediatric patient on RRT. Each country mortality rate is plotted against the number of patient follow-up years.

The latter is used to indicate the degree of reliability for the rate, as in countries with a small number of

patients, estimates may be imprecise due to increased variability in smaller populations. The 95% and 99%

control limits (which correspond to approximately 2 and 3 standard deviations, respectively) form a ‘funnel’

around the European average. Countries that fall outside these limits are doing either better or worse

compared to the European average.

0

10

20

30

40

50

60

70

80

90

0 1000 2000 3000 4000 5000 6000Mor

talit

y ra

te (

deat

hs/1

000

follo

w-u

p ye

ars)

Patient follow-up years

European average 95% control limits 99% control limits Country


60

Figure 2. Spine plot displaying country crude 5-year country mortality rates, derived from the funnel plot.

Mortality rates that lie within the central grey segment of the plot do not differ significantly from the European

average. Countries that fall outside the 95% and 99% control limits (which correspond to approximately 2 and

3 standard deviations, respectively) are performing either better or worse compared to the European average.

0 20 40 60 80

MEEEALBAMK

SILTBG

ISUASKBYRS

CZHUHRBEPT

RONO

FICHPL

DKATGRSE

NLRUESFR

UK

Mortality rate (deaths/1000 follow-up years)

<99% control limits

99%-95% control limits



>99% control limits

Mortality rate

European average


61

Table 1. Univariate and multivariate hazard ratios for patient- and country-level determinants of mortality, and

the effect of each determinant on the variation in country mortality rates. PCV = proportional change in

variance. AH= antihypertensives. GH= growth hormone.

Univariate Multivariate Variation

HR (95% CI) P-value Confounders aHR (95% CI) P-value Variance (SE)

PCV

Baseline model - - - - - 0·24 (0·11) Ref

Macroeconomics

1. GDP per capita (per1 SD increase) 0·79 (0·62-1·02) 0·07 - - -

0·18 (0·09) -25%

2. Public health expenditure (per1 SD increase)

0·73 (0·61-0·86) <·0001 1 0·69 (0·52-0·91) 0·008 0·08 (0·06) -67%

3. Private heath expenditure (per1 SD increase)

0·88 (0·73-1·06) 0·18 1 0·87 (0·74-1·03) 0·11 0·22 (0·10) -8%

Child mortality

4. Neonatal mortality rate (per1 SD increase)

1·31 (1·13-1·53) 0·0005 1,2 1·21 (0·97-1·51) 0·10 0·12 (0·07) -50%

5. Under5 mortality rate (per 1 SD increase)

1·32 (1·13-1·53) 0·0004 1,2 1·21 (0·96-1·53) 0·17 0·12 (0·07) -50%

Renal service indicators

6. Paediatric RRT incidence (per1 SD increase)

0·80 (0·63-1·02) 0·07 1,2 1·02 (0·76-1·36) 0·92 0·18 (0·09) -25%

7. Transplantation rate (per1 SD increase)

0·85 (0·70-1·03) 0·10 1,2 0·86 (0·70-1·06) 0·16 0·19 (0·09) -21%

8. Proportion pre-emptive Tx (per1 SD increase)

0·85 (0·71-1·02) 0·07 1,2 1·00 (0·76-1·31) 0·98 0·19 (0·09) -21%

9. No. centres pmcˣ (per1 SD increase)

0·91 (0·75-1·10) 0·31 1,2 0·92 (0·78-1·08) 0·30 0·23 (0·11) -4%

10. No. paediatric nephrologists pmc (per1 SD increase)*

0·93 (0·84-1·03) 0·16 1,2 0·91 (0·85-0·98) 0·02 0·25

(0·12) 4%


62

11. Reimbursement AH** (for yes) 0·98 (0·53-1·82) 0·95 1,2 1·16 (0·72-1·87) 0·53 0·30

(0·15) 0%

12. Reimbursement GH** (for yes) 0·80 (0·33-1·96) 0·62 1,2 1·49 (0·65-3·43) 0·35 0·32

(0·15) 7%

Patient factors

13. Age at RRT 1,2,3,4,17 0·29 (0·12) 21%

0-1 years 5·81 (4·34-7·78) <·0001 6·49 (4·80-8·77) <·0001

2-5 years 2·40 (1·69-3·41) <·0001 2·60 (1·82-3·70) <·0001

6-13 years 1·45 (1·06-1·98) 0·02 1·51 (1·10-2·07) 0·01

13-18 years Ref Ref Ref Ref

14. Initial treatment modality

1,2,3,4,12,17

0·21 (0·09) -13%

HD 2·91 (1·93-4·38) <·0001 2·32 (1·53-3·52) 0·001

PD 3·43 (2·29-5·12) <·0001 1·76 (1·16-2·68) 0·008

Tx Ref Ref Ref Ref

15. Time under treatment prior to dialysis (per1 SD increase)***

0·54 (0·42-0·69) <·0001 1,2,3,4,13,17 0·88 (0·67-1·17) 0·39 0·31 (0·23) 29%

16. Gender - 0·24

(0·11) 0%

Female 1·18 (0·96-1·44) 0·13 - -

Male Ref Ref Ref Ref

17. Primary renal disease

-

0·26 (0·11) 8%

Glomerulonephritis 1·29 (0·93-1·80) 0·12 - -

Cystic 1·13 (0·76-1·69) 0·54 - -

Hereditary 1·60 (1·04-2·46) 0·03 - -

Ischemic 1·91 (0·93-3·95) 0·08 - -

HUS 1·19 (0·68-2·09) 0·54 - -

Metabolic 1·65 (0·91-3·01) 0·10 - -

Vasculitis 2·17 (1·13-4·16) 0·02 - -

Miscellaneous 2·66 (1·91-3·70) <·0001 - -

Unknown 2·08 (1·46-2·97) <·0001 - -

CAKUT Ref Ref Ref Ref


63

Country-level determinants

We studied the effect of country macroeconomics, country child mortality rates, and renal

health service indicators on mortality risk and their influence on the variation in country

mortality rates (table 1). An increase in country public health expenditure was strongly

associated with a decreased mortality risk (aHR per SD increase: 0·69, 95% CI 0·52-0·91), and

explained 67% of the variation in country mortality (visualized in figure 3), whereas private

health expenditure (HR per SD increase: 0·88, 95% CI 0·73-1·06) had no significant effect. An

increase in GDP per capita (HR per SD increase: 0·79, 95% CI 0·62-1·02, p=0·07, explained

25%) showed a protective trend with mortality risk. After adjustment for patient age

distribution (not as a confounder, but as a mediator), this association reached statistical

significance (aHR per SD increase: 0·74, 95% CI 0·58-0.96, p=.02), suggesting that a lower

acceptance of high risk young patients in countries with limited resources may be somewhat

masking the relationship between GDP and mortality. Increases in both neonatal (HR per SD

increase: 1·31, 95% CI 1·13-1·53) and under 5 (HR per SD increase: 1·32, 95% CI 1·13-1·53)

mortality rates were associated with an increased mortality risk, and explained 50% of the

variation in country mortality rates. However, these latter effects were reduced after

adjustment for macroeconomic factors (aHR per SD increase neonatal mortality: 1·21, 95% CI

0·97-1·51, under 5 mortality: 1·21, 95% CI 0·96-1·53). Similarly, the protective trends found

between mortality risk and RRT incidence and the proportion of pre-emptive transplantations

were also reduced after adjustment for country macroeconomics.

Interactions between country-level and patient-level determinants

We identified an interaction between country GDP per capita and initial dialysis modality. In

the wealthiest countries, with a GDP per capita of more than $35 000 (AT, BE, CH, DK, FI,

FR, IS, NL, NO, SE, UK), there was no significant mortality risk difference between initial

dialysis modalities (HD vs. PD, aHR: 1·24, 95% CI 0·89-1·73, adjusted for age at RRT, PRD,

gender), whereas in countries with a GDP per capita of less than $35 000, patients starting

RRT on HD had an increased mortality risk compared to those starting on PD (HD vs. PD,

aHR: 1·66, 95% CI 1·19-2·30), as illustrated by the Kaplan-Meier curves in figure 4. We found

no other interactions between patient- and country level determinants.


64

Figure 3. Country unadjusted (blue), patient age adjusted (red), and public health expenditure adjusted (green)

hazard ratios (exponentiated frailties) and 95% confidence intervals.

0.1 1 10

FR

NL

PT

DK

HR

IS

HU

RS

LT

SI

AT

CH

FI

UK

UA

MK

BY

ES

AL

GR

EE

SK

ME

NO

BE

BA

CZ

SE

BG

PL

RU

RO

Hazard Ratio

Crude HR Age adjusted HR Public health expenditure adjusted HR


65

DISCUSSION

Considerable variation exists in mortality rates of children treated with renal replacement

therapy across Europe. Most of this variation was attributable to an increased mortality risk in

several larger Eastern European countries compared to Northern, Southern, and Western

European countries, where the mortality risk was mostly similar. The current study provides a

novel disentanglement of the explanatory effect of both country and patient level factors on

differences in country mortality risk. We demonstrate that country differences in the effect

and distribution of patient characteristics, such as age at RRT onset, may conceal the true

variation in country mortality risk, which we found to be primarily attributable to disparities in

public health expenditure.

Economic welfare is a key determinant of health and access to health services. The effect of

country macroeconomics on country RRT mortality risk in our population is understandable

given the complexity and costs involved in the provision of renal care to children by a multi-

professional paediatric team. Restricted public healthcare financing in particular is detrimental

to the survival probabilities of children on RRT, whereas private, or out-of-pocket, health

expenditure has little effect on mortality risk. This is not unexpected considering that the

majority of paediatric patients are treated in (public) academic centers, and that most of the

direct costs are fully reimbursed [93]. We also demonstrate an indirect effect of country

macroeconomics on country mortality risk, as the effects of various renal service indicators

and child mortality rates were attenuated after adjustment for macroeconomic indicators.

Schaefer et al. previously demonstrated that country mortality rates in paediatric patients

treated with PD were strongly affected by country gross national income (GNI), independent

of patient age and the presence of comorbidities [23]. They included developing countries, and

thus a wider range in both country GNI and mortality rates compared to our analyses of

European middle- and high income countries. Nonetheless, despite the smaller range in

country macroeconomics in our study, we demonstrate that health financing disparities across

Europe are still adversely affecting mortality risk in the paediatric RRT population.

Interestingly, the opposite effect of macroeconomics was previously demonstrated in the adult

dialysis population, where a higher country GDP per capita and healthcare expenditure were

associated with an increased country RRT mortality risk [20]. The authors attributed this

association to a higher acceptance of patients with a poor health condition in wealthier


66

countries, and that increased health care spending does not necessarily result in more effective

care. In children, where favourable macroeconomic conditions will also promote the inclusion

of younger patients with severe comorbidities in RRT programs, resource spending appears to

be more effective in terms of promoting patient survival.

The amplification of RRT mortality risk variation after adjusting for patient age demonstrates

that countries differ in their ability to accept and successfully treat the youngest children, who

are typically the most complex and costly to treat. We previously established a higher

incidence of RRT in wealthier countries due to the acceptance of younger patients [93]. One

may therefore expect a higher mortality rate in wealthier countries, as the youngest patients

bear the highest mortality risk. Here we demonstrate the opposite, finding higher survival

rates in wealthier countries despite the acceptance of younger and presumably more medically

complex patients (i.e. Finland). Vice versa, lower survival rates were found in countries

burdened with economic constraints, despite the lower acceptance of younger patients [79,

93]. This finding is a cause for concern, as non-acceptance to RRT implies an underestimation

of ESRD mortality (as these patients go unregistered), thus further exacerbating the

inequalities in care and mortality caused by economic disparities.

Interestingly, the high survival rate in pre-emptively transplant recipients was similar in both

the wealthiest and less wealthy countries. Furthermore, in the wealthiest European countries,

we found no significant difference in mortality risk between initial dialysis modality, whereas in

the less wealthy countries, patients starting RRT on HD had a significantly worse survival

compared to those starting on PD. This suggests that the majority of excess mortality found in

poorer countries occurs predominantly in the haemodialysis population. This may be due to

either a poorer performance on HD in these countries, or that patients are sicker at

treatment initiation and are therefore started on HD.

Child mortality rates reflect the health of the general paediatric population, as well as the level

of economic development (for which we adjust in our multivariable analyses), and the

accessibility and quality of paediatric (and obstetric) health services. Country-specific child

mortality rates were associated with mortality on RRT and explained a large portion of the

variation in country RRT mortality risk. This possibly reflects the impact of the quality of


67

paediatric health systems on the effectiveness of paediatric RRT care, as well as how the

intrinsic mortality risk in the general paediatric population affects mortality on RRT [100].

However, the association was weakened after adjustment for macroeconomic factors,

suggesting that the quality of country’s paediatric health care system is - to some extent -

reliant on country wealth and health expenditure.

The variation in paediatric RRT mortality rates was limited across Western, Northern, and

Southern European countries. The majority of variation in mortality rates across Europe was

attributable to several larger Eastern European countries, where patients had a significantly

higher mortality risk compared to the European average. Since the fall of communism, many

Eastern European countries have undergone dramatic changes in health care systems and

financing, and have achieved substantial progress regarding the availability and effectiveness of

renal services [81, 93, 101, 102]. Although the gap between Western and Eastern Europe has

narrowed progressively over the past decades, many countries in Eastern Europe remain

burdened under stringent austerity measures, limited health care budgets, and higher child

mortality rates; factors which we demonstrate here to strongly affect the RRT mortality risk

on a country level. Furthermore, after their accession to the EU, many Eastern and Central

European countries experienced an outflow of health professionals to higher-income

countries, and a consequent loss in educational health care investment [103–105]. This may

cause larger problems in the future, given the inverse association found between RRT

mortality and the number of paediatric nephrologists. In support of this premise, we found a

positive trend (p=0·06, independent of patient age) between the number of paediatric

nephrologists in a country and the time under treatment of a nephrologist prior to RRT (as a

marker for timely referral and speed of disease progression).

The study strengths include its large sample size, low loss to follow-up, and the European-wide

inclusion of patients. Additionally, patient variables such as gender, primary renal disease, and

treatment modality are collected by the registry in a standardized manner using an uniform

coding system. Likewise, country level indicators are collected through umbrella organizations

such as the World Bank for the purpose of country comparisons. Furthermore, the funnel-plot

and frailty model techniques both provide a novel approach to describe and explain variation

in paediatric RRT survival between countries. An important limitation is the lack of data


68

regarding children with ESRD who were not registered due to various reasons; 1) patients

who were not accepted on RRT (i.e. in children with severe comorbid conditions and a

perceived unacceptable quality of life), 2) patients who died prior to treatment initiation, or 3)

patients who did not fulfill the national registry inclusion criteria. Other limitations include the

incomplete coverage of the study period for several countries (although including data from

2007 onwards had no meaningful effect on the estimates), and a low number of events and

follow-up time in smaller countries, which may impact the reliability of our country mortality

rates due to random variation. Unfortunately, Germany and Italy were excluded from the

study due to the fact that either transplant or dialysis patients are exclusively registered, and

not the full RRT population. As transplant patients have higher survival rates compared to

dialysis patients, including these countries would have introduced bias to the results. Lastly,

multiple testing may form an issue in table 1, where the association between mortality risk and

multiple indicators is tested, however, even if we were to use a more conservative p-value (for

instance 0.01 instead of 0.05), this would not alter the interpretation of the results.

Although all European Union Member States have made commitments towards reducing

inequalities in access to health care and in health outcomes, considerable international

variation persists in mortality rates in the paediatric RRT population across Europe, most of

which was attributable to an excess mortality risk for patients treated in several Eastern

European countries. The majority of this variation was explained by disparities in public health

expenditure, which seems to limit the availability and quality of paediatric renal care.

Moreover, country differences in their ability to successfully treat the youngest patients, who

are typically the most complex and costly to treat, seemed to be an important source of

disparity within Europe. These results call for improvements to be made on both clinical- and

policy-levels to reduce inequities in RRT mortality rates. To achieve this, we advocate further

standardization of treatment guidelines and medical training for paediatric nephrologists across

Europe, for example through the European Society for Paediatric Nephrology (ESPN)

recommendations for the training of Paediatric Nephrologists, information exchange through

international fellowships, and the provision of Continuing Medical Education courses [106].

Furthermore, in line with previous EU commitments, we recommend that national and

European policy-makers involved in health care financing should pursue a uniform and high


69

quality of paediatric renal care across Europe, although this may prove challenging with varying

national health priorities, especially in times of austerity.

Figure 4. Adjusted Kaplan-Meier plot for survival by initial treatment modality, stratified by GDP per capita,

using covariate values, PRD group = CAKUT, gender = male, age group at RRT initiation = 6-12 years. HD =

haemodialysis, PD= peritoneal dialysis, Tx = pre-emptive transplantation.


70

Appendix 1. Conceptual framework describing the hypothesized causal pathways between various country

indicator groups and country RRT mortality rates, with macroeconomic indicators at the highest hierarchical

level. We hypothesized that macroeconomics may affect RRT mortality rates through two main pathways. The

first pathway assumes that macroeconomics dictate both organizational and quality aspects of paediatric renal

care, for instance by influencing reimbursement rates (organizational) or the number of available treatment

facilities (organizational), as well as the availability of a specialized multidisciplinary team (quality) or access to

expensive medications such growth hormone therapy (quality). We hypothesized that organizational aspects of

renal care may determine the characteristics of the paediatric RRT population in a country, such as the

proportion of patients that receive a pre-emptive transplant or the average patient age at start RRT. The

second pathway recognizes economic welfare as a key determinant of general population health, which

consequently contributes directly to the intrinsic mortality risk in the paediatric RRT population.


71

Appendix 2. Description of country-level determinants.

Country Indicator

Description

GDP per capita

Gross domestic product (GDP) per capita based on purchasing power parity (PPP), is a measure for country wealth. The PPP method allows for the international comparison of economies.

Public health expenditure Public health expenditure consists is expressed as the percentage of national GDP that a government spends on health care.

Private health expenditure Private health expenditure includes direct household (out-of-pocket) spending, and private insurance, expressed as a percentage of national GDP.

Neonatal mortality rate Neonatal mortality rate is the number of neonates dying before reaching 28 days of age, per 1000 live births.

Under 5 mortality rate Under 5 mortality rate is the number of children dying before reaching 5 years of age, per 1,000 live births.

Paediatric RRT incidence Age-adjusted RRT incidence per million children under the age of 15.

Transplantation rate The number of transplantations, pre-emptive or otherwise, expressed per million children.

Proportion of pre-emptive transplantations

The percentage of patients receiving a pre-emptive transplantation, as a proportion of all incident patients.

Centres providing paediatric RRT

The number of centres providing paediatric RRT, expressed per million children.

Number of paediatric nephrologists

The number of paediatric nephrologists per million children.

Reimbursement of anti-hypertensive medications

Reimbursement of anti-hypertensive treatment, defined as >90% reimbursement of costs.

Reimbursement of growth hormone treatment

Reimbursement of growth hormone treatment, defined as >90% reimbursement of costs.


72

Appendix 3. Country 5-year mortality rates.

Country N Deaths Follow-up years Mortality rate

AL 6 0 13·2 0·0

AT 203 8 665·1 12·0

BA 18 2 24·4 81·9

BE 110 6 348·8 17·2

BG 34 5 72·9 68·6

BY 55 2 149·0 13·4

CH 141 6 536·7 11·2

CZ 70 6 155·6 38·6

DK 174 5 607·9 8·2

EE 3 0 8·9 0·0

ES 759 38 2768·6 13·7

FI 148 6 523·7 11·5

FR 1094 31 3383·3 9·2

GR 227 11 764·3 14·4

HR 82 2 318·3 6·3

HU 73 2 236·1 8·5

IS 23 0 89·0 0·0

LT 25 0 61·3 0·0

ME 3 0 2·6 0·0

MK 8 0 26·4 0·0

NL 460 15 1589·6 9·4

NO 129 7 466·2 15·0

PL 224 24 601·5 39·9

PT 129 3 390·0 7·7

RO 199 22 463·7 47·4

RS 47 1 152·0 6·6

RU 582 61 1733·9 35·2

SE 270 22 844·7 26·0

SI 17 0 55·3 0·0

SK 47 2 127·4 15·7

UA 32 1 92·5 10·8

UK 1716 77 5804·9 13·3


73

Appendix 4. Spine plot displaying country crude 1-year country mortality rates, derived from the funnel plot.

Mortality rates that lie within the central grey segment of the plot do not differ significantly from the European

average. Countries that fall outside the 95% and 99% control limits (which correspond to approximately 2 and

3 standard deviations, respectively) are performing either better or worse compared to the European average.

0 20 40 60 80

MEEEAL

MKBASI

LTIS

BGUARSSKBYCZHUHRBE

NOPT

CHFI

RODKATPL

GRSE

NLRUESFR

UK

Mortality rate (deaths/1000 follow-up years)

<99% control limits




>99% control limits

Mortality rate

European average

5

Survival in children

requiring chronic renal

replacement therapy

Nicholas C Chesnaye, Karlijn J van Stralen, Marjolein Bonthuis,

Jérôme Harambat, Jaap W Groothoff, Kitty J Jager

Pediatr Nephrol 2017 May; Epub ahead of print

ABSTRACT

Survival in the paediatric end-stage renal disease (ESRD) population has improved substantially

over the past decades. Nonetheless, mortality remains at least 30 times higher than that of

healthy peers. Patient survival is multifactorial, dependent on various patient and treatment

characteristics, as well as on the degree of economic welfare of the country in which a patient

is treated. In this educational review we aim to delineate the current evidence regarding

mortality risk in the paediatric ESRD population, and provide paediatric nephrologists with an

up-to-date knowledge base required to counsel affected families.

Survival in paediatric RRT

77

INTRODUCTION

Approximately 9 out of every million children under 20 years of age in the developed world

require renal replacement therapy (RRT) for the treatment of end-stage renal disease (ESRD)

[5]. Mortality risk in these children is multifactorial, owing to the complex nature and multiple

causes of ESRD in this population, and is at least 30 times higher than that of healthy peers

[79, 107]. Although other patient-related outcomes such as growth, psychosocial development

and quality-of-life are of major importance, prolongation of patient survival may be arguably

the most relevant clinical goal. As ESRD in children is a rare condition, the statistical power

needed to accurately assess (risk factors related to) survival has been limited. Over the past

years, various (inter)national registries have been instrumental in providing sufficient data to

advance epidemiological research, and expand the evidence regarding outcomes and treatment

guidelines for this population. In this review, we aim to delineate the current evidence base

regarding mortality risk in the paediatric RRT population, and provide paediatric nephrologists

with up-to-date data to counsel affected families.

IMPROVEMENTS IN PATIENT SURVIVAL

Since the introduction of the first paediatric chronic RRT programs during the 1960s,

substantial advances in renal medicine have been achieved (Box 1) [108, 109], and survival has

improved significantly, especially in the youngest patients. Historic registry data from Australia

and New Zealand (ANZDATA registry) cite a 10-year mortality rate of 110 deaths per 1000

patient years during the 1960s, which was halved with each subsequent decade, stabilizing at

18 deaths per 1000 patient years during the 90s [8]. In European dialysis patients, the 5-year

mortality risk decreased by 36% from 1980-1984 to 1995-2000, and by 79% in the subgroup of

patients aged 0-4 years [68]. In the US, dialysis survival improved during 1990-2010, with each

5-year increment decreasing mortality by 12% decrease in children over 5 years of age, and by

20% in children under 5 years [34]. In neonates and infants initiating dialysis, the majority now

survive long enough on dialysis to reach the minimum age and body weight required for

successful transplantation [86, 110].


78

Post-transplant survival has also improved over time. The mortality risk in European first renal

transplant recipients decreased by 42% for the period 1995-2000 compared with 1980-1984

[68]. Between 1990-2010 in the US, each additional calendar year led to a 3% decrease in

mortality risk, which was 5% for children under 5 years of age. Improvements were most

pronounced during the first year post-transplant [111]. The 5-year survival for deceased donor

recipients improved from 91.2% during 1987-1995 to 96.4% during 2005-2013, and from

95.1% to 97.1% for living donor recipients [11].

Presently, the overall 5-year survival for paediatric RRT patients is approximately 90%, and is

similar across high-income countries (Table 1). In Europe, survival currently ranges from 82%

to 96% at 10 years, and from 76% to 89% at 20 years. Long-term survival probabilities for

European patients are presented by age group and initial treatment modality in table 2

(personal communication; Anneke Kramer, 25 January 2017).

Box 1. Key developments in paediatric renal medicine

The introduction of continuous ambulatory peritoneal dialysis [100]

The development of haemodiafiltration [101]

The development of home HD programs [102]

The introduction of portable PD devices [103]

Improvements in pre-dialysis care [104]

Introduction of the “Y-set” catheter connection for PD [105]

The use of bicarbonate-buffer for dialysis [106]

The addition of amino acids to dialysate [107]

The development of erythropoietin [108]

The development of growth hormone therapy [109]

An increased percentage of pre-emptive Tx [90]

Innovation of immunosuppressive drugs [110]

Improvements in nutrition [111]


79

Table 1. Five-year crude survival probabilities of paediatric RRT patients by country and period [35, 62, 66, 79,

112, 113]. 1 Four-year survival probability; 2 Incident dialysis patients only.

Country/Area Period Survival

Australia and New Zealand 1963-2002 83%

United States 2004-2008 89%

Canada 1992-2007 92%

Europe1 2009-2011 94%

Japan 2006-2011 92%

Taiwan2 1995-2004 88%

Table 2. Long-term crude survival for patients initiating RRT between 1990 and 2014 by age group and initial

treatment modality, using European Renal Association – European Dialysis and Transplant Association (ERA-

EDTA) data for the countries Austria, Bosnia and Herzegovina, Denmark, Spain, Finland, France, Greece,

Iceland, The Netherlands, Norway, Romania, Serbia, Sweden, and Scotland (personal communication; Anneke

Kramer, 25 January 2017). HD = haemodialysis, PD = peritoneal dialysis, Tx = transplantation.

5-yr 10-yr 15-yr 20-yr

Overall 94% 90% 87% 83%

Age

0-1 85% 82% 79% 76%

2-5 92% 88% 83% 81%

6-12 95% 93% 90% 85%

13-18 95% 92% 88% 85%

First RRT modality

HD 94% 90% 86% 82%

PD 92% 88% 85% 82%

Tx 97% 96% 93% 89%


80

FACTORS ASSOCIATED WITH MORTALITY

Age

Age at dialysis initiation is a key determinant of patient survival. Registry data has consistently

shown that, compared with adolescents, mortality risk is approximately 4 times higher in

children < 5 years of age at dialysis initiation, and 1.5 times higher in children aged > 5 years of

age [8, 34, 79, 114]. Mortality risk remains the highest in neonatal and infant dialysis patients

[114, 115], whom are technically challenging to treat due to small body size, a high risk of

infection, difficulties in nutrition and growth, and a high prevalence of (severe) comorbidities

[116, 117]. These challenges and a perceived unacceptable quality of life form important

factors in the decision to withhold or withdraw treatment in some of these children [117–

120]. Moreover, transplantation is often not feasible due to the small size of the child relative

to the large donor kidney, and is usually recommended after reaching a minimum of 18

months of age or a weight of 10 kg. Growth retardation, which is highly prevalent in these

children, delays reaching the recommended transplant weight/age, thus further delaying

transplantation and increasing time on dialysis, which in turn increases the mortality risk in this

already vulnerable population [117, 121]. Nonetheless, relatively good clinical outcomes have

been reported and survival has improved significantly in this group. An international

collaboration recently demonstrated a 5-year survival of 76% and a transplant probability of

55%, concluding that relatively good survival may be achieved in neonates, despite the high

prevalence (73%) of comorbidities [86].

Sex

No studies have specifically investigated a possible effect of sex on mortality in the paediatric

ESRD population, but girls seem to have a higher mortality risk than boys [107]. In the US,

girls over the age of 5 on dialysis had a 27% increased mortality risk compared with boys,

although this effect was less pronounced in younger children [34]. In the US transplant

population, girls had a 18% higher cardiovascular-related and an 37% higher infection-related

mortality risk compared with boys [36]. A potential explanation was suggested by a European

study demonstrating a 23% decreased probability of pre-emptive transplantation in girls

compared with boys. This disparity was mostly explained by the fact that girls tended to

progress faster to ESRD compared with boys and by differences in age and primary renal


81

disease distribution. Other potential non-medical factors such as patient, parental, and

physician attitudes towards transplantation may also play a role [122].

Race

Patient race has also been shown to affect mortality risk in the paediatric RRT population. In

the US, black race was associated with a 25% higher risk of death compared with white race in

first transplant recipients [123], and a 64% higher risk of death in dialysis patients. The

likelihood of transplantation was also lower in both black and Hispanic dialysis patients [124].

Furthermore, black children were 1.6 times more likely to die from cardiovascular causes

before the age of 30 compared with white children [125]. The former has been attributed to a

higher incidence of hypertension, arrhythmia, cardiomyopathy, and valvular heart disease in

black patients [126, 127]. Also in CKD stages 1-3, black children were more likely to have

elevated systolic and diastolic blood pressure compared with non-black children [128]. In

Europe, black and Asian patients were less likely to receive a transplant, and Asian patients had

a 2.5-fold higher mortality risk compared with white patients [129]. The latter was reduced

after adjustment for primary renal disease, suggesting that differences in renal disease

distribution between races explains a part of these disparities.

Primary renal disease

Congenital anomalies of the kidney and urinary tract (CAKUT) and glomerulonephritis form

the most common aetiologies of renal disease in children, accounting for at least half of all

paediatric ESRD patients [79, 113]. Patients suffering from CAKUT have the best survival

probabilities of all primary renal disease groups, although survival varies by aetiology [34, 115,

130]. In infants and neonates, those with renal hypo/dysplasia, congenital nephrotic syndrome,

polycystic disease, and other/unknown had a 2 to 4 times increased mortality risk compared

with those with obstructive uropathy [110]. Poor patient survival has also been described in

patients with secondary glomerulonephritis, vasculitis, systemic lupus erythematosus, and

primary hyperoxaluria [130–132].


82

Anthropometry

Children that are either underweight or obese at ESRD onset have an increased mortality risk.

In the US, this U-shaped association was demonstrated in both dialysis and transplant patients,

with mortality risk increasing by 26% for every 2 SD increase or decrease from the 0.5 BMI

SDS reference value [133]. In children with a high BMI, volume overload, edema, or

comorbidity may explain the increased mortality risk. In underweight children, disease severity

and malnutrition may be accountable. Low serum albumin (<3.5 g/dL), a marker for

malnutrition or inflammation, was indeed associated with a 90% increased risk of death [33].

Similarly, Ku et al. found that both obese (17% increase) and underweight (26% increase)

children were at increased risk of mortality. Interestingly, they found that obese children were

less likely to receive a transplant especially from a living donor, and that this attenuated their

increased mortality risk [134].

Growth failure in the paediatric RRT population may reflect disease severity and is associated

with increased mortality [135]. In the US, every SDS decrease in height increased mortality

risk by 14%. This effect was particularly evident in children under 14 years of age, but was

similar across treatment modalities [133]. A NAPRTCS study echoed these results,

demonstrating that mortality risk was twice as high in children initiating dialysis with a height

SDS of less than 2.5 compared with children with a normal height [136]. More recently in the

US, both short (<3rd percentile) and tall (>3rd percentile) stature at RRT initiation were

associated with an increased risk of death compared with less extreme heights, although the

latter was limited to a small group of children with an elevated BMI (>95th percentile) and

white race [137].

Comorbidity

Extra-renal comorbidity is common in the paediatric ESRD population. The UK Renal Registry

reported that at the onset of RRT in 2009-2013, 19.3% of paediatric patients had at least one

comorbidity, and 9.5% had two or more comorbidities. Syndromic diagnosis (8%),

developmental delay (7%), and congenital abnormality (7%) were the most frequently reported

comorbidities [138]. Multiple studies have shown that the presence of comorbidity is an

important predictor of mortality [34, 139], especially in patients with cognitive (5-year survival

probability of 63%), cardiac (73%), and pulmonary (50%) abnormalities [140]. In a single-centre


83

study from the UK, 76% of the dialysis patients who died had a comorbid condition, resulting

in a 7.5 times increased mortality compared with those without comorbidities [141]. Several

studies have shown that particularly the youngest patients with co-morbid conditions have an

increased mortality risk, especially those with pulmonary hypoplasia [142–145].

RRT modality

It has been well established that (pre-emptive) renal transplantation offers better survival

probabilities compared with dialysis [8, 146]. Nonetheless, approximately 80% of paediatric

patients will initiate RRT on dialysis to bridge the preparation time needed for transplantation,

or will require dialysis after graft loss [79]. Survival comparisons by dialysis modality in a

randomized clinical trial (RCT) setting have proved extremely difficult [147]. Consequently,

survival comparisons remain reliant on observational studies [33–35, 148–150]. In adults, there

seems to be a consistent trend showing a survival advantage during the first few years on PD,

especially in younger, healthier, and non-diabetic patients [27–32, 151]. In the paediatric

dialysis population, recent registry data from Europe and the US demonstrate a 21%-32%

reduced mortality risk in children initiating dialysis on PD [34, 148, 149]. In the US, this

treatment effect was only present in children younger than 5 years of age, whereas in Europe

this effect was less pronounced in children younger than 5 years and absent in infants [34,

148]. Furthermore, European data show that this treatment effect was stronger during the

first year of dialysis, in older children, and in those with a short time under treatment of a

nephrologist prior to starting dialysis (figure 1). As the latter may serve as a proxy for timely

referral and the speed of disease progression, this may prelude to indication bias due to

unmeasured case-mix confounders, as sicker patients are more likely selected to start dialysis

on HD [148].

Time on RRT

Time spent on dialysis has been shown to impact mortality risk, which is highest during the

first year of treatment, and reflects the intrinsic mortality risk of initiating dialysis. In the US,

mortality rates reach 48 per 1000 patient years during the first month, peak during the second

month of dialysis at 57, then slowly decrease to 28 during months 9-12. Rates of mortality due

to cardiovascular disease and infection show similar patterns [152].


84

The duration of living with a functioning graft has been shown to decrease patient mortality

risk. In the US, in first transplant recipients, mortality was highest during the first post-

transplant year, after which mortality risk decreased (albeit not significantly) by 1% for each

additional follow-up year. This effect was stronger for cardiovascular-specific mortality, which

decreased by 16% for each follow-up year, suggesting that transplantation has no cumulative

negative effect on the cardiovascular health in young recipients. However, returning to dialysis

after graft failure was associated with a 4.4-fold increase in overall mortality risk and a 7.8-fold

increase in cardiovascular mortality risk [36].

Residual renal function

In adult dialysis patients, a decrease in residual renal function has been associated with an

increase in mortality risk [153, 154]. Data is lacking in the paediatric population. Two single-

centre US studies demonstrated that infants with oligoanuria had a higher mortality risk

compared with infants with residual renal function [144, 145], and others have demonstrated a

positive effect of residual renal function on growth and nutrition [155–157].

GFR at RRT initiation

The literature discussing the relationship between GFR at dialysis initiation and mortality risk

in adults is conflicting [158–160], and this question has not yet been studied in children,

although a study from the US found that children with a higher GFR at dialysis initiation had a

decreased risk of hospitalization for hypertension and pulmonary edema [161]. A single RCT

has tackled this question in adults, finding no difference in survival between late and early

starters, although the difference (2.2 ml/min/1.73m²) in GFR between groups was smaller than

anticipated. Nonetheless, dialysis initiation was delayed by 6 months amongst the late starters,

which is favourable for both patients and costs [162].


85

CAUSES OF DEATH

Cardiovascular disease and infection-related mortality

Cardiovascular disease (CVD) and infection-related mortality are the major causes of death in

the paediatric RRT population, accountable for approximately 30% and 20% of deaths,

respectively, although these rates vary strongly by country, age, race, the definition used, and

treatment modality [70, 79, 125, 152]. In Europe, infections were the leading cause of death in

those on PD and those with a functioning graft, whereas cardiovascular causes of death

dominated in patients on HD [79]. In the US, a 4.5 times increased risk of CVD death in

dialysis patients compared with transplant recipients has been reported [125]. An increased

CVD mortality risk for dialysis patients was also cited in Australia and New Zealand, where

between 1963-2002, CVD death accounted for 57% of deaths in children on HD, 43% in

children on PD, but only for 30% in those with a functioning transplant [8]. Both CVD and

infection-related mortality have decreased over the past decades in the US [34]. Vogelzang et

al. studied changes in causes of death in adults after long-term RRT since childhood in the

Netherlands, finding that CVD mortality risk had decreased by 91% since the 70s, whereas

infection-related mortality risk had doubled over time. The decrease in CVD mortality was

attributed to an increased awareness amongst nephrologists of the burden of cardiovascular

disease and a subsequent strict cardiovascular management in these patients [69].

Malignancy-related mortality

Malignancy-related death occurs more often in transplant recipients compared with those on

dialysis, and is likely caused by an impaired tumour immune surveillance due to

immunosuppression [36, 107, 163–165]. In Australia and New Zealand, malignancies

accounted for 14% of deaths among transplant recipients, compared with only 1% and 2%

percent of deaths among patients on HD and PD, respectively, with most deaths occurring

after 10 years of RRT [8]. Furthermore, paediatric transplant recipients had a 15-30 times

increased risk of developing a malignancy compared with the general population [166]. In the

Netherlands, 30 years after paediatric transplantation, 41% of survivors had developed cancer,

and 31% had developed a second de novo cancer during the first year after initial diagnosis.

Malignancies were responsible for 13% of all deaths in the cohort. The overall incidence of

malignancy was more than 20-fold higher compared with the general population with a notable

increase in risk starting after 20 years of follow-up [167].


86

INTERNATIONAL DISPARITIES IN SURVIVAL

As economic welfare is a key determinant of health and access to health services, in low and

middle income countries the provision of chronic RRT is fraught with challenges. The

complexity and cost involved in the provision of renal care to children, a lack of financial and

human resources, different health priorities, and an inadequate health infrastructure have

obvious consequences for access to RRT and the survival probabilities of patients in these

countries [168, 169]. In 2010, at least half of the 4.9 million people requiring RRT worldwide

died prematurely because they did not have access to treatment [3]. Specifically in children, it

has been suggested that possibly no more than 10% of those requiring RRT have access to

treatment, and that most of these preventable deaths occurred in low- and middle- income

countries [170]. The few studies available in lower-income countries, where renal registries

are often lacking, confirm these disparities. In Jamaica, between 2001-2006, of all ESRD

patients under 12 years of age at diagnosis, 62.5% died due to restricted access to RRT [171].

In a tertiary hospital in South-West Nigeria, between 2005 and 2012, the median survival time

of 51 admitted paediatric ESRD patients was only 47 days. Of these, 82% had received an

acute dose of dialysis, however, continuation of RRT was not possible due to financial

constraints, likely resulting in death shortly after discharge [172]. In two tertiary hospitals in

Vietnam, between 2001 and 2005, only 27% of admitted paediatric ESRD patients received

RRT. The remainder were treated conservatively due to a lack of financial resources [173]. In

a tertiary care hospital in India, 61% of admitted paediatric ESRD patients were either treated

conservatively or opted against further treatment due to the high cost of RRT, likely resulting

in death [174]. As only a fraction of children requiring RRT globally actually receive treatment,

and an equitable and universal provision of costly RRT is unrealistic in the short term, the

largest gains in survival are likely to be made by delaying progression of CKD and preventing

ESRD [168, 175].

Even amongst high- and middle-income countries, survival probabilities of paediatric RRT may

vary. We recently demonstrated that considerable international variation exists in mortality

rates across Europe, mostly attributable to an excess mortality risk for patients treated in

several Eastern European countries. Most of this variation was explained by disparities in

country public health expenditure, which limits the availability and quality of paediatric renal

care services. In addition, differences in a country’s ability to accept and successfully treat the


87

youngest children, who are the most complex and costly to treat, formed an additional source

of disparity within Europe. Economic constraints in Europe were also associated with a lower

incidence of RRT [93]. As non-acceptance to RRT implies an underestimation of ESRD

mortality (as these deaths go unregistered), the inequalities in mortality caused by economic

constraints will be exacerbated. In addition, considerable country variation persists in

transplant rates, donor source, and time on the transplant waiting list which, given the

beneficial effect of transplantation, will affect patient survival indirectly [42].

RECOMMENDATIONS FOR LONG-TERM FOLLOW-UP

THROUGH ADULT LIFE

The increased mortality risk of paediatric onset ESRD carries on through adulthood, with life

expectancy reduced by 40–50 years in dialysis patients and by 20–30 years in transplant

patients [176]. Cardiovascular disease is highly prevalent amongst young adults after a lengthy

exposure to RRT, but has been shown to be reversible [177–180]. Strict monitoring of

cardiovascular disease, and intensified antihypertensive and antilipaemic therapy should

therefore be a priority in this population. Furthermore, as the majority of paediatric onset

ESRD patients will have received a transplant prior to transitioning to adult care, continued

compliance to immunosuppression regimens is of the upmost importance, especially given that

up to 53% of adolescents have been reported to be non-compliant [181–183]. Moreover, due

to prolonged exposure to immunosuppression in these patients, adult nephrologists should be

attentive to the increased risk of infections, and the development of skin cancers 10-15 years

post-transplantation.

KNOWLEDGE GAPS

National and international registries for paediatric RRT have been instrumental in describing

survival and establishing factors associated with mortality in this population. However, data

from middle- and lower-income countries remain scarce. The forthcoming IPNA Registry aims

to consolidate existing registry data and fill in the gaps by collecting data globally [184].

Worldwide reporting of paediatric RRT is essential in order to reveal international disparities

regarding treatment and mortality rates, increase the awareness regarding these disparities in

the paediatric nephrology community, and provide the evidence required to advocate policy

change and inform budgetary decisions on various levels of government.


88

Furthermore, although associations between mortality and various patient and treatment

related factors have been studied in the adult RRT population, simple extrapolation of these

results to children is often not valid given the differences in disease aetiology and progression.

Small samples sizes and a low number of adverse events often impede epidemiological

research in the paediatric RRT population. Nonetheless, with continued support and

commitment, the volume of registry data will increase over time, hopefully enabling studies to

fill in the knowledge gaps concerning the determinants of mortality, specifically in the

paediatric RRT population [185].

LIMITATIONS

Several factors limited our ability to investigate the mortality risk in the paediatric RRT

population. First, children with ESRD who are not accepted on RRT, or died prior to

treatment initiation, are not registered. Second, in registries patients are frequently lost to

follow-up when transferred to adult care, precluding registration of premature death during

(early) adulthood, and thirdly, studies often focus on mortality risk on either dialysis or

transplantation instead of throughout the entire RRT trajectory. Lastly, in contrast to adult

patients, virtually all children with ESRD are considered transplantable. Consequently, long-

term dialysis studies are scarce and subject to negative selection of non-transplantable

patients.

SUMMARY

Patient survival has improved substantially over the past decades in both the paediatric dialysis

and transplant population, especially in the youngest patients. First and foremost, as global

disparities persist in the provision of paediatric renal care, patient survival is primarily

dependent on access to treatment. In patients receiving RRT, survival is largely dependent on

country health expenditure, disease aetiology, patient age, transplant feasibility, growth failure,

sex, BMI, race, and the presence of comorbidities.


89

Figure 1. Cumulative incidence plots by dialysis modality and A) age group at the start of renal replacement

therapy, and B) the time under treatment by a nephrologist. Reproduced with minor modification and

permission from Kidney International [148]. HD = haemodialysis, PD = peritoneal dialysis.


90

SUMMARY BOX

Patient survival has improved substantially over the past decades in both the dialysis

and transplant population, and although the youngest patients bear the highest

mortality risk, they also show the greatest improvement in survival over time.

Patient survival is multifactorial, largely dependent on access to treatment, country

health expenditure, disease aetiology, patient age, transplant feasibility, growth failure,

sex, BMI, race, and the presence of comorbidities.

Although comparisons between dialysis modalities are hindered by selection bias and

residual confounding, patients selected to start dialysis on PD seem to have an initial

survival advantage over those starting on HD.

Global disparities persist in the provision of RRT and outcomes in the paediatric

ESRD population, even amongst middle- and higher income countries.

Demographics

6

Mortality Risk in European

Children with End-Stage

Renal Disease on Dialysis

Results from the


Nicholas C Chesnaye, Franz Schaefer, Jaap W Groothoff,

Marjolein Bonthuis, Gyorgy Reusz, James G Heaf, Malcolm Lewis,

Elisabeth Maurer, Dušan Paripović, Ilona Zagozdzon, Karlijn J van

Stralen, Kitty J Jager

Kidney Int 2016 Jun; 89(6): 1355–1362

ABSTRACT

We aimed to describe survival in European paediatric dialysis patients and compare the

differential mortality risk between patients starting on haemodialysis (HD) and peritoneal

dialysis (PD). Data for 6,473 patients under 19 years of age were extracted from the

ESPN/ERA-EDTA Registry for 36 countries for the period 2000–2013. Hazard ratios were

adjusted for age at start of dialysis, gender, primary renal disease, and country. A secondary

analysis was performed on a propensity-score matched cohort (PSM). Overall 5-year survival

in European children starting on dialysis was 89.5% (95% CI 87.7%-91.0%). The mortality rate

was 28.0 deaths per 1000 patient years overall, 36.0 during the first year of dialysis, and 49.4

for patients aged 0-5 years. Cardiovascular events (18.3%) and infections (17.0%) were the

main causes of death. Children selected to start dialysis on HD had an increased mortality risk

compared to PD (aHR: 1.39, 95% CI 1.06–1.82, PSM HR: 1.46 95% CI 1.06-2.00), especially

during the first year of dialysis (HD/PD aHR: 1.70, 95% CI 1.22 – 2.38, PSM HR: 1.79, 95% CI

1.20-2.66), when starting above the age of 5 years (HD/PD aHR: 1.58, 95% CI 1.03 – 2.43, PSM

HR: 1.87, 95% CI 1.17-2.98), and when children have been seen by a nephrologist for only a

short time prior to starting dialysis (N=1681, HD/PD aHR: 6.55, 95% CI 2.35-18.28, PSM HR:

2.93, 95% CI 1.04-8.23). As unmeasured case-mix differences and selection bias may explain

the higher mortality risk in the HD population, these results should be interpreted with

caution.

Mortality risk on dialysis

93

INTRODUCTION

End-stage renal disease (ESRD) in children is a rare and severe condition, and requires renal

replacement therapy (RRT) to sustain life. Renal transplantation is the preferred treatment

modality in terms of outcomes, yet most patients will start RRT on dialysis to bridge the

preparation time needed for transplantation. Although patient survival in these children has

increased substantially over the past decades, mortality is still approximately 55 times higher

than in the general paediatric population, and occurs predominantly in the dialysis population

[79].

Several factors have been shown to affect the mortality risk in the paediatric RRT population,

the most influential being age at RRT initiation, transplantation, time on RRT, primary renal

disease (PRD), and the presence of co-morbidities [26, 34]. The few studies that explored the

effect of initial dialysis modality on mortality risk in children show conflicting results [8, 34,

186]. In Europe, no such study has previously been undertaken on an international scale, and

the rarity of paediatric ESRD has limited exploration of the heterogeneity of treatment effect

across patient subgroups and time-dependent treatment effects, as have been demonstrated in

the adult population [27–32].

The current study therefore aims to 1) describe survival in European paediatric dialysis

patients, 2) compare the mortality risk between patients starting RRT on HD and PD, and 3)

explore the differential mortality risk in the dialysis population by examining treatment-

subgroup interactions by gender, primary renal disease (PRD), age at start of RRT, co-

morbidity presence at start of RRT, and the time under treatment by a nephrologist prior to

dialysis as a marker for timely referral and the speed of disease progression.


94

METHODS

Data source and study population

This cohort study was performed using incident patient data from the ESPN/ERA-EDTA

registry database for 36 European countries [51]. Austria, Belgium, Croatia, Denmark, Finland,

Greece, Iceland, Italy, the Netherlands, Norway, Spain, Sweden, Switzerland, and The United

Kingdom provided data from January 1, 2000, to December 31, 2013. France from 2004,

Czech Republic, Hungary, Lithuania, Macedonia, Portugal, Romania, Russia, Serbia, Slovakia,

and Slovenia from 2007, Belarus, Bulgaria, Estonia, Montenegro, and Poland from 2008,

Albania, Turkey, and Ukraine from 2010, Bosnia & Herzegovina from 2011, Moldova for 2011

and 2012, and Georgia for 2013. The number of HD and PD patients per country is provided

in appendix 5.

6,473 patients were included under the age of 19 at initiation of dialysis, starting treatment

between January 1, 2000, and December 31, 2013 on either HD or PD. The initial treatment

modality was defined as treatment at day 30 as some patients start on HD to bridge the

preparation time for PD or transplantation. Data was extracted on patients’ date of birth,

gender, the time under treatment by a nephrologist prior to dialysis (defined as the time

between the first visit to the paediatric nephrologist and dialysis start), PRD, the presence of 1

or more comorbidities at start of dialysis, and events such as death, changes in treatment, and

transfer out of the Registry. 19 out of 35 countries (N=2294) provided data on the time under

treatment by a nephrologist prior to dialysis, and 26 countries (N=1725) provided data on the

presence of comorbidities at treatment start. Age groups were defined as starting RRT

between the ages 0-5 and 6-18 years. PRDs were classified following the ERA-EDTA grouping

of PRD codes for children [54]. Causes of death were defined by the ERA-EDTA coding

system, whereas ‘cardiac failure’, ‘cardiac arrest/sudden death other causes’, and ‘myocardial

ischemia and infarction’ were combined to ‘cardiovascular mortality’[59].


The primary outcome studied was all-cause death on dialysis. Baseline characteristics of PD

and HD patients were compared using Pearson’s chi-square tests. Crude and adjusted

cumulative incidence curves were used to describe the survival probabilities and mortality risk

differences between groups were estimated using Cox regression. Differential treatment-


95

effects across patient subgroups were identified by testing interaction terms in the Cox model.

All analyses were adjusted for age group, PRD group, gender, and country. We adjusted for

age groups (ages 0-1, 2-5, 6-12, and 13-18) instead of continuous age as the latter did not

show a linear relationship with mortality. Death was assigned to a patient’s initial dialysis

modality (at day 30) regardless of dialysis modality changes. Patients were censored when

renal function recovered, when lost to follow-up, reaching end of study, at transplantation, or

after 5 years of follow-up, whichever came first. As a secondary analysis, we constructed a

propensity-score matched cohort. The propensity-score is the probability of treatment

assignment, and was determined for each patient based on age at RRT start, gender, PRD, and

country. HD and PD patients with similar propensity scores were then matched on a 1-to-1

ratio. The use of propensity-scores reduces the effect of the selection bias by ensuring a

similar distribution of available baseline characteristics between treatment groups.

Furthermore, as a sensitivity analysis, we censored patients at the first treatment switch

(therefore removing time spent on switched dialysis modality) in order to assess the sole

effect of initial dialysis modality on mortality. To improve interpretability, we categorized

patient time under treatment by a nephrologist as either long (> 5 months) or short (< 5

months), determined by finding the cut-off value that gives the maximum difference in survival

[187]. As patients requiring acute dialysis are generally limited to HD as primary treatment

modality, we excluded patients requiring dialysis within 1 month after their visit to the

nephrologist to improve comparability.

RESULTS

Patient characteristics

Between January 1, 2000, and December 31, 2013, we identified 6,473 children under 19 years

of age starting RRT on dialysis in Europe. Of these patients, 30.9% started at age 0-5 years and

69.1% at 6-18 years, 56.1% were boys and 47.8% started dialysis on HD. Most ESRD was

caused by congenital anomalies of the kidney and urinary tract (CAKUT, 32.4%), followed by

glomerulopathies (18.3%), cystic kidney diseases (9.3%), and hereditary nephropathies (7.3%).

Of the group of miscellaneous PRDs (8.9%), kidney tumours (1.0%) and unspecified interstitial

nephritis (1.0%) were the most common. The number of patients recovering renal function

was low for patients starting on HD (N=51, 0.8%) and PD (N=33, 0.5%). The number lost to


96

follow-up due to transfer to an adult centre was higher in patients starting on HD (N=328,

5.1%) compared to patients starting on PD (N=137, 2.1%), reflecting the older HD dialysis

population transitioning sooner to adult centres. Patient characteristics by initial dialysis

modality are provided in table 1.

Table 1A. Patient characteristics by initial dialysis modality. PSM = propensity-score matched. *Available for

1725 patients and for 1018 patients in the PSM dataset. † Available for 1681 patients and for 1129 patients in

the PSM cohort.

Ages 0-5 Age 6-18

HD (500) PD (1498) P value HD (2591) PD (1884) P value

Gender Male 302 (60.4%) 934 (62.4%) 0.44 1411 (54.5%) 987 (52.4%) 0.17

Primary renal disease CAKUT 116 (23.2%) 582 (38.9%) <.0001 780 (34.6%) 621 (33.0%) <.0001

Glomerulonephritis 81 (16.2%) 185 (12.4%) 555 (21.4%) 365 (19.4%) Cystic 42 (8.4%) 132 (8.8%) 204 (7.9%) 223 (11.8%) Hereditary 51 (10.2%) 152 (10.2%) 170 (6.6%) 98 (5.2%) Ischemic 18 (3.6%) 47 (3.1%) 31 (1.2%) 18 (1.0%) HUS 34 (6.8%) 87 (5.8%) 84 (3.2%) 63 (3.3%) Metabolic 22 (4.4%) 31 (2.1%) 63 (2.4%) 47 (2.5%) Vasculitis 3 (0.6%) 1 (0.1%) 102 (3.9%) 45 (2.4%) Miscellaneous 75 (15.0%) 147 (9.8%) 237 (9.2%) 117 (6.2%) Unknown 58 (11.6%) 134 (9.0%) 365 (14.1%) 287 (15.2%) Comorbidity at RRT start* At least 1 31 (47.2%) 190 (40.3%) 0.86 227 (38.4%) 221 (38.0%) 0.9

Time under treatment of a nephrologist†

1 - 5 months 33 (22.3%) 115 (27.5%) 0.21 69 (13.0%) 70 (12.0%) 0.61


97

Table 1B. Patient characteristics by initial dialysis modality. PSM = propensity-score matched. *Available for

1725 patients and for 1018 patients in the PSM dataset. † Available for 1681 patients and for 1129 patients in

the PSM cohort.

PSM Ages 0-5 PSM Age 6-18

HD (458) PD (458) P value HD (1434) PD (1434) P value

Gender

Male 278 (60.7%) 277 (60.5%) 0.95 766 (53.4%) 765 (53.4%) 0.97


CAKUT 114 (24.9%) 128 (28.0%) 0.94 466 (32.5%) 468 (32.6%) 0.99

Glomerulonephritis 70 (15.3) 67 (14.6%)

284 (19.8%) 281 (19.6%)

Cystic 41 (9.0%) 42 (9.2%)

147 (10.3%) 149 (10.4%)

Hereditary 45 (9.8%) 42 (9.2%)

79 (5.5%) 79 (5.5%)

Ischemic 18 (3.9%) 13 (2.8%)

14 (1.0%) 16 (1.1%)

HUS 31 (6.8%) 37 (8.1%)

45 (3.1%) 49 (3.4%)

Metabolic 18 (3.9%) 14 (3.1%)

42 (2.9%) 37 (2.6%)

Vasculitis 2 (0.4%) 1 (0.2%)

44 (3.1%) 42 (2.9%)

Miscellaneous 64 (14.0%) 58 (12.7%)

99 (6.9%) 102 (7.1%)

Unknown 55 (12.0%) 56 (12.2%)

214 (14.9%) 211 (14.7%)

Comorbidity at RRT start*

At least 1 28 (36.8%) 30 (48.4%) 0.17 166 (37.3%) 164 (37.7%) 0.9


1 - 5 months 33 (23.6%) 37 (24.0%) 0.93 54 (12.6%) 44 (10.8%) 0.42

Overall mortality in patients on dialysis

A total of 306 deaths occurred during 10,910 patient years, equivalent to a mortality rate of

28.0 deaths per 1000 patient years (py) during the first 5 years of dialysis treatment, while

censoring for transplantation. Overall survival at 1, 2, and 5 years was 96.6% (95% CI 96.0%-

97.0%), 94.5% (95% CI 93.8%-95.2%), and 89.5% (95% CI 87.7%-91.0%), respectively. Mortality

was highest during the first year of dialysis (189 deaths, 36.0 deaths per 1000 py) and in the

youngest patients (ages 0-5, 189 deaths, 49.4 deaths per 1000 py), and declined progressively

with time on dialysis and with age at dialysis initiation.


98

Causes of death

Cardiovascular mortality (18.3%) and infection (17.0%) were the main known causes of death,

followed by cerebrovascular accidents (7.5%), withdrawal (4.9%), and malignancies (5.2%).

Cause of death was missing in 26.1% of cases (21.4% in HD, 29.7% in PD). Within the group of

cardiovascular deaths, the most common cause was cardiac arrest / sudden death (54.4%),

followed by fluid overload / pulmonary oedema (16.1%). Within the group of infection-related

deaths, sepsis was the most common cause of death (61.5%), followed by pulmonary infections

(13.5%). Of the miscellaneous causes of death, the most common cause was haemorrhage

(due to surgery or otherwise, 9.8%) followed by pulmonary embolus (8.2%). The cause of

death varied by current treatment modality (appendix 1), with cardiovascular mortality (20.5%)

as the predominant cause of death in HD patients, whereas infection was the main cause of

death in PD (19.5%) patients. The cause of death did not differ between the first year on

dialysis and the years thereafter. In a sensitivity analysis excluding 4 countries with a high

percentage of missing deaths, the ranking of causes of death remained unchanged.

Mortality risk comparison between HD and PD

In patients initiating dialysis under 5 years of age, the 5-year crude mortality rate was 57.0

deaths per 1000 py for patients selected to start RRT on HD vs. 47.3 deaths per 1000 py for

those initiating on PD . In patients initiating dialysis above 5 years of age, the mortality rate

was 20.6 deaths per 1000 py for those starting on HD, and 11.1 deaths per 1000 py for those

starting on PD. After adjustment for PRD, age at start of dialysis, gender, and country, patients

selected to start RRT on HD had a 39% increased risk of death compared to patients starting

on PD (HD vs. PD, aHR: 1.44, 95% CI 1.06 – 1.82). Analysis of the propensity-score matched

cohort (PSM, N=3784), matched on age at start RRT, PRD, gender, and country, showed

similar results (HD vs. PD, aHR: 1.46, 95% CI 1.06 – 2.00). In a sensitivity analysis, censoring

patients at first dialysis modality switch, the HR profiles remained similar (HD vs. PD, aHR:

1.44, 95% CI 1.08 – 1.93), reflecting that death occurs predominantly on the initially selected

dialysis modality. After additional adjustment for the time under treatment by a nephrologist

prior to dialysis (N=2294), patients selected to start RRT on HD had a 63% increased risk of

death compared to patients starting on PD (HD vs. PD, aHR: 1.63, 95% CI 1.06 – 2.52).


99

Sub-group analyses

We identified a significant treatment-subgroup interaction effect by patient age at start of

dialysis (p <.0001) and by the time under treatment by a nephrologist prior to dialysis initiation

(p=0.01). After stratification by age group, the data suggested a somewhat weaker association

in younger patients (HD vs. PD, aHR: 1.24, 95% CI 0.85 – 1.79) compared to older patients

(HD vs. PD, aHR: 1.58, 95% CI 1.03 – 2.43). The propensity-score matched cohort provided

similar results (HD vs. PD, HR for ages 0-5: 1.18, 95% CI 0.77-1.82, HR for ages 6-18: 1.87,

95% CI 1.17-2.98). The crude, adjusted, and PSM cumulative incidence plots stratified by age

group are presented in figure 1.

We studied the effect of the duration of time under treatment by a nephrologist prior to

dialysis initiation as a marker for the timeliness of referral and the speed of disease

progression. Overall, the median time under treatment by a nephrologist was 0.88 (IQR 0.07 -

3.90) years for PD patients and 1.61 (IQR 0.07 – 6.11) years for HD patients. This difference

was due to a relatively high proportion of young (aged 0-2 years) PD patients (appendix 2).

Within the group of patients with a short time under treatment by a nephrologist (1-5

months), the median time was similar for patients starting on HD (N=102, 75 days) compared

to PD (N=185, 71 days), but was lengthier for HD patients in the group with a long time

under treatment by a nephrologist (> 5 months, HD patients; N=577, 4.4 years, PD patients;

N=817, 3.1 years). Patient characteristics are presented in appendix 3, and figure 2 shows the

crude, adjusted, and PSM cumulative incidence plots stratified by these categories. There was

no significant difference in survival for patients under treatment by a nephrologist for >5

months (HD vs. PD, aHR: 0.79, 95% CI 0.38-1.63, PSM HR: 0.96, 95% CI 0.44-2.11). However,

among patients under treatment by a nephrologist between 1-5 months, those selected to

start on HD had a significantly poorer survival compared to those starting on PD (HD vs. PD,

aHR: 6.55, 95% CI 2.35-18.28, PSM HR: 2.93, 95% CI 1.04-8.23). Further stratification by age

group provided similar results (appendix 4).The HR profiles remained similar in a sensitivity

analysis excluding patients under 2 years of age, and in a separate analysis including patients

with <1 month of time under treatment of a nephrologist.


100

Table 2A. Mortality rates and hazard ratios for HD vs. PD calculated separately for each patient sub-group,

and adjusted for age group at RRT, gender, PRD group, and country. There were no deaths in the PD group

for vasculitis patients. PSM = propensity-score matched. *Available for 1725 patients and for 1018 patients in

the PSM dataset. † Available for 1681 patients and for 1129 patients in the PSM cohort.

HD PD PSM HD PSM PD

Deaths/Years Deaths/Years Deaths/Years Deaths/Years

Overall 131 / 4878 175 / 6031 93 / 3024 66 / 3118

Age at RRT start 0-5 48 / 842 141 / 2981 43 / 776 39 / 860

6-18 83 / 4036 34 / 3051 50 / 2248 27 / 2258

Gender Female 64 / 2162 78 / 2590 46 / 1344 31 / 1418

Male 67 / 2716 97 / 3441 47 / 1679 35 / 1700

Primary renal disease Glomerulonephritis 17 / 1028 27 / 932 11 / 570 13 / 545

Cystic 5 / 368 17 / 562 5 / 278 3 / 287

Hereditary 7 / 299 14 / 393 6 / 175 5 / 204

Ischemic feb-77 5 / 133 feb-59 jan-52

HUS 5 / 194 6 / 338 5 / 122 4 / 178

Metabolic 4 / 142 4 / 135 2 / 104 mrt-81

Vasculitis 9 / 145 0 / 60 apr-65 0 / 55

Miscellaneous 34 / 494 16 / 497 22 / 276 14 / 264

Unknown 21 / 637 27 / 713 13 / 412 14 / 411

CAKUT 27 / 1493 59 / 2269 23 / 962 9 / 1040

Comorbidity at RRT start* At least 1 13 / 476 34 / 861 11 / 368 12 / 389

None 9 / 832 24 / 1361 8 / 634 3 / 612


1 - 5 months 14 / 155 13 / 357 13 / 139 5 / 160

> 5 months 13 / 771 28 / 1255 12 / 644 13 / 669


101

Table 2B. Mortality rates and hazard ratios for HD vs. PD calculated separately for each patient sub-group,

and adjusted for age group at RRT, gender, PRD group, and country. There were no deaths in the PD group

for vasculitis patients.PSM = propensity-score matched. *Available for 1725 patients and for 1018 patients in

the PSM dataset. † Available for 1681 patients and for 1129 patients in the PSM cohort.

Crude Adjusted PSM

HR (95% CI) HR (95% CI) HR (95% CI)

Overall 0.91 (0.73-1.14) 1.39 (1.06-1.82) 1.46 (1.06-2.00)

Age at RRT start

0-5 1.16 (0.84-1.62) 1.24 (0.85-1.79) 1.18 (0.77-1.82)

6-18 1.84 (1.23-2.74) 1.58 (1.03-2.43) 1.87 (1.17-2.98)

Gender

Female 0.97 (0.69-1.34) 1.36 (0.93-1.99) 1.56 (0.99-2.46)

Male 0.86 (0.63-1.18) 1.39 (0.95-2.04) 1.37 (0.88-2.12)


Glomerulonephritis 0.57 (0.31-1.04) 0.78 (0.39-1.56) 0.82 (0.37-1.83)

Cystic 0.45 (0.17-1.22) 1.53 (0.41-5.70) 1.66 (0.40-6.97)

Hereditary 0.61 (0.24-1.51) 5.3 (1.59-17.67) 1.22 (0.37-4.06)

Ischemic 0.65 (0.13-3.39) 1.17 (0.14-9.90) 1.91 (0.17-21.05)

HUS 1.33 (0.40-4.37) 2.83 (0.71-11.29) 1.67 (0.45-6.25)

Metabolic 0.41 (0.05-3.11) 1.10 (0.28-4.24) 0.60 (0.10-3.57)

Vasculitis - - -

Miscellaneous 2.08 (1.15-3.78) 1.85 (0.94-3.63) 1.59 (0.81-3.11)

Unknown 0.87 (0.49-1.53) 1.08 (0.54-2.16) 0.93 (0.44-1.98)

CAKUT 0.68 (0.43-1.08) 1.74 (1.00-3.03) 2.78 (1.29-6.02)

Comorbidity at RRT start*

At least 1 0.69 (0.36-1.30) 0.85 (0.40-1.81) 2.63 (0.70-9.91)

None 0.63 (0.29-1.36) 1.22 (0.51-2.94) 1.34 (0.48-3.76)


1 - 5 months 2.47 (1.16-5.27) 6.55 (2.35-18.28) 2.93 (1.04-8.23)

> 5 months 0.75 (0.39-1.46) 0.79 (0.38-1.63) 0.96 (0.44-2.11)


102

Time-dependent treatment effect

The mortality risk difference between HD and PD was not constant over time. During the

initial period of treatment there was a survival advantage for PD, after which the mortality risk

became similar to HD. Specifically, during the first year on dialysis, patients starting RRT on

HD had a 70% increased risk of death compared to patients starting on PD (HD vs. PD, aHR:

1.70, 95% CI 1.22 – 2.38, PSM HR: 1.79, 95% CI 1.20-2.66).

In the sub-group analyses, the treatment effect during the first years on dialysis increased in

the younger patients, whilst remaining statistically insignificant (HD vs. PD, aHR: 1.47, 95% CI

0.95 – 2.27, PSM HR: 1.42, 95% CI 0.85-2.35), and was even stronger in older patients (HD vs.

PD, aHR: 2.17, 95% CI 1.15 – 4.09, PSM HR: 2.60, 95% CI 1.33-5.08). Furthermore, the time-

dependent effect during the first year on dialysis remained absent in patients with > 5 months

of time under treatment by a nephrologist prior to dialysis (HD vs. PD, aHR: 0.94, 95% CI 0.38

– 2.36, PSM HR: 1.06, 95% CI 0.37-3.01), and remained present among patients with 1-5

months under treatment by a nephrologist prior to dialysis (HD vs. PD, aHR: 16.08, 95% CI

3.30– 78.43, PSM HR: 4.12, 95% CI 0.87-19.4).

Dialysis modality and transplantation rate

We studied the transplantation rate across dialysis modalities as a higher rate of

transplantation in either dialysis modality could possibly lead to informative censoring, as

healthier patients would be transplanted and censored, leaving the less healthy patients as the

remainder of the cohort. Patients starting on HD had a somewhat shorter median time to Tx

(1.76 years, 95% CI 1.64-1.87) compared to PD patients (1.86 years, 95% CI 1.78-1.95), and an

overall similar 5-year likelihood of transplantation (HD vs. PD, HR: 1.02 95% CI 0.96 – 1.09,

PSM HR: 0.95, 95% CI 0.87-1.03), with increased likelihood during the first year on dialysis

(HD vs. PD, HR: 1.33, 95% CI 1.21 – 1.46). This difference decreased after adjustment for age

at RRT start (HD vs. PD during first year on dialysis, aHR: 1.12 95% CI 1.01 – 1.24, PSM HR:

1.07, 95% CI 0.95-1.21), reflecting the lower transplantation probability in the younger PD

population.


103

DISCUSSION

In this study, we describe all-cause mortality in the largest cohort of European paediatric ESRD

patients studied to date, including 6,473 patients who initiated dialysis between 2000 and 2013.

Overall survival in these children is good, with almost 90% surviving after 5 years of dialysis,

however, mortality risk varies substantially within this population. Comparing HD versus PD,

we demonstrated an increased mortality risk in children selected to start dialysis on HD,

especially during the first year of dialysis, in children starting dialysis at an older age, and in

children with a short time under treatment by a nephrologist prior to dialysis. These

differential mortality risks highlight the importance of focussed clinical management in these

sub-groups.

The increased mortality risk observed in European paediatric patients selected to initiate

treatment on HD compared to PD is partially in agreement with previous observational

studies in the paediatric dialysis population. In the US, Wong et. al. reported an increased, but

non-significant, adjusted mortality risk for 1,723 paediatric patients (<18 years of age) starting

RRT on HD compared to PD between 1995-98 [33]. In a larger and more recent US study,

Mitsnefes et. al. observed a significantly higher adjusted risk of death in children younger than

5 years initiating HD compared to PD between 1990-2010, but in contrast to our results, this

effect was absent in children older than 5 years [34]. In Taiwan, Lin et. al. found no difference

in crude survival between paediatric (<19 years of age) patients starting on either HD or PD

between 1995 and 2004, similar to our crude analysis [35]. In a per protocol analysis,

McDonald et al. found a higher but non-significant adjusted mortality risk for 1,634 patients

(<20 years of age) starting on PD compared to HD between 1963-2002 in the ANZDATA

registry, although survival from that era may not be extrapolated to the current one [8].

We demonstrate that the mortality risk difference between HD and PD varies over time, with

an initial survival advantage for patients selected to start on PD during the first one to two

years on dialysis, after which the mortality risk becomes similar to those selected to start on

HD. This time-dependent effect is consistent with several studies performed in the adult

dialysis population, and has been attributed to an improved preservation of residual renal

function in PD patients during the first period of dialysis, followed by an increasing


104

deterioration of the peritoneal membrane over time, reducing dialysis adequacy and potentially

increasing the mortality risk on PD over a prolonged period of dialysis [188–193].

We explored the effect of the time under treatment by a nephrologist prior to dialysis as an

indicator of the timeliness of referral and the speed of renal disease progression. It has

previously been shown that paediatric patients that are referred late have poorer clinical and

biochemical status and a reduced rate of pre-emptive transplantation compared to those

referred earlier [194–196], whereas the speed of renal disease progression is largely

dependent on the underlying renal disease and CKD management [65, 197]. Importantly, we

demonstrate that in patients with sufficient time under treatment by a nephrologist prior to

dialysis (due to either timely referral and/or a relatively slow progression of renal disease),

modality choice does not influence the survival outcome. Conversely, in patients with a

relatively short time under treatment by a nephrologist prior to dialysis (but excluding

unplanned crash starts who may be limited to HD due to the acuteness of dialysis initiation)

we found a poorer prognosis for patients selected to start RRT on HD, even after adjustment

for important confounders such as renal disease and age at dialysis initiation. Similar results

have been demonstrated in adults, where patients starting RRT with ‘planned’ PD have a

similar mortality risk compared to ‘planned’ HD patients with an AV-fistula and ‘unplanned’ PD

patients, but where patients starting ‘unplanned’ HD on a central venous catheter had a higher

mortality risk [198, 199]. Indeed, vascular access type has been demonstrated as an important

modifier of the relationship between dialysis modality and survival, and given that the survival

advantage in PD patients was only found when comparing patients who initiated HD with a

central venous catheter, it has been suggested that case-mix differences that coincide with HD

vascular access type, such as timely access to pre-dialysis care, are likely to explain the higher

mortality risk in the HD population [190]. Disentanglement of the relationships between

mortality on dialysis, the timeliness of referral, disease progression speed, and vascular access

type certainly deserves further investigation in the paediatric population.

An important caveat when comparing dialysis modalities using observational data is the issue of

residual confounding due to unmeasured variables. Dialysis modality choice in children

depends on many factors, such as patient age, family and patient preference, the timeliness of

referral and acuteness of disease, social conditions, logistical issues, facility preference and

technical capability, and various medical aspects. Due to the inherent nature of observational


105

data, we are unable to assess unmeasured variables that may affect both dialysis choice and

mortality risk. Moreover, as sicker and more complex patients may be selected to start on HD

as initial modality choice, this could have introduced some degree of selection bias to our

results [140, 142]. In an attempt to account for this, we performed a propensity-score

matched analysis, thus eliminating baseline differences in the available determinants of dialysis

choice between groups. Furthermore, although the presence of one or more comorbidities at

dialysis initiation was associated with mortality, it did not affect the HR between HD and PD.

Indeed, according to a recent European study auditing 14 dialysis units in 11 countries, dialysis

modality choice in children is less influenced by the presence of comorbidities, as is more the

case in adult patients [200]. Nonetheless, despite our propensity-score matched analyses, the

mortality risk difference between HD and PD could still be explained by unmeasured case-mix

differences, such as details on the type and severity of comorbidity, malnutrition, hypertension,

access type, and various metabolic factors at dialysis initiation, which may have introduced

some degree of selection bias, and therefore the results should be interpreted conservatively.

Notwithstanding the potential bias from unmeasured variables, our findings may have

important implications for the clinical management of children with ESRD. Our data support

the preferential use of PD as initial treatment modality particularly in children who are

referred to specialist care late, leaving little time to adapt to end-stage renal disease prior to

dialysis initiation. This appears to be applicable independent of the presence of comorbidities.

The study strengths include its large sample size which provides the necessary power to

perform stratified analyses, sufficient follow-up time with a low loss to follow-up, and the

European-wide inclusion of patients. Limitations include the high percentage missing causes of

death, which may limit the precision of our results and our ability to detect differences

between sub-groups, although the ranking of cause of death remained the same when

excluding countries with a large number of missing values. In addition, although we do not

adjust for potential regional or centre-level confounders, we adjust for country, which we

consider adequate to generalize our results to the European patient population.

We conclude that although overall survival in European children with ESRD starting RRT on

dialysis is relatively good, mortality risk is amplified during the first year on dialysis and in the

youngest patients. Importantly, we demonstrate an increased mortality risk in children


106

selected to start dialysis on HD compared to PD, especially during the first year of dialysis, in

older children, and in patients with a short duration of time under treatment by a nephrologist

prior to dialysis. The identification of vulnerable sub-groups in the paediatric dialysis

population suggests scope for further improvement through targeted preventive approaches.

However, as sicker and more complex patients may be selected to start on HD as initial

modality choice, unmeasured case-mix differences may explain the higher mortality risk in the

HD population, and therefore these results should be interpreted with caution.


107

Figure 1. Crude (A), adjusted (B), and propensity-score matched (C) cumulative incidence plots by dialysis

modality and age group at RRT start, using covariate values, PRD group = CAKUT, gender = male, age group at

RRT initiation = 2-5 years, country = UK. The plots were truncated at 3 years due to an insufficient number of

patients remaining in the cohort.


108

Figure 2. Crude (A),adjusted (B), and propensity-score matched (C) cumulative incidence plots by dialysis

modality and time under treatment by a nephrologist, using covariate values, PRD group = CAKUT, gender =

male, age group at RRT initiation = 2-5 years, country = UK. The plots were truncated at 3 years due to an

insufficient number of patients remaining in the cohort.


109

Appendix 1. Cause of death by dialysis modality. The percentage of malignancies as the cause of death in

patients on peritoneal dialysis is 1.7%.

Appendix 2. Distribution of time under treatment by a nephrologist by dialysis modality. The lighter segments

represent the proportion of patients starting renal replacement therapy RRT between 0 and 2 years of age.

13.6% 19.7%

20.5%17.9%

7.6%7.5%

9.8%4.5%

5.2%

20.5%20.8%

23.5% 27.2%

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

HD PD

Unknown

Miscellaneous

Withdrawal

Malignancy

Cerebrovascular

Cardiovascular

Infections

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0 10+

Pre-dialysis period

PD HD


110

Appendix 3. Patient characteristics by categorized time under treatment by a nephrologist prior to dialysis.

HD (577) PD (817) P value

HD (102) PD (185) P value

> 5 months > 5 months 1-5 months 1-5 months

Age <.0001

0-5 19.9 37.1 <.0001 32.4 62.2 6-18 80.1 62.9 67.7 37.8 Gender 0.18

Male 58.4 60.7 0.39 52.4 61.1 Primary renal disease 0.001

CAKUT 38.1 43.2 0.04 21.6 41.1 Glomerulonephritis 18.4 18.6 17.7 12.4 Cystic 9.9 8.5 8.8 10.3 Hereditary 8.2 9.6 4.9 10.8 Ischemic 1 1.6 3.9 2.2 HUS 2.3 3.3 5.9 8.1 Metabolic 4 2.5 1 0 Vasculitis 3.3 1.4 2.9 2.7 Miscellaneous 6.9 4.9 19.6 7.6 Unknown 8 6.6 13.7 4.9 Comorbidity at RRT start ( 77% missing) 0.17

At least 1 61.2 58 0.58 65.2 47.7


111

Appendix 4. Crude (A) and adjusted (B) cumulative incidence plots by dialysis modality and time under

treatment by a nephrologist, and age group, using covariate values, PRD group = CAKUT, sex = male, age

group at RRT initiation = 2-5 years / 6-13 years, country = UK. The plots were truncated at 3 years due to an

insufficient number of patients remaining in the cohort.


112

Appendix 5. Number of HD and PD patients initiating RRT between 2000 and 2013 in each country.

HD PD

Albania 6 0

Austria 97 52

Belarus 19 35

Belgium 42 55 Bosnia and Herzegovina

15 3

Bulgaria 27 6

Croatia 19 56

Czech Republic 31 34

Denmark 71 53

Estonia 1 1

Finland 52 91

France 614 228

Georgia 2 0

FYR of Macedonia 0 8

Greece 118 85

Hungary 20 39

Iceland 6 7

Italy 212 383

Lithuania 15 9

Moldova 4 0

Montenegro 0 3

Norway 48 24

Poland 79 129

Portugal 23 94

Republic of Serbia 31 12

Romania 149 40

Russia 273 260

Slovakia 19 27

Slovenia 14 3

Spain 256 295

Sweden 59 126

Switzerland 60 46

the Netherlands 161 167

Turkey 100 212

Ukraine 19 11

United Kingdom 429 788

Total 3091 3382

Demographics

7

Infants Requiring Dialysis:

Outcomes of Hemodialysis

and Peritoneal Dialysis

Enrico Vidal, Karlijn J van Stralen, Nicholas C Chesnaye,

Marjolein Bonthuis, Christer Holmberg, Aleksandra Zurowska,

Antonella Trivelli, José Eduardo Esteves Da Silva, Maria

Herthelius, Brigitte Adams, Anna Bjerre, Augustina Jankauskiene,

Polina Miteva, Khadizha Emirova, Aysun K Bayazit, Christoph J

Mache, Ana Sánchez-Moreno, Jérôme Harambat, Jaap W

Groothoff, Kitty J Jager, Franz Schaefer, Enrico Verrina

Am J Kidney Dis. 2017 May; 69(5): 617-625

ABSTRACT

Background: The impact of different dialysis modalities on clinical outcomes has not been

explored before in young infants with end-stage renal disease.

Methods: Study design: Cohort study. Setting & Participants: Data were extracted from the

ESPN/ERA-EDTA Registry. This analysis included 1063 children aged <12 months who initiated

renal replacement therapy (RRT) from 1991 to 2013. Factor: Type of dialysis modality.

Outcomes & Measurements: Differences between infants treated with peritoneal dialysis (PD)

or hemodialysis (HD) in patient survival, technique survival, and access to kidney

transplantation were examined using Cox regression analysis while adjusting for age at dialysis

initiation, gender, underlying renal disease, and country of residence.

Results: 917 infants initiated dialysis on PD and 146 on HD. Median age at dialysis start was

4.5 (IQR 0.7-7.9) months and median body weight 5.7 (IQR 3.7-7.5) kg. While the groups

were homogeneous regarding age and gender, children treated with PD more often had

CAKUT (48 vs. 27%), whereas those on HD suffered more frequently from metabolic

disorders (12 vs. 4%). Risk factors for death were younger age at RRT initiation (HR: 0.94, 95%

CI 0.90-0.97) and non-CAKUT etiology of ESRD (HR: 1.49, 95% CI 1.08-2.04). Mortality risk

and likelihood of transplantation were equal in PD and HD patients, whereas HD patients had

a higher risk of changing dialysis treatment (aHR: 1.64, 95% CI 1.17-2.31). Limitations: Inability

to control for unmeasured confounders not included in the Registry database and missing data

(i.e. comorbidities).

Conclusions: Despite a widespread preconception that HD should be reserved for cases

where PD is not feasible, in Europe we found one in eight infants in need for chronic dialysis

to be started on HD. Patient characteristics at dialysis initiation, prospective survival, and time

to transplantation were very similar for infants commenced on HD or PD.

Infant dialysis

115

INTRODUCTION

The management of infants requiring chronic dialysis represents a significant challenge for

pediatric nephrologists. Difficulties in feeding and maintaining fluid balance, growth failure,

increased infection risks, and the presence of co-morbidities complicate the management of

chronic renal failure in children <1 year of age [117]. Consequently, mortality rates in infants

on dialysis are substantially higher than in older children [115].

In a multinational survey performed in the late 1990’s, only 50% of pediatric nephrologists

recommended initiation of RRT in infants with end-stage renal disease (ESRD) [201]. Since

then, this attitude has been partially modified by reports demonstrating favorable results in

growth, development, and renal transplantation in infants on dialysis given careful medical and

nutritional management [143, 202–205]. The number of infants on RRT has increased over the

past decades and according to the 2011 North American Pediatric Renal Trials and

Collaborative Studies (NAPRTCS) Report, 13.2% of patients were less than 2 years old at

dialysis initiation [68, 114].

Maintenance PD represents the preferred dialysis modality in infants [86, 120, 143, 202, 203].

Advantages over HD include potentially better preservation of residual kidney function [206],

less dietary restrictions, avoidance of central vascular access placement, and the option to

perform dialysis at home, though this requires a labor-intensive effort from the family [207].

The experience of treating infants with HD is limited [121, 208–211]. HD in infants is

technically difficult and requires highly qualified nursing staff. However, in cases where PD is

contraindicated for clinical reasons, fails, or is inappropriate due to psychosocial problems, HD

remains the only alternative treatment until renal transplantation is feasible [145].

To our knowledge, no reports have compared the long-term outcomes of both dialysis

modalities in infants. We therefore sought to compare the clinical characteristics and

outcomes of PD and HD patients in a large cohort of children starting dialysis under 1 year of

age.

Infant dialysis

116

METHODS

Study population

We analyzed data of 1081 infants who initiated RRT before 12 months of age between January

1, 1991 and December 31, 2013. The cohort included all patients collected within the

framework of the European Society for Paediatric Nephrology/European Renal Association

and European Dialysis and Transplant Association (ESPN/ERA-EDTA) Registry. Countries

initiating infants on dialysis during the study period were: Austria, Belarus, Belgium, Bosnia-

Herzegovina, Bulgaria, Croatia, Czech Republic, Denmark, Finland, France, Germany, Greece,

Hungary, Italy, Lithuania, the Netherlands, Norway, Poland, Portugal, Romania, Russia, Serbia,

Slovakia, Slovenia, Spain, Sweden, Switzerland, Turkey, and the United Kingdom. Patient

numbers per country are included in appendix 1.

We excluded patients who received a pre-emptive renal transplantation (n=10) and patients

whose dialysis modality was not clearly specified (n=8). Patients entered the study on day 1 of

dialysis and were then stratified by modality on day 30. For patients who died within the first

month of treatment, the last treatment modality prior to death was considered for analysis.

Data collection

Age, gender, primary renal disease, initial treatment modality and any subsequent changes are

obligatory information in the ESPN/ERA-EDTA Registry. Other parameters such as body

weight, height, blood pressure, serum creatinine, albumin, hemoglobin, and parathyroid

hormone (PTH) levels at baseline and during follow-up are provided on a voluntary basis, as

well as the reasons for modality failure. Primary renal disease and causes of death were

determined by the patients’ nephrologists and classified according to the ERA-EDTA coding

system [54]. No ethics committee approval or informed consent was required as the

ESPN/ERA-EDTA Registry is based on observational and anonymized patient data collection.


The primary outcome studied was patient survival by dialysis modality. Secondary outcomes

included comparison of clinical characteristics at dialysis onset, technique survival, and the

likelihood of transplantation in infants receiving PD and HD. The primary analysis was

performed on an “intention-to-treat” (ITT) basis where patients were assigned to the initial

Infant dialysis

117

dialysis modality (at day 30). As infants often tend to switch between modalities, we also

performed a “per-protocol” (PP) analysis, assigning patients to the treatment they actually

received. For both the ITT and PP analyses, patients were censored at transplantation, when

renal function recovered, when lost to follow-up, reaching end of study period (December 31,

2013), or after 5 years of follow-up, whichever came first. Cumulative incidence competing

risk curves were constructed for death (with transplantation as a competing risk),

transplantation (with death as a competing risk), and modality switching (with both death and

transplantation as competing risks) [57]. Cox regression was used to adjust for possible

confounders, including age at start of dialysis, gender, and underlying renal disease. Due to the

low number of patients in some smaller countries, and that some countries have either no HD

or no PD patients, it was not possible to adjust for country as a fixed effect without making

the model unstable. As an alternative to adjust for a potential country effect on clinical

outcomes, a random country factor was added to the Cox model using the shared frailty

model. This random effect allows patients within the same country to share a baseline hazard

while allowing the hazard function to differ between countries, and therefore allows the model

to account for the effects of unobserved heterogeneity between countries.

Demographic baseline and clinical characteristics were described with medians and

interquartile ranges (IQR) or proportions, as appropriate. The Student’s t was used to test for

differences between treatment groups for normally distributed continuous variables, the

Wilcoxon test for non-normally distributed continuous variables, and the χ2 test for

categorical variables. Estimated glomerular filtration rate (eGFR) was calculated by using the

updated Schwartz formula [55]. Linear mixed models were used to compare mean levels of

serum albumin, hemoglobin (Hb), blood pressure standard deviation scores (SDS), and PTH

between the two treatment groups, whilst adjusting for multiple measurements within a

patient, as well as for confounders. Height values were normalized to SDS for chronological

age using recent national or European height-for-age charts [212]. As serum Hb changes

during the first year of life, age specific SDS for Hb were calculated using KDIGO reference

values. For the analyses of clinical and biochemical parameters, all measurements during the

first year of dialysis were used except for the baseline measurements. Statistical tests were

two-tailed and were considered significant when p<0.05. Data were analyzed using the SAS

software (version 9.4, SAS Institute, Cary, NC, USA).

Infant dialysis

118

RESULTS

Patient characteristics

We identified a total of 1063 infants starting on dialysis. Of these, 919 started on PD and 144

on HD. At day 30, 14 PD patients had switched to HD and 12 HD patients had switched to

PD. Fourteen patients died before day 30 (12 on PD and 2 on HD). Dialysis was initiated in

649 (61%) children at age 0-6 months and in 414 (39%) children at age 7-12 months. The

baseline patient characteristics by initial dialysis modality are shown in table 1, whereas the

estimated mean for the clinical and biochemical parameters during the first year of dialysis are

reported in table 2. We found a higher proportion of hypoalbuminemic infants on PD, likely

resulting from increased protein losses via the peritoneal membrane that at this age is often

characterized by a hyperpermeable state. Conversely, infants on HD presented with

significantly lower hemoglobin levels, possibly related to significant blood losses with the

extracorporeal systems or more relevant fluid overload at the time of blood sampling, which is

usually performed immediately before dialysis.

In infants receiving PD, automated cycler regimens were applied in 71% of cases (out of the

605 patients for whom this information was available), whereas 29% of infants initially received

manual intermittent or continuous ambulatory PD. Nearly all HD patients received in-center

HD, except for one case treated with home-HD. For the 131 patients for whom this was

known, 90% were treated with bicarbonate HD and 10% with hemodiafiltration. For 21

patients, we had information on the number of HD treatment sessions per week and the

duration of each session. Ten out of 21 patients had 3 days of HD per week, while the

remaining patients had 2 (1 case), 4 (2 cases), 5 (4 cases), 6 (2 cases), or 7 days (2 cases) per

week. Total hours of HD per week were highly variable, ranging from 6 to 35, with a median

of 12 hours. Information on the type of vascular access was available for 15 patients; a central

line was used in 14 cases (median age at implantation of 8.4 months) and an arteriovenous

graft was used in 1 case (placed at 7.5 months of age).

Infant dialysis

119

Table 1. Baseline patient characteristics by initial dialysis modality. P-values refer to comparison between PD

and HD. a Adjusted for age at start.

Available data All patients PD HD

(n=1063) (n=917) (n=146)

N (%) Median (IQR) Median (IQR) Median (IQR) p

Age (months) 1063 (100%) 4.5 (0.7-7.9) 4.3 (0.7-7.9) 5.1 (1.3-7.9) 0.4

Female gender (%) 1063 (100%) 33.2 32.4 38.4 0.2

Body weight (kg) 576 (54%) 5.7 (3.7-7.5) 5.5 (3.6-7.5) 6.3 (4.2-8.0) 0.06 a

Height (cm) 473 (44%) 60 (52-67) 60 (52-67) 62 (55-67) 0.2 a

Height SDS 473 (44%) -1.1 (-2.4- -0.3) -1.3 (-2.4-0.2) -0.9 (-2.6-0.5) 0.2

BMI (kg/m2) 491 (44%) 16.6 (15.3-18.8) 16.6 (15.3-18.9) 16.5 (15.4-18.7) 0.9

eGFR (ml/min/1.73 m2) 313 (29%) 6.1 (4.4-8.4) 6.1 (4.4-8.0) 6.3 (4.2-8.8) 0.7

Primary diagnostic group (%) 1063 (100%) <0.001

CAKUT 45.3 48.4 27.1 Glomerulonephritis 4.7 4.7 4.7 Cystic kidney disease 8.3 8.1 9.3 Hereditary nephropathy 15.4 15.9 12.4 Ischemic renal failure 4.7 4.2 7.8 HUS 3.1 3.3 2.3 Metabolic disorders 5.5 4.1 12.4 Vasculitis 0.2 0 1.6 Miscellaneous 9.4 8 17.8 Unknown 3.5 3.3 4.7

Table 2A. Unadjusted mean clinical and biochemical parameters during the first year of dialysis treatment. P-

values refer to comparison between PD and HD. N = patients, n = measurements. BMI = Body Mass Index;

PTH = serum parathyroid hormone.

n N Mean (95% CI)

BMI (kg/m2) 1920 705 16.1 (15.9-16.3)

Systolic BP SDS (mmHg) 1095 496 1.2 (1.0-1.4)

Diastolic BP SDS (mmHg) 983 434 1.7 (1.5-1.8)

Hemoglobin SDS 1068 498 -1.62 (-1.84 to -1.40)

Serum albumin (g/dL) 977 491 32.5 (31.8-33.2)

PTH (pg/mL) 892 422 496 (438-555)

Infant dialysis

120

Table 2B. Unadjusted mean clinical and biochemical parameters during the first year of dialysis treatment. P-

values refer to comparison between PD and HD. N = patients, n = measurements. BMI = Body Mass Index;

PTH = serum parathyroid hormone.

PD HD

p

n N Mean (95% CI) n N Mean (95% CI)

BMI (kg/m2) 1666 615 16.1 (15.9-16.3) 254 90 16.3 (15.7-16.9) 0.6

Systolic BP SDS (mmHg) 974 438 1.1 (1.0-1.3) 121 58 1.5 (1.0-2.0) 0.2

Diastolic BP SDS (mmHg) 877 388 1.6 (1.5-1.8) 106 46 2.1 (1.7-2.5) 0.03

Hemoglobin SDS 900 423 -1.40 (-1.64 to -1.15) 168 75 -2.73 (-3.28 to -2.17) <0.001

Serum albumin (g/dL) 878 434 32.1 (31.3-32.8) 99 57 36.4 (34.2-38.6) <0.001

PTH (pg/mL) 765 360 500 (433-568) 127 62 474 (321-628) 0.7

Patient survival and cause of death

The overall 5-year crude mortality rate in the entire cohort of infants receiving dialysis was

52.3 deaths per 1000 patients year (py). The overall cumulative incidence of death at 1, 2 and

5 years was 10.0% (95% CI 8.10%-11.7%), 13.1% (95% CI 11.0%-15.2%) and 16.1% (95% CI

13.8%-18.5%), respectively. Causes of death were infections (25.1%), cardiovascular disease

(13.6%), withdrawing ESRD treatment (6.8%), respiratory failure due to fluid overload (3.1%),

cerebrovascular accident (5.8%), malignancy (2.1%), miscellaneous (23.6%) and unknown/not

available causes (19.9%). Among the 26 deaths for cardiovascular disease, the specific reported

causes were sudden cardiac arrest (50%), myocardial infarction (4%), hypertensive cardiac

failure (4%), and unknown causes of cardiac failure (42%). There were no significant differences

in cause of death between children starting dialysis before and after the year 2000. Causes of

death according to dialysis modality were also comparable.

Younger age at the start of RRT was a significant risk factor for death, with a 5% lower risk

per month of later start (HR: 0.95, 95% CI 0.90-0.97; p<0.001). A significantly higher risk of

death was found in patients with non-CAKUT diseases (HR: 1.49; 95% CI 1.08-2.04; p=0.03),

while there was no significant mortality risk difference by gender (female vs. male, HR: 1.28;

95% CI 0.95-1.71) or between children starting dialysis before and after the year 2000 (post-

2000 vs. pre-2000, HR: 0.93; 95% CI 0.67-1.29). Survival was also similar across countries

(country frailties are presented in appendix 1).

Infant dialysis

121

Mortality risk comparison between HD and PD

The crude 5-year mortality rate was 51.0 deaths per 1000 py for PD and 62.2 deaths per 1000

py for HD. The 5-year cumulative incidence of death is presented by dialysis modality in table

3 and figure 1A. In the ITT analysis, whilst censoring for transplantation, crude (HR: 1.08, 95%

CI 0.69-1.68) and adjusted (aHR: 1.06, 95% CI 0.67-1.67) hazard ratios did not differ

significantly between treatment groups. HD vs. PD hazard ratios did not differ significantly

between children starting dialysis before and after the year 2000 (p-value for interaction term

= 0.64). Among infants whose initial dialysis modality was PD, 135 of 143 deaths occurred

while still on PD and 8 died while switched to HD. Among HD patients, 19 of 23 deaths

occurred while still on HD and 4 were on PD. In the PP analysis, crude (HR: 0.76, 95% CI

0.47-1.22) and adjusted (aHR: 0.73, 95% CI 0.45-1.18) hazard ratios did not differ significantly

between treatment groups.

Experience and skills in treating infants on HD may vary across European countries. To

explore the potential impact on survival of a country’s experience in treating infants on HD,

we first looked at the interaction effect between country and dialysis modality on mortality

and found that this was statistically insignificant (type 3 test, p=0.19). In addition, we added the

ratio HD/PD patients (HD vs PD, HR: 1.00, 95% CI 0.62-1.62) and the proportion of HD

patients per country (HD vs PD, HR: 0.96, 95% CI 0.59-1.57) as a proxy for HD country

experience to the Cox model, which had little effect on the hazard ratio. Survival remained

similar after excluding those countries that had no infants treated with HD (aHR: 1.07, 95% CI

0.64-1.70).

Table 3. The cumulative incidence of death (with transplantation as a competing risk), modality switching (with

both death and transplantation as competing risks), and transplantation (with death as a competing risk), and

corresponding adjusted hazard ratios for HD vs. PD patients.

Outcome

Overall HD PD HD vs. PD adjusted

hazard ratio (95% CI) 5-year cumulative 5-year cumulative 5-year cumulative

incidence % (95% CI) incidence % (95% CI) incidence % (95% CI)

Death 16.1 (13.8-18.5) 16.3 (9.60-23.1) 16.1 (13.6-18.7) 1.06 (0.67-1.67)

Dialysis switch 25.5 (22.7-28.3) 30.9 (23.1-38.7) 24.6 (27.5-21.6) 1.54 (1.07-2.20)

Transplantation 70.2 (67.1-73.4) 69.0 (60.2-77.9) 70.5 (67.1-73.8) 0.95 (0.70-1.29)

Infant dialysis

122

Figure 1. Cumulative incidence curves for A) death (with transplantation as a competing risk), B) modality

switching (with both death and transplantation as competing risks), and C) transplantation (with death as a

competing risk).

Infant dialysis

123

Technique survival

The overall cumulative incidence for dialysis modality switching at 1, 2, and 5 years was 14.5%

(95% CI 12.4%-16.7%), 19.7% (95% CI 17.3%-22.2%), and 25.5% (95% CI22.7%-28.3%),

respectively. The 5-year cumulative incidence for dialysis modality switching is presented by

dialysis modality in table 3 and figure 1B. Patients on HD had a 1.64-fold higher risk of changing

dialysis treatment (95% CI 1.17-2.31; p=0.004), as compared to patients on PD. This effect

remained even after adjustment for confounders (aHR: 1.54, 95% CI 1.07-2.20; p=0.02) and

was stronger during the first year of dialysis (aHR: 2.79, 95% CI 1.81-3.99). We registered 198

modality failures among PD and 44 among HD infants. Reasons for modality failure are

reported in detail for patients where this information was available in appendix 2: peritonitis

(63%) was the main cause of failure in PD patients followed by exit site or tunnel infection

(13%), and patient/family choice (56%) was the main cause in HD patients followed by vascular

access failure (20%).

Overall, older patients had a lower risk of changing the type of dialysis (HR: 0.96 per month,

95% CI 0.93-0.99; p=0.03). This was not the case among PD patients (HR: 0.98 per month,

95% CI 0.95-1.02; p=0.4), but was strongly present among HD patients (HR: 0.82 per month,

95% CI 0.75-0.91; p<0.001). Among patients starting on PD and compared to CAKUT, those

with metabolic disorders were more likely to change to HD (aHR: 6.29, 95% CI 3.32-11.94;

p<0.001), as were patients with hereditary nephropathies (aHR: 1.75, 95% CI 1.04-2.95;

p=0.04). The likelihood of changing from HD to PD was not affected by the underlying renal

disease. There were differences in the likelihood of switching dialysis modalities between

countries (appendix 1). Compared to other European counties, the UK had a significant

increased risk of modality switching (HR: 1.90, 95% CI 1.31-2.77).

Time to transplantation

Within 5 years after start of dialysis, 70.2% (95% CI 67.1%-73.4%) of all patients had received a

kidney transplant. Information about the donor source was available in 524 out of 608

transplants, showing that 63% of patients had received a deceased donor and 37% a living-

related donor. The 5-year cumulative incidence of transplantation is presented by dialysis

modality in table 3 and figure 1C. The probability of receiving a transplant did not differ

Infant dialysis

124

significantly between the two treatment groups (HR: 1.03, 95% CI 0.78-1.37), even after

adjustment for age, gender, and primary renal disease (aHR: 0.95, 95% CI 0.70-1.29).

Factors affecting the chance of transplantation included age at dialysis initiation and primary

renal disease. Older patients were more likely to receive a transplant (HR: 1.05 per month

increase in age, 95% CI 1.02-1.07; p<0.0001), as were patients with glomerulonephritis

(compared to CAKUT, adjusted for age, aHR: 1.65, 95% CI 1.09-2.48; p=0.02), hereditary

nephropathies (compared to CAKUT, adjusted for age, aHR: 1.54, 95% CI 1.15-2.06; p=0.004)

and metabolic disorders (compared to CAKUT, HR: 2.23, 95% CI 1.43-3.47; p<0.001). The

chance of receiving a transplant also differed significantly between countries, with notably

Scandinavian countries showing higher transplant rates (appendix 1).

DISCUSSION

In this study, we report on the largest cohort of infants receiving maintenance dialysis ever

examined. Overall survival in infants was 84% at 5 years after commencing dialysis, with similar

mortality rates and transplant access in PD and HD patients, but a higher risk of early

technique failure among those treated with HD.

Mortality rates in children receiving chronic dialysis are at least 30 times higher than in the

general pediatric population, with even higher relative risks in very young children [8]. Five

published reports have described the short and long-term survival of infants receiving

maintenance dialysis [115, 117, 145, 204, 213] which ranged from 62% to 87% at 1 year and

from 50% to 79% at 5 years. Our study places the average European infant on dialysis in the

upper range of the reported survival. Recent pediatric RRT studies describe a trend of

improving patient survival. Among 628 infants on chronic PD in the NAPRTCS database, the 3-

year survival on dialysis improved from 75.8% to 84.6% between the periods 1992-1999 and

2000-2012 [110], and survival in infants who initiated chronic dialysis before 1 year of age

approached that of older children in the more recent cohort. Based on previous studies, a

mortality “risk profile” seems evident; apart from age at dialysis initiation, survival is influenced

by small-for-gestational-age birth [117], primary renal disease [115, 142], the presence of co-

morbidities [142, 145, 203], and residual urinary output [206]. Our study provides

Infant dialysis

125

corroborative evidence that early age at initiation of RRT and non-CAKUT etiology of ESRD

are predictors of death on dialysis.

To date, the lack of sufficiently large infant cohorts has precluded the analysis of the impact of

dialysis modality on survival in infants. The high rate of infants starting RRT in Europe and the

establishment of a pan-European population based pediatric RRT registry allowed us to analyze

short- and long-term mortality in this age group by dialysis modality. In our cohort, 13.5% of

ESRD infants were started on HD. This proportion is higher than that reported in the 2011

NAPRTCS report [114], where 70 (8.2%) out of 927 children aged 0-1 year initiated dialysis

on HD. Analyzing survival in more than 1600 children and adolescents with ESRD in Australia

and New Zealand, McDonald et al. found no differences in mortality risk between HD and PD

patients [8]. However, only 26 out of 1634 children included in this study were younger than

12 months. In a large US cohort of children initiating dialysis between 1990 and 2010 [34],

Mitsnefes et al. reported a protective role of PD as compared with HD in children younger

than 5 years at RRT start, but the proportion of infants was again negligible. In the current

study, we found no difference in mortality risk between infants selected to start dialysis on PD

or HD, respectively. Extracorporeal RRT is generally considered a reserve technology in

infants, to be used when PD fails [120]. Current recommendations suggest HD as the initial

modality in infants with metabolic disorders and in those with clinical contraindications for PD.

Our findings suggest that HD is an equally safe alternative when PD fails, it is contraindicated

or in those settings in which PD is unavailable or unfeasible.

Our results show that the overall probability for shifting the dialysis modality was higher in

infants treated with HD as compared with PD. We did not find previous studies comparing

technique survival in small children on chronic dialysis, since in most single-center case-series

younger children were almost exclusively treated with PD. In our study, HD was most often

withdrawn because of parental decision and poor central line function. HD in infants is most

often performed in-center and with a median time of 12 hours per week. This schedule

relieves families from burden of home therapy, but still requires a great effort; small patients

have to be brought regularly to the pediatric dialysis unit, creating potential problems in the

parents’ work environment. The maintenance of a safe and efficient vascular access is also

crucial in small children requiring RRT. Poor central line function due to catheter malposition

Infant dialysis

126

or thrombosis, and line infections are the most common limiting factors in achieving successful

HD. When HD was used in infants for a continuous period of 3 months or longer, Shroff et al.

found an infection rate leading to line revision of 35% [209], a value that is higher than the

rates reported in other series including older children [214]. In our study, 20% of patients

(where this information was available) had HD access failure, whereas two recent single-

center studies reported remarkably low catheter infection rates and prolonged catheter

survival times in infants receiving chronic HD [121, 211]. While PD enables preservation of

vascular access for future use, when prescribing chronic HD in small children, both the

immediate impact and potential long-term sequelae of a central vascular access positioned

early in patients who will have a long period of RRT ahead of them should be considered.

Experienced personnel devoted to the care and handling of HD catheters may represent a

crucial factor for both catheter survival and outcome of infants receiving this mode of therapy.

Since small body size often precludes pre-emptive transplantation, infants usually spend a

longer period on dialysis than older children. In our case, more than half of patients had

received a kidney transplant after 3 years of dialysis, and 70% after 5 years. Importantly, the

choice of pre-transplant dialysis modality did not influence access to transplantation. This

concept has never been analyzed in children, although it is known in the adult dialysis

population [215].

We are aware of the limitations of this registry study covering a long period of time during

which the management of infants with ESRD may have changed (although era had little effect

on the outcomes studied). Firstly, our ability to control for confounders (i.e. co-morbidities,

urine output, patient’s socioeconomic status and ethnicity) was limited by large amounts of

missing data. In addition, we cannot exclude the possibility of residual confounding due to

unmeasured variables or potential confounding by indication. In evaluating the association

between exposure and outcomes, we used hard measures; however, there may be other

outcome measures, such as quality of life, growth and development, nutritional status, and

cardiovascular function that may be deemed equally important when discussing the long-term

picture of infants receiving dialysis.

Infant dialysis

127

Despite these limitations, most of which are inherent to observational research, to our

knowledge this is the largest study performed to date to compare clinical outcomes in infants

on PD and HD. The study provides evidence that may help physicians in the decision-making

process when facing the management of ESRD in infants. According to our results, patient

survival and access to kidney transplantation appeared similar for infants initiating dialysis on

HD and PD, suggesting that HD may represent a safe and effective alternative dialysis modality

in infants with ESRD accepted for RRT. The choice of dialysis modality in this age group should

take into account specific benefits and drawbacks of either technique, thus individualizing the

choice that best fits the needs of the patient and family.

Appendix 2.

Cause HD PD

N (%) N (%)

Abdominal complications (except for peritonitis in PD) 0 (0) 2 (4.3)

Cardiovascular instability 1 (4) 0 (0)

Dialysis access failure 5 (20) 1 (2.2)

Inadequate dialysis 1 (4) 1 (2.2)

Patient/family choice 14 (56) 3 (6.5)

Peritoneal catheter infection 0 (0) 6 (13)

Peritoneal membrane failure 0 (0) 1 (2.2)

Peritonitis 0 (0) 29 (63)

Other 4 (16) 3 (6.5)

Total 25 (100) 46 (100)

Infant dialysis

128

Appendix 1.

HD PD Total Patient survival Modality switching Transplantation

(n) (n) (n) HR (95% C.I.) HR (95% C.I.) HR (95% C.I.)

Austria 7 26 33 1.07 (0.66-1.73) 1.11 (0.55-2.25) 2.15 (1.18-3.92)

Bosnia and Herzegovina 6 1 7 1.07 (0.61-1.87) 0.85 (0.33-2.19) 0.55 (0.11-2.85)

Belgium 10 23 33 0.88 (0.54-1.44) 0.70 (0.35-1.40) 0.77 (0.40-1.47)

Bulgaria 1 0 1 1.00 (0.55-1.80) 0.97 (0.32-2.95) 0.99 (0.13-7.81)

Belarus 1 3 4 1.02 (0.58-1.81) 1.63 (0.57-4.63) 0.89 (0.27-2.88)

Switzerland 2 26 28 0.82 (0.50-1.37) 1.40 (0.72-2.72) 0.91 (0.48-1.72)

Czech Republic 2 6 8 1.22 (0.71-2.12) 0.91 (0.35-2.36) 0.30 (0.07-1.22)

Germany 2 14 16 1.00 (0.58-1.72) 0.65 (0.25-1.68) 3.41 (1.45-8.05)

Denmark 3 13 16 0.98 (0.57-1.68) 0.85 (0.36-1.99) 1.98 (0.96-4.08)

Spain 23 62 85 1.15 (0.75-1.78) 0.87 (0.49-1.55) 2.59 (1.55-4.32)

Finland 1 93 94 0.66 (0.41-1.06) 1.48 (0.85-2.57) 4.63 (2.74-7.84)

France 12 61 73 0.92 (0.59-1.45) 1.40 (0.84-2.35) 1.87 (1.11-3.16)

Greece 0 19 19 0.95 (0.56-1.58) 1.01 (0.47-2.17) 0.19 (0.07-0.53)

Croatia 1 6 7 1.09 (0.62-1.92) 2.66 (1.00-7.09) 0.44 (0.13-1.54)

Hungary 0 5 5 0.92 (0.52-1.63) 1.20 (0.47-3.08) 0.67 (0.21-2.07)

Italy 6 121 127 0.77 (0.49-1.21) 1.26 (0.69-2.29) 1.15 (0.63-2.11)

Lithuania 0 4 4 0.93 (0.53-1.65) 0.72 (0.27-1.94) 0.32 (0.08-1.37)

The Netherlands 7 48 55 1.04 (0.67-1.63) 1.50 (0.90-2.50) 0.96 (0.55-1.68)

Norway 3 12 15 1.01 (0.58-1.74) 0.92 (0.39-2.20) 5.25 (2.50-11.03)

Poland 0 34 34 1.20 (0.75-1.93) 0.59 (0.27-1.30) 0.97 (0.51-1.86)

Portugal 0 13 13 1.00 (0.58-1.72) 0.83 (0.36-1.94) 0.51 (0.21-1.23)

Romania 3 5 8 1.20 (0.68-2.10) 0.80 (0.29-2.23) 1.02 (0.24-4.34)

Serbia 1 1 2 0.98 (0.55-1.76) 0.79 (0.28-2.19) 1.58 (0.43-5.79)

Russia 3 49 52 1.21 (0.77-1.91) 0.59 (0.29-1.19) 0.69 (0.36-1.34)

Sweden 0 51 51 1.36 (0.86-2.17) 1.16 (0.59-2.28) 5.87 (3.40-10.15)

Slovenia 2 1 3 0.96 (0.54-1.72) 1.26 (0.43-3.71) 0.58 (0.11-3.06)

Slovakia 0 2 2 0.97 (0.54-1.73) 0.90 (0.31-2.62) 0.75 (0.12-4.67)

Turkey 0 25 25 0.96 (0.58-1.58) 0.41 (0.17-0.95) 0.16 (0.05-0.59)

United Kingdom 50 193 243 0.96 (0.69-1.33) 1.90 (1.31-2.77) 1.27 (0.79-2.05)


8

The association of donor and

recipient age with graft survival

in paediatric renal transplant

recipients: an ESPN/ERA-EDTA

Registry study

Nicholas C Chesnaye, Karlijn J van Stralen, Marjolein Bonthuis,

Jaap W Groothoff, Jérôme Harambat, Franz Schaefer, Nur

Canpolat, Arnaud Garnier, James Heaf, Huib de Jong, Søren

Schwartz Sørensen, Burkhard Tönshoff, Kitty J Jager

Nephrol Dial Transplant 2017 Jul; Accepted

ABSTRACT

Background: The impact of donor age in paediatric kidney transplantation is unclear. We

therefore examined the association of donor-recipient age combinations with graft survival in

children.

Methods: Data for 4686 first kidney transplantations performed in 13 countries in 1990-2013

were extracted from the ESPN/ERA-EDTA Registry. The effect of donor and recipient age

combinations on graft failure risk, stratified by donor source, was estimated using Kaplan-

Meier survival curves and Cox regression, whilst adjusting for sex, primary renal diseases with

a high risk of recurrence, pre-emptive transplantation, year of transplantation, and country.

Results: The risk of graft failure in kidneys from older (50-75 years old) donors was similar to

that of younger living donors (aHR 0.74, 95% CI 0.38-1.47). Deceased donor (DD) age was

non-linearly associated with graft survival, with the highest risk of graft failure found in the

youngest donor age group 0-5 years (compared with donor ages 12-19 years, aHR 1.69, 95%

CI 1.26-2.26), especially among the youngest recipients (0-11 years). DD age had little effect

on graft failure in recipients ages 12-19.

Conclusions: Our results suggest that donation from older LDs provide excellent graft

outcomes in all paediatric recipients. For young recipients, the allocation of DD over the age

of 5 should be prioritized.

Graft survival

131

INTRODUCTION

It has been well established that in children with end-stage renal disease, renal transplantation

offers optimal patient survival probabilities, cognitive development, quality of life, and growth

[6–10]. Deceased donor (DD) kidney allocation policies aim to reduce waiting times and

provide high-quality grafts to the best-matched recipients in order to improve post-transplant

patient- and graft survival. The donor-recipient matching process is often based on a

composite points-based system involving factors such as waiting time, HLA and blood group

matching, percentage of panel-reactive antibodies, distance between donor and recipient, and

medical urgency.

In the United States and most European countries, a deceased donor-recipient ‘young-for-

young’ matching policy has been implemented, as this would reduce the number of size

mismatches, the risk of hypoperfusion, and graft non-function [37–42]. In addition, grafts from

paediatric donors show a superior long-term kidney function compared with grafts from adult

donors, likely due to their ability to adapt to the growing child [43, 44]. Conversely, earlier

reports have shown a higher risk of graft loss in recipients of (very) young donors due to

surgical complications, high rates of graft thrombosis, early rejection, and hyperfiltration injury

[39, 40, 45–47].

It is known that living donation offers better long-term graft survival and improved growth

compared with deceased donation [11, 48, 49]. In adult transplant recipients, it has been

established that advanced living donor (LD) age is associated with poorer graft survival

compared with younger LDs [216–218]. In paediatric LD transplantation, it remains unclear

whether utilizing kidneys from elderly donors, such as from grandparents, may affect graft

survival compared with kidneys from younger donors.

Although recipient and donor age are both known to affect graft survival, a potential

interaction effect between recipient- and donor age on graft survival has not previously been

explored in depth, and physicians may question whether to accept an organ from a

grandparent or from a DD. The current study therefore aims to optimize the utilization of

donor grafts by examining how the relationship between donor age and recipient age affects

graft survival in paediatric kidney transplant recipients.

Graft survival

132

METHODS

Data source and study population

This observational cohort study was performed using incident patient data from the European

Society for Paediatric Nephrology / European Renal Association – European Dialysis and

Transplantation Association (ESPN/ERA-EDTA) Registry database for 13 European countries

where donor age has been reported [51]. Denmark, Finland, France, Greece, Norway,

Portugal, Spain, and The United Kingdom provided data from January 1, 1990, to December

31, 2013, Belgium from 2006, Belarus from 2010, Germany from 2012, and the Netherlands

and Turkey from 2007. Data was extracted on patients’ date of birth, sex, primary renal

disease (PRD), date of first transplantation, and events such as graft failure, death, changes in

renal replacement therapy (RRT) modality, and transfer out of the Registry, as well as on

donor source and age.


The primary outcome studied was 5-year graft survival. The association between donor age

and recipient age on graft failure risk, stratified by donor source, was estimated using Kaplan-

Meier survival curves and Cox regression models. After examination of the splined effect of

both donor and recipient age, age categories were defined for DD age groups as 0-5, 6-11, 12-

19, 20-49, and 50+, and for recipient age groups as 0-5, 6-11, and 12-19 years at

transplantation. All analyses were stratified by donor source and adjusted for recipient age at

transplantation, donor age, sex, PRD with a high risk of recurrence (defined as focal segmental

glomerulosclerosis, membrano-proliferative glomerulonephritis, haemolytic uraemic syndrome,

oxalosis, or systemic lupus erythematosus), pre-emptive transplantation, calendar year of

transplantation, and country. As DD age and recipient age displayed a non-linear relationship

with graft loss, both were included in the Cox model using restricted cubic splines. Patients

were censored when lost to follow-up, at end of study, or after 5 or 10 years of follow-up,

whichever came first. In countries providing data over the full study period (Denmark, Finland,

France, Norway, Spain, and United Kingdom, representing 90.2% of available data), we

compared whether graft survival had improved between the periods 1990-2000 and 2000-

2013. We also studied death-censored graft survival in a sensitivity analysis. Results were

similar and therefore not described. Analyses were performed using SAS 9.4 and R version

3.3.1.

Graft survival

133

RESULTS

Recipient and donor characteristics

A total of 4 686 patients received a first kidney transplant between 1990 and 2013 across 13

European countries. The majority of grafts were obtained from DD (N=3517, 75.1%). Patient

characteristics stratified by donor source are presented in table 1. In countries providing data

over the full study period (N=4229, 90.2%), the proportion of LD transplants increased from

19.7% in 1990-2000 to 25.4% in 2000-2013 (p<0.0001). Amongst LD, 78.6% were between 30-

50 years of age, reflecting that the majority of living donation likely comes from parents.

Among DD, 61.4% received a young-for-young (both recipient and donor under 19 years of

age) donor, which varied by country (Appendix 1). The percentage of deceased adult donation

(>20 years of age) was 44.5% in adolescent recipients (12-19 years old), whereas this was

29.4% in recipients under 12 years of age.

Table 1. Patient characteristics stratified by donor source.

Deceased donor Living donor p-value

N=3517 N=1168

Median age at RRT (IQR), years 10.6 (5.2-14.2) 11.5 (5.6-15.2) 0.0001

Median age at first Tx (IQR) 11.7 (6.6-15.2) 11.9 (6.3-15.7) 0.36

Median donor age (p5, p25, p75, p95)) 16 (3, 10, 28, 49) 41 (28, 36, 46, 55) <0.0001

Gender (male) 2095 (59.6%) 700 (59.9%) 0.86

High risk of reccurent PRD 585 (16.6%) 122 (10.4%) <0.0001

Pre-emptive Tx 746 (21.2%) 504 (43.1%) <0.0001

Graft survival

A total number of 537 graft failures occurred during 16 221 patient follow-up years,

corresponding with a 5-year graft failure rate of 33.1 per 1000 patient years. The overall 5-

year and 10-year graft survival rates were 86.3% (95% CI 85.2%-87.4%) and 73.2% (95% CI

71.1%-75.2%). In DD recipients, the 5-year and 10-year graft survival rates were 85.1% (95%

CI 83.7%-86.3%) and 72.0% (95% CI 69.7%-74.3%) compared with 90.1% (95% CI 87.8%-

92.0%) and 76.9% (95% CI 72.1%-81.0%) in LD recipients (log-rank p-value < 0.0001). DD

recipients had a higher risk of graft failure compared with LD recipients (HR 1.68, 95% CI

1.24-2.12). In countries providing data over the full study period (N=4229, 90.2%), graft failure

Graft survival

134

risk did not change significantly over time for DD recipients (1990-2000 vs. 2000-2013, HR

1.03 95% CI 0.84-1.26), whereas in LD recipients graft survival improved over time (1990-

2000 vs. 2000-2013, HR 1.60 95% CI 1.02-2.52, appendix 2). This improvement was stronger

after adjustment for country differences and changes over time in the rate of pre-emptive

transplantation (HR 1.77 95% CI 1.10-2.83). Early graft failure, defined as graft failure within 3

months post-transplant, accounted for more than half (N=287, 53.4%) of the graft failures.

The association between recipient age and graft survival

Graft survival varied by recipient age. In recipients aged 0-5, the 5-year and 10-year graft

survival rates were 84.2% (95% CI 81.5%-86.6%) and 75.2% (95% CI 71.7%-78.4%), in

recipients ages 6-11, 89.9% (95% CI 87.9%-91.5%) and 76.7% (95% CI 73.4%-79.6%), and in

recipients ages 12-19 the 5-year graft survival was 84.6% (95% CI 82.8%-86.3%). In adolescents

aged 12-19 years, there were insufficient patients with complete follow-up to accurately assess

10-year graft survival, as patients are lost to follow up when transferred to adult care. During

5 years of follow-up, recipient age was non-linearly associated with graft loss (figure 1,

p<0.0001) and did not differ significantly by transplant source (p-value for interaction term =

0.80) or study period (p-value for interaction term = 0.94, appendix 3). Compared with the

recipient age group 6-11, graft failure risk was higher in patients transplanted under 5 years of

age (aHR 1.60, 95% CI 1.24-2.07) and during adolescence (aHR 1.54, 95% CI 1.24-1.93). Similar

results were obtained for 10-years of follow-up (<5 years vs. 6-11 years aHR 1.23, 95% CI

1.00-1.51).

Notably, the percentage of early graft failure (within the first three months post-

transplantation) was higher in recipients under 5 years of age (63.4% of all the grafts lost in the

first 5 years), compared with recipients ages 6-11 (42.0%), and recipients ages 12-19 (40.1%),

possibly reflecting the surgical difficulties of transplantation in the youngest patients. In patients

with a functioning graft at 3 months post-transplant, compared with the recipient age group 6-

11, graft failure risk was similar in patients transplanted under 5 years of age (aHR 1.01, 95%

CI 0.69-1.48) and higher during adolescence (aHR 1.71, 95% CI 1.28-2.29).

Graft survival

135

Figure 1. Effect of recipient age at transplantation on the hazard of graft loss with 95% confidence bands,

adjusted for donor age, for primary renal disease with a high risk of disease recurrence, sex, pre-emptive

transplantation, calendar year of transplantation, and transplant source.

The association between donor age and graft survival

The age of a LD was not associated with graft survival (aHR 1.00 per additional year, 95% CI

0.97-1.03, figure 2A), even when comparing older LD (N=180, 50-75 years old) with younger

(N=989, 18-50 years old) LD (aHR 0.74, 95% CI 0.38-1.47). DD age was non-linearly

associated with graft survival (figure 2B, p<0.0001). The adjusted risk of graft failure was

highest in recipients of the youngest donor grafts, decreasing exponentially until donor age of

approximately 12, after which the risk of graft failure gradually rose. Compared with donor

ages 12-19, the youngest DD age group 0-5 showed the highest risk of graft failure (aHR 1.69,

95% CI 1.26-2.27). This effect remained after excluding donors under 2 years of age (aHR

1.50, 95% CI 1.09-2.06) and did not differ by study period (p-value for interaction term =

0.53). Hazard ratios for DD age groups 6-11 (aHR 1.20, 95% CI 0.90-1.60), 20-49 (aHR 1.13,

95% CI 0.87-1.46), and 50+ (aHR 1.44, 95% CI 0.87-2.37) were not statistically significantly

different compared with the donor age group 12-19.

Graft survival

136

Figure 2. Effect of living (A) and deceased (B) donor age on the hazard of graft loss with 95% confidence

bands, adjusted for recipient age at transplantation, PRD with a high risk of disease recurrence, sex, calendar

year of transplantation, and pre-emptive transplantation.

A.

B.

Graft survival

137

Graft survival in donor-recipient age combinations

In all recipient age groups, living donation provided the lowest adjusted risk of graft failure. To

answer the question which DD age provides the best graft survival in the absence of a suitable

LD, we focused on the comparison of donor and recipient age combinations. Hazard ratios

and Kaplan Meier curves are presented in table 2 and figure 3. The non-linear effect of DD age

on graft failure is presented by recipient age group in appendix 4. In the youngest recipients’

ages 0-5, the risk of graft failure was highest in DD ages 0-5 (compared to DD 12-19, aHR

2.01, 95% CI 1.11-3.67). In transplant recipients ages 6-11, compared with the DD age group

12-19, graft failure risk was higher in those receiving a graft from a 0-5 year old DD (aHR 2.38,

95% CI 1.31-4.32), whereas other donor age groups showed similar graft failure risks. In

adolescent recipients’ ages 12-19, graft survival was similar across all DD age groups,

suggesting that adolescence itself is a more influential determinant of graft failure.

DISCUSSION

We showed that recipient- and donor-age combinations strongly affect graft survival in the

largest cohort of European paediatric renal transplant recipients studied to date. First, we

demonstrated a non-linear effect of recipient age on graft survival, finding a higher rate of graft

failure in the youngest and adolescent recipients. Second, we described a non-linear effect of

DD age on graft survival, with the highest rates of graft failure occurring in recipients of the

youngest donor grafts. Importantly, LD age did not affect the risk of graft failure, independent

of recipient age. Lastly, we established a high risk of graft failure in younger, pre-adolescent,

recipients receiving grafts from DD under 5 years of age, whereas in adolescent recipients,

DD age seemed less important than adolescence itself.

Graft failure was highest in the youngest and adolescent transplant recipients. In adolescents,

the poor graft survival has been attributed to poor compliance to immunosuppression

regimens [219–221]. Recipients under 5 years of age show a higher risk of graft failure

especially during the first three months post-transplantation, most likely reflecting the

consequences of surgical difficulties of transplantation in the youngest patients [222, 223].

After having successfully bridged this initial high risk period, long-term graft survival was equal

compared with patients aged 6-11 and superior compared with adolescents.

Graft survival

138

Table 2. Five-year graft survival and hazard ratios for graft failure risk by donor-recipient age groups. Hazard

ratios were adjusted for primary renal disease with a high risk of disease recurrence, sex, country, and pre-

emptive transplantation.

Recipient age Donor age Source N 5-yr graft survival Hazard ratio

Crude Adjusted

0-5 0-5 DD 152 74.9 (68.2-82.3) 2.49 (1.42-4.37) 2.01 (1.11-3.67)

0-5 6-11 DD 161 80.9 (74.8-87.5) 1.75 (0.97-3.15) 1.51 (0.82-2.76)

0-5 12-19 DD 166 88.7 (83.9-93.8) Reference Reference

0-5 20-49 DD 148 83.1 (77.1-89.5) 1.56 (0.84-2.87) 1.47 (0.77-2.80)

0-5 50+ DD 15 87.1 (71.9-100) 1.24 (0.29-5.32) 1.26 (0.27-5.83)

0-5 All LD 242 90.0 (86.1-94.2) 0.88 (0.47-1.65) 0.77 (0.38-1.56)

6-11 0-5 DD 145 80.8 (74.5-87.6) 2.91 (1.63-5.2) 2.38 (1.31-4.32)

6-11 6-11 DD 201 89.2 (84.8-93.8) 1.47 (0.79-2.73) 1.22 (0.64-2.32)


6-11 20-49 DD 273 90.3 (86.8-94.2) 1.36 (0.75-2.46) 1.19 (0.64-2.21)

6-11 50+ DD 37 81.7 (69.3-96.4) 2.65 (1.06-6.59) 1.63 (0.60-4.42)

6-11 All LD 282 93.9 (90.8-97.1) 0.76 (0.38-1.50) 0.69 (0.31-1.51)

12-19 0-5 DD 132 83.0 (76.5-90.0) 1.26 (0.78-2.03) 1.21 (0.74-1.96)

12-19 6-11 DD 257 82.9 (77.9-88.1) 1.13 (0.77-1.65) 1.06 (0.71-1.57)


12-19 20-49 DD 797 84.6 (81.4-87.9) 0.98 (0.72-1.32) 0.99 (0.72-1.35)

12-19 50+ DD 86 76.6 (65.3-89.7) 1.41 (0.80-2.48) 1.36 (0.74-2.49)

12-19 All LD 645 87.5 (84.0-91.1) 0.68 (0.48-0.96) 0.70 (0.47-1.05)

Similar results have been demonstrated in the UK and Ireland [224], and by the 2014

NAPRTCS report, which shows the poorest graft survival in DD recipients under 2 years of

age, especially during the initial post-transplant period [11]. In contrast to our results,

NAPRTCS data showed the best graft survival with LD transplantation in this recipient age

group [11]. However, improvement may be possible, as in the Nordic countries within

Scandiatransplant, graft survival rates have improved greatly over the past decades, especially

among recipients under 2 years of age receiving a DD; between the periods 1982-1996 and

1997-2012, the number of transplantations doubled in this group, and the 1-year and 3-year

graft survival rates improved from 70% to 94.6%, and from 60% to 94.6%, and has become

superior to the graft survival of older recipient age groups. The authors attributed this

Graft survival

139

improvement to better surgical techniques and changes in anticoagulation and

immunosuppression protocols [225].

DD age is of major importance for graft survival. Similar to our findings, early reports from the

US described a U-shaped association between DD age and graft failure, with higher rates of

graft failure occurring in the youngest and oldest donor kidneys, and an ‘optimal’ donor age

determined at 20-25 years of age [39, 40, 222, 226, 227]. Between 1998 and 2007, the

Collaborative Transplant Study found poor outcomes in donors older than 49 and in those

younger than 11, but similar 10-year survival rates for DD ages 11-49, recommending that

kidneys from DDs in this age range be allocated to paediatric recipients. They also reported a

reluctance to procure the youngest DD kidneys for transplantation, demonstrating a

considerable decline in the proportion of donor ages 0-5, decreasing from 16.4% during 1988-

1997 to 7.5% during 1998-2007 [228]. More recently, the NAPRTCS 2014 annual transplant

report cited poor graft survival in recipients of donors under 2 and above 50 years of age, but

comparable graft survival rates within this range [114]. In contrast to the NAPRTCS, we found

no significant higher risk of graft loss in DDs above 50 years of age, although we did find a

small upwards trend in increasing risk with older age. However, the number of older donors in

our study may be too small to obtain statistical significance.

We demonstrated high rates of graft loss in recipients of donors under 5 years of age,

especially in pre-adolescent recipients. This effect was retained after excluding the high risk

donors under 2 years of age. The poor outcomes found in the youngest donor kidneys has

previously been attributed to surgical complications, high rates of graft thrombosis, early

rejection, and hyperfiltration injury due to size mismatching [39, 40, 45, 46]. Nonetheless,

multiple studies have shown that even the youngest donors provide acceptable graft outcomes

in both adults and children when transplanted en bloc [229–233]. Since December 2004 in the

Eurotransplant region, en bloc procurement is compulsory in donors under 2 years of age and

recommended in donors between 2 and 5 years of age [234]. In other European countries, the

en bloc transplantation policy differs, but remained the same during the study period (Appendix

5). As en bloc transplantation constitutes a small percentage of all paediatric transplantations,

any potential confounding effect will be marginal.

Graft survival

140

Figure 3. Kaplan-Meier survival curves for 5-year graft survival, stratified by deceased donor age groups,

recipient age groups, and donor source.

Graft survival

141

We found a significantly better graft survival among recipients of LD compared with DD

kidneys, and an improvement in graft survival in LD over time. Similar to the NAPRTCS and

ANZDATA reports, we demonstrated an increase in the use of LD between the periods

1990-2000 and 2000-2013 [63, 165]. In contrast to the adult transplant population where

older LDs showed poorer graft survival rates [216–218], we found no association between LD

age and graft loss, even in donors between 50 and 75 years of age, although the number of

older living donors in our study may be too small to obtain statistical significance. In line with

our results, a USRDS study found no effect of living LD age on graft survival in patients under

21 years of age [235], and two single-centre studies found no difference in graft survival

between paediatric recipients receiving a LD transplant from a grandparent or a parent [236,

237]. Furthermore, Dale Shall et al. demonstrated in the US that paediatric recipients of LDs

up to the age of 54 years provide higher long-term graft survival rates compared with all DD

age groups. Graft survival was only reduced in recipients of LD over 55 years old after five

years of follow-up, but was still similar to that of the ideal 18-34 year old DD [227].

Altogether, these data suggest that older LDs provide excellent graft outcomes in paediatric

recipients, independent of recipient age, and should be preferred, given adequate matching,

over kidneys from DDs.

Some limitations of our study need to be acknowledged, including missing data on the cause of

graft failure, which could further help differentiate early from long-term graft failure, and the

issue of unmeasured donor and recipient variables, such as the percentage of panel-reactive

antibodies, the number of human leukocyte antigen mismatches, the cold ischaemia time,

immunosuppression regimens, ethnicity, and whether or not the kidneys were transplanted en

bloc, which may have led to residual confounding. Furthermore, as France, the UK, and Spain

compose almost 80% of the total study population and only 13 countries were included in this

analysis, our results are hardly generalizable to the rest of Europe.

Although many countries have implemented a ‘young-for-young’ donation policy, where

‘younger’ DDs are allocated preferentially to children, allocation schemes vary widely in their

definition of ‘young’. For example, Eurotransplant preferentially allocates donors under 16 to

recipients under 16, whereas in the UK, recipients under 18 are preferentially allocated well-

matched donors under 50, and in Serbia recipients under 21 have regional priority for donors

Graft survival

142

under 30 [42]. As both recipient and DD age show a non-linear effect on graft failure risk, a

differential graft failure risk exists dependent on recipient and DD age combinations. We

therefore recommend fine-tuning the young-for-young allocation schemes by increasing the

granularity of the definitions of ‘young’. Specifically, we recommend prioritizing the allocation

of DDs over the age of 5 (with no specific upper age limit) to pre-adolescent recipients, and a

cautionary use of ‘youngest-for-youngest’ allocation. In adolescents, we found poor graft

survival rates for DD transplantation across the entire DD age range, precluding any specific

donor age recommendations for allocation policy.

Graft survival

143

Appendix 1. The number and percentage of young-for-young deceased donor transplantations (recipient and

donor both under 19 years of age) by country.

Country N (%)

BE 15 (28.3)

BY 6 (16.7)

DE 14 (53.8)

DK 25 (31.6)

ES 431 (60.9)

FI 47 (29.0)

FR 1156 (73.2)

GR 3 (20.0)

NL 14 (20.6)

NO 8 (29.6)

PT 64 (71.1)

TR 3 (60.0)

UK 375 (56.1)

Total 2161 (61.4)

Appendix 2. Graft survival and numbers at risk for DD and LD by transplantation era.

Graft survival

144

Appendix 3. Graft survival and numbers at risk by recipient age and transplantation era.

Graft survival

145

Appendix 4. Effect of deceased donor age on the hazard of graft loss with 95% confidence bands, stratified by

recipient age group at transplantation, and adjusted for PRD with a high risk of disease recurrence, sex,

calendar year of transplantation, and pre-emptive transplantation. A) Recipients ages 0-5, B) recipients ages 6-

11, C) recipients ages 12-19.

A

Graft survival

146

B

C

Graft survival

147

Appendix 5. Survey regarding en bloc transplantation policy.

Country

Accept donors <

6 years old?

If yes, only for young recipients?

If yes, only when

procured en bloc?

Accept donors betw

een 6-11 years old?

If yes, only for young recipients?

If yes, only when

procured en bloc?

Has this policy changed

between 1990-2013?

Denmark Yes No Only for very young donors below 2-4 years

Yes No No No

Finland Yes Preferably No Yes

Preferably No No

France Yes, but decrease over time in <3 and no <1 Yes No Yes No No No

Greece Yes No No Yes No No No

Norway Yes No Only when donor is <2 or 9-12kg Yes No No No

Portugal Yes, reject < 2 years of age

No No Yes No No No

Spain Yes, generally reject

<3 years old Preferably

Only when donor is <3 and recipient is teenager Yes No No No

United Kingdom

Some centres reject < 5 donors No Yes Yes No No No

Belgium Yes No Only with good size match Yes No No No

Belarus Yes Yes Yes Yes No No No

Germany Yes No No Yes No No No

Netherlands Yes No Yes Yes No No No

Turkey Yes Preferably No Yes Yes No No

9

General Discussion

Discussion

150

GENERAL DISCUSSION

In this thesis, we aimed to reveal health inequalities and provide information to improve

outcomes in the European paediatric RRT population. In this chapter, we will discuss the main

findings and their implications for health policy and clinical practice.

MAIN RESULTS

Considerable disparities exist in RRT incidence rates and mortality risk between

European countries. The lowest incidence rates and highest mortality risk were found

in several Eastern and Central European countries, whereas the opposite was evident

in most Western and Northern European countries.

Country differences regarding the genetic susceptibility to certain renal diseases

played only a marginal role in explaining the differences in RRT incidence and survival

between countries.

Disparities in RRT incidence and mortality risk between countries were largely

explained by differences in country macroeconomics, which limit the availability and

quality of paediatric renal care in countries burdened under financial constraints,

especially in the youngest children.

Children selected to start on HD had an increased mortality risk compared with

those on PD, especially during the first year of dialysis, and when seen by a

nephrologist for a shorter time prior to dialysis. This treatment effect was less

pronounced in patients under 5 years of age at dialysis initiation and in the infant

dialysis population.

Grafts from older living donors provided excellent graft survival across all recipient

age groups.

The youngest deceased donors showed the poorest graft survival in the youngest

recipients, whereas deceased donor age had little effect on graft failure in adolescent

recipients.

Discussion

151

European disparities in the paediatric RRT population

As defined by the WHO, the term ‘health disparity’ may best be described as “the unfair and

avoidable differences in health status seen within and between countries”, and has been

recognized as a key area for improvement by all WHO member states [238]. Correspondingly,

two of the main aims of the European health policy framework (Health 2020) are to 1)

significantly reduce health inequalities and 2) ensure health systems that are universal,

equitable, and of high quality [14]. While the concept of health disparities may be viewed from

various angles, such as race, gender, or socioeconomic status, in this thesis we focus on

geographical health disparities. The first step necessary to achieve equitable health across

Europe is to measure the magnitude of existing disparity.

We reveal considerable disparities in the provision and quality of paediatric renal care across

Europe. RRT incidence varied from 0.0 cases per million children (pmc) in Malta to 9.8 pmc in

the UK (IQR 1.8-7.7), and mortality rates varied from 0.0 deaths per 1000 patient years in

Iceland to 81.9 in Bosnia and Herzegovina (IQR 6.5-16.1). Although random variation may

partly explain these differences, especially due to the inclusion of several smaller countries

with a limited number of patients, a clear geographical pattern was evident. Compared to

Western, Northern, and Southern European regions, where the RRT incidence rate was

between 7-8 cases pmc, the Eastern European region treats a relatively low number of

children, with an incidence rate of 3.6 pmc. Similarly, most of the variation we demonstrate in

country mortality rates across Europe was attributable to an excess mortality risk in several

Eastern European countries, whereas mortality risk was mostly similar in patients treated in

other regions. Having defined the inequities in chapters 3 and 4, the next step in achieving

equitable health was to explore potential factors explaining these disparities.

Disease distribution explains little of the variation in country RRT incidence

and mortality rates

Variation in rates of paediatric RRT across Europe may be caused by geographical differences

in disease occurrence. This has been previously demonstrated in the adult RRT population,

where differences in the occurrence of diabetes and hypertension in the general population,

the two main causes of ESRD in adults, explained 79% of the variation in incidence between

Germany, England, and Wales [239]. In addition, national studies in France and Denmark

Discussion

152

associated regional variation in incidence of RRT in adults with the prevalence of diabetes

[240, 241]. Globally, the EVEREST study found an association between general population

diabetes prevalence and RRT incidence for diabetic ESRD, although surprisingly this was not

the case for overall RRT incidence [19]. Furthermore, as chronic kidney disease has a genetic

component, country variation in adult RRT incidence has also been linked to genetic variation

[242, 243].

In children, international differences in the disease occurrence of rare disorders with a genetic

component, such as childhood cancers, have been linked to geographical variations in genetic

susceptibility [244, 245]. As nearly all cases of paediatric onset ESRD consist of rare disorders

with at least some genetic origin, differences in treatment rates across Europe could be

explained by geographical differences in disease occurrence. However, in chapter 3, we

demonstrate that relative differences in disease distribution only play a marginal role in

explaining the variation in RRT incidence across European regions. A notable exception was

the higher incidence of hereditary nephropathies in Northern Europe, which was likely due to

the higher incidence of Finnish type nephropathy common in Scandinavian countries, and

explained 8% of the higher RRT incidence found in Northern Europe relative to the rest of

Europe.

Although the relative distribution of renal disease was similar across Europe, country

differences in the ability to successfully treat various diseases may still contribute to variation

in country mortality risk. However, in chapter 4, we demonstrate that this factor only

increased the variation in mortality risk between countries by 8%, suggesting that countries are

more or less equally capable of treating varying renal diseases. In line with our results, Hölttä

et al. demonstrated that patients suffering from congenital nephrotic syndrome of the Finnish

type treated in Finland, where nephrologists have ample experience in treating this disease,

had similar survival probabilities compared to RRT patients with Finnish type treated in other

European countries [246]. Subsequently, having eliminated relative differences in disease

occurrence as a major cause for geographical variation in incidence and mortality risk, we then

focused on country-level factors affecting access-to-care and quality of care.

Discussion

153

Macroeconomics

It has been well established that economic welfare is an important determinant of population

health and access to health services. Health care expenditure is determined by the volume and

cost of health care in a country, with wealthier countries tending to spend more on health

care [247]. Health care expenses have grown steadily over the past decades, driven primarily

by medical advances in technology [248, 249], and correlate well with health spending on

ESRD patients [250, 251]. In the adult RRT population, the EVEREST study demonstrated that

a higher country wealth and health care expenditure were associated with a higher RRT

incidence [19]. Similarly, Schaefer et al. previously demonstrated that country mortality rates

in were strongly affected by gross national income in countries across the globe in the

paediatric PD population [23], and Harambat et al. revealed that disparities in transplantation

rates across Europe were related to economic differences [42].

In the European paediatric RRT population, we found higher treatment rates in wealthier

countries, which tend to spend more on health care and where patients bear less out-of-

pocket health expenditures. The association between these macroeconomic indicators and the

provision of care is understandable given the complexity and cost involved in the provision of

renal care to children by a multi-professional paediatric team in an academic setting.

Furthermore, in countries with limited funds available for health care spending, the health

agenda may prioritize resources towards the more dominant high-burden diseases, thus

allocating less funds towards the expensive treatment of rare diseases [81]. Nonetheless, these

results indicate that the need for paediatric RRT is not being met by governments burdened

under financial constraints, and is a cause for concern, as non-acceptance to RRT implies

mortality. Encouragingly, we identified a ceiling effect in countries spending >7.5% of GDP on

healthcare, suggesting that RRT for all children with ESRD is attainable with healthcare

spending around this margin. A similar ceiling effect was identified by a WHO study in the

adult RRT population, where country wealth above a GDP per capita of $20 000 per year had

little effect on RRT incidence [25].

Similarly, the majority of variation in country mortality rates across Europe was explained by

differences in country public health financing, with restricted public health expenditure

adversely affecting mortality risk in our population. Importantly, this implies that in countries

Discussion

154

with limited spending on health services, children in need of RRT are not only dying due to

limited access to treatment, but also as a result of substandard care. Interestingly, the opposite

was found in the adult dialysis population, where greater country wealth was associated with

an increased mortality risk, likely reflecting the acceptance of older and more frail patients in

countries with sufficient resources [20]. This implies that the allocation of health care funds to

paediatric RRT, and a subsequent higher acceptance of younger high-risk patients, is more

effective in terms of patient survival compared to more financial resources allocated to high-

risk patients in adult RRT.

The youngest patients

Despite the substantial improvements made over the past decades regarding treatment of the

youngest patients [34, 68], the provision of RRT to these patients remains technically

challenging due to small body size, higher risk of infection, difficulties in nutrition and growth,

and a high incidence of severe comorbidities [116, 117]. Although the improvements in

survival over the past decades have been the greatest in these young patients, mortality risk

remains the highest of all paediatric patient age groups. We show that the disparities regarding

both country RRT incidence rates and mortality risk were particularly evident in this age

category. Importantly, we demonstrate that differences in country macroeconomics

disproportionately affect access to treatment for the youngest patients, finding that wealthier

countries, spending more on health care, were accepting patients for treatment at a younger

age. This was not unexpected given that the youngest patients are the most complex and

costly to treat, but implies that poorer European countries are lacking the capacity to treat

these vulnerable patients [12].

Counterintuitively, we also found the highest survival rates in countries spending the most on

health care, despite the higher acceptance of complex young patients whom bear the highest

mortality risk. Vice versa, in countries burdened under financial constraints, where access-to-

care was limited for the youngest patients, we found the worst survival rates. As non-

acceptance to RRT implies an underestimation of ESRD mortality, the disparities in mortality

caused by differences in country macroeconomics are amplified in countries with limited

resources. Furthermore, initially contradicting our hypothesis, we found no association

between country GDP per capita and mortality on RRT. However, after adjustment for patient

Discussion

155

age distribution as a mediator, an inverse relationship became statistically significant, suggesting

that, in countries with limited resources, the lower acceptance of young, high-risk, patients

was masking the inverse association between GDP per capita and mortality.

As an expected unacceptable quality of life forms an important factor in the decision to

withhold or withdraw treatment [117–120], country differences in the physicians’ willingness

to treat the youngest patients at the limit of viability may also play a role in explaining the

variation in incidence rates [252]. Indeed, the acceptance of these patients is not self-evident,

as a recent survey indicated that 70% of paediatric nephrologists sometimes refuse RRT to

children under 1 month of age, and 50% to children between 1-12 months of age [205]. As this

survey was conducted in affluent countries (Canada, Germany, United States, Japan, and the

UK), one may speculate that countries with limited (financial) capacity may further pre-dispose

physicians towards an even higher non-acceptance of these complex and costly patients,

compared to the wealthier countries included in the survey.

Country indicators of access to care and quality of care

Although we found clear associations between macroeconomic indicators and the provision

and outcomes of paediatric RRT, the exact mechanisms explaining these relationships remain

unclear. In adults, factors such as the availability of renal services, timing of dialysis initiation,

travel time to dialysis facilities, and renal service organizational indicators all play a role in

explaining the variation of RRT incidence [18, 19, 25, 239, 253–255]. Therefore, we studied

the relationship between other country indicators of access to care and quality of care, hoping

to find actionable indicators and evidence that may explain the causal pathway downstream

from macroeconomics.

We studied the effect of physical access-to-care indicators on RRT incidence, finding higher

incidence rates in more urbanized countries and in countries with a high paediatric population

density, suggesting that health services are physically more accessible in countries where

patients may expect shorter average travel times and lower costs compared with more rural

countries with a low paediatric population density. Similarly, in the UK, on a regional level of

analysis, it has been shown that adults who live in (rural) areas that are further from a dialysis

unit, have a lower chance of starting RRT [18, 256]. Conversely, we found no association with

Discussion

156

the density of paediatric RRT centres and RRT incidence. This may be explained by the

improved availability of paediatric renal care throughout Europe over the past years [68]. In

1998, 90% of European countries were able to provide paediatric dialysis and 55% paediatric

transplantation, with only 30% of Eastern European countries able to provide the latter [81]. In

our current survey, all European countries were able to provide at least one modality of

paediatric dialysis, which may explain the lack of association between the density of RRT

centres and incidence rates.

We studied the effect of country child mortality rates, as a proxy for the effectiveness and

accessibility of paediatric health services in general. We found an inverse trend between

neonatal mortality and RRT incidence in the youngest patients, independent of country wealth,

suggesting that the effectiveness of a country’s paediatric health care system is associated with

access to RRT in the youngest patients. Moreover, a countries’ neonatal mortality rate was

inversely associated with mortality on RRT and explained a large portion of the variation in

country RRT mortality risk, reflecting the impact of the quality of paediatric health systems on

the effectiveness of paediatric RRT care. However, the association between neonatal mortality

and both RRT incidence and mortality was attenuated after adjustment for macroeconomic

indicators, reflecting how the quality and accessibility of paediatric renal care services are – to

some extent – reliant on country wealth and public health expenditure. Similarly, we observed

attenuating effects of macroeconomic indicators on the relationship between mortality risk

and RRT incidence, transplantation rate, and the proportion of pre-emptive transplantations.

Closing the East-West health gap

We reveal that the variation in access to paediatric RRT and mortality rates was limited across

Western, Northern, and Southern European countries, and mainly attributable to Eastern

European countries, where patients had a significantly lower access to RRT and higher

mortality risk compared to the remainder of Europe. Despite the commitments and efforts

made by all European countries, our results demonstrate how disparities regarding the

accessibility and quality of paediatric renal care have yet to be eliminated across Europe.

Unfortunately, these results may also exemplify disparities in the quality and provision of

specialized care for other complex and costly paediatric disorders across Europe. Since the fall

of communism, many Eastern European countries have undergone dramatic changes in health

Discussion

157

care infrastructure and financing, and have achieved substantial progress regarding the

availability and effectiveness of renal services [81, 101, 102, 257]. While the gap between

Western and Eastern Europe has narrowed progressively over the past decades, many

countries in Eastern Europe remain burdened under limited health care budgets, with public

health care expenditure ranging from 2-3% (as a percentage of GDP) in the lowest ranking

countries of Europe (Albania, Cyprus, Ukraine, Montenegro, and Russia), to 9-10% in the

highest ranking countries (Sweden, The Netherlands, Denmark, France, and Austria) [75].

Financial support may be available through the EU cohesion policy, which primarily aims to

reduce social and economic disparities across Europe through project investments by various

EU funding bodies. Investments in health care infrastructure may help reduce these disparities,

although only countries with a GDP < 90% of the EU27 average are eligible. Furthermore,

after their accession to the EU, many Eastern and Central European countries experienced

‘brain drain’; an outflow of health professionals to higher-income countries [103–105]. This

may potentially cause larger problems in the future, given the inverse association found

between RRT mortality and the number of paediatric nephrologists working in a country per

million children.

In efforts to close the East-West health gap, national and European policy-makers have

pursued a uniform and high quality level of care for the prevention and treatment of rare

diseases across Europe. However, to date, evidence has been lacking regarding European

disparities specific to the field of paediatric onset ESRD. By revealing the magnitude of health-

care inequalities across Europe, we hope to increase the awareness in the paediatric

nephrology community and amongst policy makers on the European, national, and regional

levels, and provide the necessary evidence required to advocate policy change and inform

budgetary decisions on various levels of government. Currently, within various national and

European policy frameworks, many countries have adopted national strategic plans to improve

(the organization of) care for patients suffering from rare diseases. However, considering the

austerity-driven cuts in healthcare budgets experienced by most European countries over the

past few years as a result of the financial crisis, implementing these plans poses a challenge for

health care policy makers, and many plans have limited funding or no funding at all [258]. In

further consolidation efforts, over the past years, the European Committee of Experts on

Rare Disease has been instrumental in establishing Centres of Expertise for Rare Diseases.

Discussion

158

These centres have been successful in improving the quality of care for these patients,

however, the organization of centres currently varies considerably between countries. Linking

existing centres of expertise throughout Europe via the European Reference Networks should

help facilitate the sharing of expertise on rare diseases between European health providers,

and expand the opportunities for cross-border diagnostics and treatment [259]. From a clinical

perspective, we advocate further standardization of treatment guidelines and medical training

for paediatric nephrologists across Europe. Examples include recommendations for the

training of paediatric nephrologists (formulated by the European Society for Paediatric

Nephrology), information exchange through international fellowships, and the provision of

Continuing Medical Education courses across Europe [106].

The bigger picture

Internationally, RRT incidence rates in other developed countries with a publicly funded

paediatric RRT programme (and thus ‘universal access’ to treatment), such as Japan and

Malaysia and Australia and New Zealand, are similar to the average European RRT incidence of

5.4 per million children. The USA forms an exception, where the incidence rate is

approximately double that of Europe, potentially as a result of differences in race and SES, and

an earlier start on RRT [61–63]. In developing countries however, the international disparities

concerning incidence and mortality rates of paediatric RRT are far greater [168–170, 260].

Liyanage et al. estimated that in 2010, at least half of the 4.9 million people requiring RRT

worldwide died prematurely because they did not have access to treatment [3]. Specifically in

children, it has been suggested that possibly no more than 10% of those requiring RRT have

access to treatment, and that most of these preventable deaths occurred in low- and middle-

income countries [170]. The few studies available in lower-income countries, where renal

registries are often lacking, confirm these disparities [171–174, 261]. Due to high cost of

treatment for relatively few individuals, many developing countries prioritize resources

towards improvements in health infrastructure, water and sanitation, and the prevention and

treatment of high-burden (infectious) diseases. As universal access to costly RRT is unrealistic

in the short term in these countries, the largest gains in survival are likely to be made by

delaying progression of CKD, for example through aggressive antihypertensive treatment, and

thus preventing ESRD [168, 175, 262].

Discussion

159

Survival on dialysis

Despite the improvements made in paediatric renal care over the past decades, mortality in

the paediatric RRT population is still between 30 and 150 times higher than that of healthy

peers and occurs predominantly in the dialysis population [8, 64, 107]. Presently, the 5-year

survival for paediatric RRT patients is approximately 94% across Europe. Patient survival is

multifactorial, dependent on various patient and treatment characteristics, such as treatment

modality. Although renal transplantation is the preferred treatment modality in terms of

outcomes, approximately 80% of patients will initiate RRT on dialysis, to bridge the

preparation time needed for transplantation, or will require dialysis after graft loss. Survival

comparisons by dialysis modality in a randomized clinical trial (RCT) setting have proved

extremely difficult. In a single RCT performed in adults, inclusion was stopped prematurely,

largely due to randomization issues regarding patient preference for initial dialysis modality

[147]. Consequently, survival comparisons have been reliant on observational studies. In

chapter 6, we demonstrate that children selected to start dialysis on HD had an increased

mortality risk compared to PD in a propensity-score matched cohort. Importantly, this risk

varied by time on dialysis and in various patient sub-groups.

Consistent with several studies performed in the adult dialysis population, the mortality risk

difference between HD and PD was not constant over time [151, 188, 189]. During the first 1

to 2 years on dialysis, we identified a survival advantage for paediatric patients selected to

initiate dialysis on PD, after which the mortality risk became similar to those selected to start

on HD. In the adult population, this initial survival advantage has previously been attributed to

an improved preservation of residual renal function in PD patients [193], and forms the

primary rationale for the ‘PD first’ strategy, which recommends that PD should be offered as

initial modality when feasible [263]. Other potential advantages of PD in adults as initial

modality include lower costs, home treatment, and a higher patient-reported quality of life [64,

264, 265].

To further explore the dialysis modality treatment effect, we stratified by time under

treatment of a nephrologist prior to dialysis as a proxy for the timeliness of referral and the

speed of disease progression. It is known that paediatric patients who are referred late have a

poorer clinical and biochemical status and a reduced rate of pre-emptive transplantation

Discussion

160

compared with those referred earlier [194–196], whereas the speed of renal disease

progression is largely dependent on the underlying renal disease and chronic kidney disease

management [65, 197]. However, to date, it was unknown whether these factors modified the

effect of dialysis modality selection on mortality. We demonstrate that in patients under

treatment by a nephrologist for a relatively short time prior to dialysis, those selected to start

on HD had an increased mortality risk, whereas modality choice had no influence on the

survival outcome in patients with sufficient time available for pre-dialysis care. This implies that

in case of late referral, and when no contra-indications are present, patients should ideally be

started on PD. Nonetheless, although we adjusted for some of the important confounders that

may influence disease severity, such as primary renal disease and age at dialysis initiation, and

excluded “crash” patients that may be limited to HD, this effect may still be the result of

indication bias due to unmeasured case-mix confounders, such as the presence of severe

comorbidities. Even so, improving policies and formulating interventions regarding the

acceptance and preparation of dialysis are necessary to reduce the proportion of late referrals

in our population.

We also identified an age-dependent treatment effect, finding that initiating dialysis on PD was

beneficial in children over 5 years, but less pronounced in children under 5 years. This is

supported by chapter 7, where – in sharp contrast to general belief – we demonstrate the

absence of a treatment effect specifically in the infant dialysis population. Importantly, dialysis

modality choice in infants was not associated with access to transplantation, suggesting that

both modalities should be viewed as equally viable options. The lack of effect in younger

patients may be explained by age itself, as patient age is a strong determinant of survival, the

absence of treatment effect in the youngest patient may be caused by the overriding adverse

effect of young patient age on mortality. Conversely, the opposite was observed in the US,

where Mitsnefes et al. observed a decreased mortality risk in children younger than 5 years

initiating PD compared to HD, but not in older children [34]. A different distribution of race in

the USA (more blacks with a higher frequency of FSGS), and differences in clinical practice may

explain this opposite effect compared to Europe. Nonetheless, in Europe, our results imply

that older paediatric patients should ideally be started on PD where feasible.

Discussion

161

Interestingly, in the wealthiest European countries, we found no significant difference in

mortality risk between initial dialysis modality, whereas in the less wealthy countries, patients

starting RRT on HD had a significantly worse survival compared to those starting on PD. This

suggests that the majority of excess mortality found in poorer countries occurs predominantly

in the HD population. This may be due to either a poorer performance on HD in these

countries, or that these patients are sicker at treatment initiation and are therefore started on

HD. Furthermore, given that the increased mortality on HD occurs mostly during the initial

period of dialysis, early mortality is likely a main contributor to differences in country

mortality risk. Similarly, in the adult population, Robinson et al demonstrated that this early,

high-risk, period on haemodialysis is responsible for a great deal of the variation found in

mortality between countries globally [21]. Consequently, early mortality could be an effective

target for intervention to reduce mortality disparities between countries.

Graft survival and deceased donor allocation policy

It has been well established that renal transplantation offers better patient survival, cognitive

development, quality of life, and growth compared to dialysis [8]. Fortunately, three-quarters

of children with ESRD receive a transplant within 4 years after RRT initiation, however,

approximately a quarter of these patients will lose their graft within 10 years after

transplantation, necessitating dialysis. In chapter 8, we demonstrated how patient and donor

related factors influence graft failure risk in our population. Firstly, receiving a graft from a

living donor should always be preferred over that of a deceased donor, as even grafts from

carefully selected older donors such as grandparents offer excellent graft survival probabilities

[236, 237]. Although living donation is generally preferred above deceased donation from a

clinical perspective, European countries vary the proportion of living donors, the reasons for

which certainly merit further study [42]. Secondly, recipient age modifies graft failure risk,

finding the poorest graft survival in the youngest and adolescent recipients. Poor graft survival

in the former has been attributed to the surgical difficulties of transplantation in the smallest

children, and is reflected by the increased risk during the first months post-transplantation

[222, 223], and the latter due to, amongst others, poor compliance to immunosuppression

regimens during adolescence [219–221]. Thirdly, in the absence of a suitable living donor, we

found that the age of the deceased donor influenced graft survival, with the highest risk of graft

failure in the youngest and oldest deceased donors. Lastly, we examined the impact of

Discussion

162

deceased donor-recipient age combinations on graft survival. We demonstrated that in the

youngest recipients, those receiving the youngest donor kidneys have the higher risk of graft

failure, whereas in adolescents, donor age seemed less important than adolescence itself. The

deceased donor-recipient matching process is often based on a composite points-based system

involving, amongst others, factors such as waiting time, HLA and blood group matching,

percentage of panel-reactive antibodies, distance between donor and recipient, and medical

urgency. Donor and recipient age are also often integrated in allocation schemes, and many

countries have adopted a ‘young-for-young’ allocation policy. Such policies allocate paediatric

donors preferentially to paediatric recipients, with the aim of reducing waiting times and

providing high-quality grafts to children, but vary across Europe in their definitions of ‘young’

[42]. Due to the non-linear effects we found regarding both recipient and deceased donor age,

these young-for-young polices are lacking in consistency. We therefore recommend fine-

tuning these allocation schemes by increasing the granularity of the definitions of ‘young’.

Specifically, in order to maximize the graft life of donor organs, we recommend prioritizing the

allocation of deceased donors over the age of 11 to younger recipients and a cautionary use of

small paediatric donors. In adolescents, we found poor graft survival rates across the entire

deceased donor age range, precluding any specific recommendations.

METHODOLOGICAL ISSUES

Variables included in the ESPN/ERA-EDTA Registry are collected from various national

sources, and therefore data may not be homogeneous. For instance, clinical measurements

may be somewhat heterogeneous between countries and centres due to differing

measurement techniques and laboratories. Conversely, core registry data such as patient

gender, treatment modality, and various events are collected in a highly standardized manner,

reducing this type of bias to a minimum. In addition, the collection of uniform data on various

country renal service indicators proved difficult due to the diversity of health landscapes

across Europe. For instance, the exact definition of a paediatric centre may have varied by

country, introducing a certain degree of measurement error to our survey results. On the

other hand, other country level variables used in these analyses were collected in a highly

standardized manner by umbrella organizations such as the World Bank for the purpose of

country comparisons.

Discussion

163

Missing data exist in various forms; missing completely at random, missing at random, and

missing not at random. If patients with missing data differ from those with complete data (i.e.

missing not at random), particularly regarding exposure and outcome variables, estimates will

be biased, especially given that we limited our analysis to complete cases only [266].

Furthermore, as not all potentially confounding variables were collected by the registry, and as

data collection for some variables was incomplete, missing data may have led to some degree

of residual confounding, and prevents us from inferring causality.

Differences in country registry practice may have introduced missing data bias to our country

comparison studies. This may have influenced our results regarding country mortality risk and

incidence rates. Moreover, in some countries, children may be sent abroad for treatment. For

instance, paediatric patients in Iceland are sent to Scandinavian countries for transplantation.

Especially in some smaller Eastern European countries with limited facilities, the more complex

children may be sent abroad for RRT. Therefore, dependent on how national registries

approach the registration of cross-border care, this may also have affected the reliability of

our incidence and mortality risk estimates.

Dialysis modality choice in children depends on many factors, such as physician and family

preference, the type and severity of comorbidities present, the presence of various

contraindications, malnutrition, hypertension, and various metabolic factors. As these case-mix

variables are not all collected by the registry in all countries, and as sicker and more complex

patients may be selected to start on HD as initial modality choice, this may have introduced

selection bias to our results, and prevents us from inferring causality between initial dialysis

modality and mortality risk.

Germany and Italy collect data on either transplant or dialysis patients. Therefore, including

these countries would have introduced bias to our results; overall RRT incidence in these

countries would have been underestimated, and as transplant patients have better survival

rates compared to dialysis patients, our survival rates will have been skewed. It was therefore

necessary to exclude Germany and Italy, which reduced the statistical power of our analyses

and limited our ability to extrapolate our results across the whole European continent.

Discussion

164

In some countries adolescents are lost to follow-up when transferred to adult care, whereas in

other countries they are followed-up. This limits our ability to produce long-term outcome

data for this age group. Moreover, as some adolescents may have initiated RRT in adult

centres, this may have led to underestimation of RRT incidence. To make matters even more

complex, the age of transition varies between countries and patients and often more complex

high-risk patients, for example, those with severe comorbidities or cognitive disorders may be

kept longer in paediatric care.

Lastly, in our survey, the explanatory factors collected reflect the situation in each country as

of 2013, whilst the incidence rate was calculated over the period of 2007–2011. As some of

the collected indicators may vary over time, this may therefore have influenced the accuracy

of our results regarding associations between incidence rates and collected indicators.

RECOMMENDATIONS FOR FURTHER RESEARCH

We demonstrate that public health expenditure forms the main determinant of quality and

access to RRT. Nonetheless, more research is needed to understand the exact mechanisms

through which macroeconomics affect access and quality of care in countries with limited

public health funding. This is problematic to approach through ecological study designs, and

calls for a more country-specific, perhaps qualitative or mixed-methods, approach. Only after

understanding the causal pathway on the country-level, can we explore potential strategies to

help reduce these disparities.

Similarly, we still do not understand which mechanisms explain the associations between

country neonatal mortality and RRT mortality and incidence rates. Neonatal mortality is used

as a proxy for the quality of country paediatric, obstetric, and perinatal care, and therefore

comprises various elements of health care. The neonatal mortality rate may be dependent on

the organization of specialized paediatric care, the level of education for paediatricians and

neonatologists, and population-related factors such as maternal health behaviours and the

intrinsic health status of children. Moreover, the ethical decision to treat neonates at the limit

of viability may also differ between countries, which will likely also effect the neonatal

mortality rate. Consequently, it is challenging to pinpoint which components of neonatal

Discussion

165

mortality are responsible for the associations with access to RRT and quality of care in our

population, and certainly warrants further investigation.

We demonstrate that patients selected to start dialysis on PD have better survival

probabilities compared to those selected to start on HD. Nonetheless, the proportion of

children starting dialysis on PD varies widely across Europe (IQR 32%-67%). Although

differences in age distribution (the youngest are preferentially started on PD) and the

competing risk of transplantation may in part explain this variation, further study is necessary

to explore the potential reasons underlying these differences, especially given the benefits of

the PD first strategy.

We demonstrate a dialysis modality treatment effect predominantly in poorer European

countries, and less so in affluent European countries. It remains unknown whether this is due

to either a poorer performance on HD in poorer countries, or that these patients are sicker

at treatment initiation and are therefore started on HD. Further research is required to

determine explanatory factors and identify opportunities to intervene. As the majority of

excess mortality occurs in these poorer countries on HD, strategies to improve survival in this

population would contribute substantially to eliminating disparities in mortality risk across

Europe.

More research is required to determine how specialized paediatric centres (for the treatment

of rare diseases) should be organized within each country and across Europe, and how this

would impact access to RRT and quality of RRT. Theoretically, the centralization of centres

would force a larger volume of patients per centre and enable paediatric nephrologists to gain

more experience, which may subsequently improve the quality of care. This may be especially

effective regarding the transplantation of the youngest patients, which requires a highly

specialized team and facilities. On the other hand, centralization of health services for rare

disease may limit access to care, and is currently topic of debate [267].

We demonstrate the contemporary effect of macroeconomics on RRT incidence and mortality

rates. However, it remains unknown how changes in country economics over time, such as

Discussion

166

caused by the economic crisis, impact the quality of and access to paediatric renal care across

Europe.

We used the time under treatment of a nephrologist as a proxy for timely referral and the

speed of disease progression. Further research is required to disentangle these two aspects

and their modifying effects on the relationship between dialysis modality choice and mortality.

As access type modifies the relationship between dialysis modality choice and mortality, is

related to the timeliness of referral, and differs between countries, this factor should also be

taken into account when studying these effects [190, 268].

The primary outcome in our longitudinal studies included in this thesis is 5 year survival.

Although the majority of deaths occur during the initial period of RRT, long-term survival data

is still required as ESRD is a life-long condition. In contrast to adult patients, virtually all

children with ESRD are considered transplantable. Consequently, long-term dialysis data are

generally scarce and subject to negative selection of non-transplantable patients. Taking the

latter in to account, further data collection and research are required to provide nephrologists

with the evidence necessary to answer patient and family questions regarding long-term

outcomes.

Summary

Samenvatting

Summary

168

SUMMARY

In children, end-stage renal disease is a very rare condition affecting approximately 5 per

million children every year in Europe. It requires life-long renal replacement therapy, consisting

of either dialysis or renal transplantation to sustain life. Due to the rarity of paediatric end-

stage renal disease, the statistical power to perform epidemiological research in this

population has been lacking. The establishment of the European-wide registry for paediatric

renal replacement therapy (ESPN/ERA-EDTA Registry) has helped alleviate this issue, and

currently collects patient data annually from 36 European countries. The studies included in

this thesis are based on data from this Registry .

In this thesis, we aimed to reveal health inequalities and improve outcomes in the European

paediatric RRT population. Despite commitments and progress made by European Union

Member States towards reducing health inequalities, we demonstrate that geographical

disparities regarding the quality and provision of RRT to children have yet to be eliminated

across Europe. Most of these disparities were attributable to an excess mortality risk and low

incidence of paediatric RRT in several Eastern European countries. Country differences in

their ability to accept and successfully treat the youngest patients, whom are the most

complex and costly to treat, formed an important source of disparity within Europe.

Geographical differences regarding the genetic susceptibility to certain renal diseases played

only a marginal role in explaining the differences in RRT incidence and survival between

countries, whereas macroeconomic indicators, in particular public health expenditure, strongly

influenced both the quality of - and access to - paediatric RRT. Importantly, this implies that in

countries with limited spending on health services, children in need of RRT are not only dying

due to limited access to treatment, but also as a result of substandard care. Considering the

austerity-driven cuts in healthcare budgets experienced by most European countries over the

past few years, our results pose a challenge for health care policy makers in their aim to

ensure universal and equal access to high-quality healthcare across Europe. Nonetheless, by

revealing the magnitude of health-care inequalities in our population across Europe, we hope

to increase the awareness amongst policy makers and in the paediatric nephrology community,

and provide the evidence necessary to advocate policy change regarding resource allocation

and clinical practice.

Summary

169

Patient survival is multifactorial, dependent on various patient and treatment characteristics,

one of which is dialysis modality selection. We identified an initial survival advantage for those

selected to initiate dialysis on PD, especially in children with a limited time under treatment of

a nephrologist prior to dialysis, and in children over 5 year years of age, implying that when no

contra-indications are present, these patients should ideally be started on PD. Specifically in

infants, dialysis modality choice was not associated with mortality, nor with access to

transplantation, suggesting that both modalities should be viewed as equally viable options in

this population. These differential treatment effects in the paediatric dialysis population

highlight the importance of focused clinical management in these subgroups.

Given the superior patient outcomes of renal transplantation compared to dialysis, it is

fortunate that three-quarters of children with ESRD receive a transplant within 4 years after

RRT initiation. However, approximately a quarter of these patients will lose their graft within

10 years after transplantation. A living donor should always be preferred over a deceased

donor, as we demonstrate that even grafts from carefully selected older donors offer excellent

graft survival probabilities. When a living donor is unavailable, the differential graft failure risk

of deceased donor age and recipient age should be taken into account during the allocation

process. Specifically, transplantation of the youngest deceased donors should be avoided in the

youngest recipients, as graft failure rates in this group were especially high, particularly directly

post-transplantation. In adolescents however, deceased donor age had little effect, likely owing

to the overriding adverse effect of adolescence on graft survival probability. Currently, both

recipient and deceased donor age definitions used by various donor allocation policies across

Europe are heterogeneous, and require reappraisal and standardization taking these results

into account.

Summary

170

SAMENVATTING

Wanneer de nieren niet of nauwelijks meer functioneren, spreken we van eindstadium

nierfalen. Zonder levenslange behandeling met nierfunctie vervangende therapie (dialyse of

transplantatie), is eindstadium nierfalen fataal. Gelukkig is deze aandoening in kinderen zeer

zeldzaam, met een incidentie van ongeveer vijf op de miljoen per jaar in Europa. Door de

kleine aantallen patiënten ontbreekt vaak de statistische power om epidemiologisch onderzoek

uit te voeren in deze populatie. Om dit te verhelpen is een Europese registratie (ESPN/ERA-

EDTA Registry) opgericht voor alle kinderen die behandeld worden met nierfunctie

vervangende therapie. De registratie verzamelt momenteel data uit 36 Europese landen en

vormt de basis van deze proefschrift.

Dit proefschrift heeft als doel de Europese verschillen in de incidentie en uitkomsten van

nierfunctie vervangende therapie bij kinderen in kaart te brengen. Ondanks recente

inspanningen van de Europese Unie om internationale verschillen in gezondheid te

verminderen, tonen wij aan dat ongelijkheid in zowel toegang tot zorg, als de kwaliteit van

zorg in deze populatie nog onaanvaardbaar groot blijkt. Het merendeel van deze verschillen

was toe te wijzen aan een hoge patiënten sterfte en een lage incidentie van nierfunctie

vervangende therapie in een aantal Oost-Europese landen. Hierin speelden de verschillen in de

capaciteit van landen om de jongste patiënten (die medisch het meest complex zijn) succesvol

in behandeling te nemen een belangrijke rol. Geografische verschillen in de relatieve distributie

van het type primaire nierziekte konden deze verschillen niet verklaren. Macro-economische

factoren daarentegen, met name overheidsuitgaven aan zorg, vormden de belangrijkste

determinanten voor zowel toegang tot, als de kwaliteit van nierfunctie vervangende therapie.

Dit impliceert dat in landen waar minder wordt uitgegeven aan zorg door de overheid,

kinderen niet alleen sterven als gevolg van een beperkte toegang tot deze levensreddende

behandeling, maar ook als gevolg van een lage kwaliteit van zorg. Vooral gezien de

bezuinigingen op de zorg als gevolg van de financiële crisis zullen deze resultaten een uitdaging

vormen voor beleidsmakers in hun streven naar een universele en eerlijke toegang tot

kwalitatief hoogwaardige zorg in Europa. Desalniettemin, door de omvang van deze verschillen

in kaart te brengen, hopen we bewustwording onder Europese/nationale beleidsmakers en

kindernefrologen te stimuleren, en kennis te verschaffen zodat deze een kader bieden voor

beleidswijzigingen zowel op klinisch gebied, als op het budgetteren van zorg.

Summary

171

Patiënt overleving in deze populatie is afhankelijk van meerdere patiënt- en

behandelingsfactoren. Eén daarvan is het selecteren van de initiële dialysemodaliteit. Wij tonen

aan dat patiënten die worden geselecteerd om met peritoneale dialyse te starten een betere

overleving hebben gedurende het eerste dialysejaar vergeleken met patiënten die starten met

hemodialyse. Dit behandeleffect was vooral evident bij kinderen die kort onder behandeling

waren van een kindernefroloog vóór dialyse, en bij kinderen ouder dan vijf jaar. Dit impliceert

dat, wanneer er geen contra-indicaties aanwezig zijn, kinderen bij voorkeur moeten starten op

peritoneale dialyse. Specifiek bij kinderen onder de één jaar, vonden wij geen associatie tussen

dialysemodaliteit en mortaliteit of transplantatiekans. De differentiële behandeleffecten bij

kinderen die starten met dialyse benadrukken het belang van gerichte klinische behandeling in

verschillende patiënt subgroepen.

Nier transplantatie bij kinderen biedt betere uitkomsten vergeleken met dialyse. In de praktijk

wordt ongeveer driekwart van alle kinderen getransplanteerd binnen 4 jaar na het starten van

nierfunctie vervangende therapie. Desondanks verliest ongeveer een kwart van deze kinderen

binnen 10 jaar hun transplantaat. Wij hebben aangetoond dat een transplantaat van een

levende donor betere overlevingskansen biedt vergeleken met een transplantaat van een

overleden donor, zelfs als de levende donor op leeftijd is. Indien een levende donor niet voor

handen is, adviseren wij tijdens het toewijzingsproces rekening te houden met het differentieel

risico op transplantaat falen dat afhankelijk is van zowel de leeftijd van de overleden donor als

de leeftijd van de patiënt. Transplantatie van de jongste overleden donors bij de jongste

patiënten geeft namelijk een hoog risico op transplantaat falen, met name direct na

transplantatie. Bij adolescente patiënten heeft de leeftijd van een overleden donor daarentegen

weinig effect op transplantaat falen. Dit komt vermoedelijk door een overheersende negatief

effect van adolescentie op de transplantaat overlevingskans. Ten tijde van het schijven van dit

proefschrift is het toewijzingsbeleid van donor nieren in Europa zeer heterogeen, met name

wat betreft de definities van donor en patiënt leeftijdscategorieën. Gezien het interactie effect

tussen de leeftijd van patiënt en overleden donor op het risico van transplantaat falen, vereist

dit evaluatie en harmonisatie van het donor toewijzingsbeleid in Europa.

Acknowledgements

Dankwoord

Acknowledgements

174

ACKNOWLEDGEMENTS

We would like to thank the patients, their parents and the staff of all the dialysis and transplant

units who have contributed data via their national registries and contact persons. We also

would like to thank E Levtchenko, D Haffner, Z Massy, A Bjerre, R Coppo, J Harambat, P

Cochat, Z Massy, and C Stefanidis for being members of the ESPN/ERA-EDTA Registry

Committee, D Shtiza, R Kramar, R Oberbauer, S Baiko, A Sukalo, K van Hoeck, F Collart, JM

des Grottes, D Pokrajac, D Roussinov, D Batinić, M Lemac, J Slavicek, T Seeman, K Vondrak,

JG Heaf, U Toots, P Finne, C Grönhagen-Riska, C Couchoud, M Lasalle, E Sahpazova, N Abazi,

N Ristoka Bojkovska, K Rascher, E Nüsken, L Weber, G von Gersdorff, F Schaefer, B

Tönshoff, K Krupka, B Höcker, L Pape, N Afentakis, A Kapogiannis, N Printza, G Reusz, Cs

Berecki, A Szabó, T Szabó, Zs Györke, E Kis, R Palsson, V Edvardsson, R Chimenz, C

Corrado, B Minale, F Paglialonga , R Roperto, G Leozappa, E Verrina, A Jankauskiene, B

Pundziene, V Said-Conti, S Gatcan, O Berbeca, N Zaikova, S Pavićević, T Leivestad, A

Zurowska, I Zagozdzon, C Mota, M Almeida, C Afonso, G Mircescu, L Garneata, EA

Molchanova, NA Tomilina, BT Bikbov, M Kostic, A Peco-Antic, B Spasojevic-Dimitrijeva, G

Milosevski-Lomic, D Paripovic, S Puric, D Kruscic, L Podracka, G Kolvek, J Buturovic-Ponikvar,

G Novljan, N Battelino, A Alonso Melgar and the Spanish Pediatric Registry, S Schön, KG

Prütz, L Backmän, M Stendahl, M Evans, B Rippe, CE Kuenhi, E Maurer, GF Laube, S Tschumi,

P Parvex, A Hoitsma, A Hemke, and all centers participating in the RichQ study, R Topaloglu,

A Duzova, D Ivanov, R Pruthi, F Braddon, S Mannings, A Cassula, MD Sinha for contributing

data to the ESPN/ERA-EDTA Registry.

Dankwoord

175

DANKWOORD Na 4 jaar heb ik eindelijk mijn proefschrift afgerond, maar dit heb ik natuurlijk niet in mijn

eentje gedaan! Hier wil ik graag iedereen bedanken die mee heeft geholpen, in welke vorm dan

ook, bij het tot stand komen van dit proefschrift. Een aantal mensen wil ik graag in het

bijzonder noemen.

Als eerste wil ik graag mijn promotores bedanken. Kitty, ik ben jou in het bijzonder dankbaar

voor alle epidemiologische kennis die je mij de afgelopen jaren hebt bijgebracht. Ik kijk met

veel plezier terug naar de samenwerking met jou de afgelopen 4 jaar, en ben daarom blij en

trots dat je mij EQUAL toevertrouwt.

Karlijn, jouw enthousiasme voor de kindernefrologie was aanstekelijk. Na onze talloze

onderzoeks besprekingen liep ik altijd terug naar mijn bureau vol met inspiratie en verse

moed. Jouw kritische en creatieve blik hebben onze papers tot een hoger niveau getild. Samen

met Kitty vormden we een efficiënt en gezellig team en dat zal ik helaas moeten missen nu je

het AMC hebt verlaten.

Marjolein, regerende koningin van de ESPN registratie, onze samenwerking aan de jaarlijkse

datastorm liep op een gegeven moment als een geoliede machine. De samenwerking ga ik

missen, het dataploegen iets minder. Gelukkig zit je nog altijd twee deurtjes verderop. Ik denk

met plezier terug aan alle congressen, van Italië tot Brazilië, die we samen vanaf mijn eerste

dag op het AMC hebben meegemaakt.

Franz, Jerome, Jaap, your clinical knowledge was essential in bridging the gap between

interesting research ideas and clinical relevance. It was my pleasure working with the very best

paediatric nephrologists of Europe (or perhaps even the world?!). Even though I’ve now

switched to adult nephrology, I am sure we will meet again soon.

(Ex-) kamergenoten, en triple-daters, jullie bedankt voor alle grapjes, gekkigheid, borrels,

wandelingetjes naar de AH en de liters koffie die we samen hebben gedronken over de

afgelopen jaren.

Dankwoord

176

ERA team, en niet te vergeten onze guest-researcher Lidwien, bedankt voor alle gezelligheid

tijdens onze diners, borrels, en team uitjes. Het is fijn om onderdeel uit te maken van een

team dat elkaar op congressen steunt met de (presentatie) stress, en daarna de overwinning

viert met borrels en (presidentiële) etentjes. Ik hoop dat er nog vele zullen volgen!

Verder wil ik iedereen van de afdeling KIK bedanken voor de altijd sympathieke, leerzame en

constructieve sfeer.

I would also like to thank the members of the doctorate committee for taking the time to

critically review my thesis and form the opposition during my defense ceremony.

Maria, wie had dat gedacht, dat ik dit nu in het Nederlands aan jou zou schrijven. Jouw kennis

van de Nederlandse taal zal die van mij snel overstijgen. Ik heb een enorme respect voor jouw

gedrevenheid; straks promoveren met duizend papers, het Nederlandse taal meester maken,

en ook nog een master transplantatienefrologie erbij! Wat een fijne jaren hebben we achter de

rug, met als hoogtepunten Brussel, Londen, Glasgow, en niet te vergeten alle Javastraat

vrimibo’s. Heel erg bedankt voor het paranimf zijn en alle bijbehorende

verantwoordelijkheden. Ik hoop dat ik het net zo goed ga doen bij jouw promotie binnenkort!

Loes, BFF, Subba-queen, buuf, en nu ook paranimf! Bedankt voor het organiseren van het feest

en alles wat daarbij komt kijken, maar vooral voor je vriendschap, inlevingsvermogen en begrip

over de afgelopen jaren. Het heeft veel voor mij betekend!

Utrechtse en Amsterdamse vrienden, fijn dat jullie altijd bereid waren om een stress

verlagende biertje met me te drinken en voor de ontelbare avonturen in het

Amsterdamse/Utrechtse nachtleven!

Chiel, Oof, Tim, mijn ‘stage’ heb ik nu afgerond, hoor. Tegen de tijd dat jullie dit lezen ben ik

ook al ‘afgestudeerd’. Laten we snel weer een biertje drinken in jullie toekomstige woonplaats,

Amsterdam.

Dankwoord

177

Peim, autonoom kunstenaar, fotomodel, grafisch ontwerper, wat kun je niet? Ik ben heel blij

met jouw ontwerp en vond het leuk om samen het creatief proces te doorlopen. Tot de

volgende WYS photo shoot!

Dad and Sophie, thank you for always having confidence in me and for all the relaxing

weekends in Assen. Mamma, je overvloed aan liefde en pedagogiek hebben me gemaakt tot

wie ik ben en daarmee ben je dus ook deels verantwoordelijk voor het tot stand komen van

deze proefschrift! Mike, straks mag jij ook promoveren. Ik heb er alle vertrouwen in dat het in

South-Hampton helemaal goed komt en ben erg benieuwd naar je avonturen daarna!

Yara, bedankt voor het luisteren naar mijn spannende verhalen over de verschillende

regressie-technieken. Je oprechtheid, empathie en liefde hebben hun sporen achtergelaten op

mij. Op naar de volgende vakantie!

178

Curriculum Vitae

Portfolio

CV & Portfolio

180

CURRICULUM VITAE

Nicholas Christopher Chesnaye werd op 13 februari 1985 geboren te Singapore. In 2003

behaalde hij zijn VWO diploma aan het Hondsrug College in Emmen. In 2007 behaalde hij zijn

BSc Biomedische Wetenschappen aan de Universiteit van Utrecht, en in 2011 zijn MSc aan de

Vrije Universiteit te Amsterdam. Ter afronding van zijn MSc liep hij stage bij de Centers for

Disease Control and Prevention in de Vereinigde Staten, waar hij een maand veldwerk heeft

verricht in Cambodja met als doel het nationale ontwormingsprogramma voor schoolkinderen

te evalueren. Nicholas heeft vervolgens zijn onderzoeksstage gelopen bij het Intituut voor

Tropische Geneeskunde Antwerpen, om in Vietnam onderzoek te doen naar de epidemiologie

van malaria. Na zijn afstuderen is Nicholas begonnen als Clinical Research Associate Trainee

bij de clinical research organisation Quintiles in Hoofddorp. In februari 2013 startte Nicholas

met zijn promotieonderzoek binnen de ESPN/ERA-EDTA Registratie op de afdeling Klinische

Informatiekunde van het Academisch Medisch Centrum te Amsterdam. Onder leiding van Prof.

Dr. Kitty Jager en Dr. Karlijn van Stralen voerde hij het onderzoek in dit proefschrift uit. Na

het afronden van zijn promotieonderzoek start Nicholas als internationale project coördinator

van de EQUAL studie.

CV & Portfolio

181

PORTFOLIO Name PhD student: Nicholas Christopher Chesnaye PhD period: March 2013 – March 2017 Promotor: Prof. Dr. Kitty Jager

Year Workload

(ECTS) 1. PhD training General courses Clinical Epidemiology 2013 0.6 Practical Biostatistics 2015 1.1 Observational Clinical Epidemiology 2015 0.6 Advanced Topics in Biostatistics 2016 2.1 Specific courses CME: Introductory Course on Epidemiology 2013 0.6 NIHES: International Comparison of Health Care Systems 2013 1.4 NDT - A Course For Reviewers To Be 2014 0.6 NIHES: Causal Mediation Analysis 2015 0.7 eBROK 2017 1.0 Seminars, workshops, and master classes Workshop 'Nephorology Registries and Genetic Renal Diseases', Genova, Italy

2013 0.6

Seminar '“Who wrote my paper” by Dr. Drummond Rennie', AMC 2013 0.3 Symposium 'Hot topics in pediatric end-stage renal disease', AMC 2013 0.3 Workshop 'Systematic review of measuring instruments', VUMC 2013 0.3 Masterclasses 'Advances in Epidemiologic Analysis', NIHES 2015 0.3 Symposium 'New Kids on the Block; wetenschappelijk onderzoek in de nefrologie', AMC

2015 0.3

Presentations (oral) Disparities in Treatment Rates of Paediatric End-Stage Renal Disease across Europe: Insights from the ESPN/ERA-EDTA Registry, ESPN/ERA-EDTA 47th congress, Porto, Portugal

2014 0.5

Mortality Risk in European Children with End-Stage Renal Disease on Dialysis, 52nd ERA-EDTA congress, London, United Kingdom

2015 0.5

Mortality Risk in European Children with End-Stage Renal Disease on Dialysis, ESPN/ERA-EDTA 48th congress, Brussel, Belgium

2015 0.5

Mortality Risk in European Children with End-Stage Renal Disease on Dialysis, 53rd ERA-EDTA congress, Vienna, Austria

2016 0.5

The Association of Donor and Recipient Age with Graft Survival in Pediatric Renal Transplant Recipients - an ESPN/ERA-EDTA Registry Study, IPNA 17th congress, Iguacu, Brazil

2016 0.5

CV & Portfolio

182

Mortality Risk Disparities in Children with End-Stage Renal Disease across Europe - An ESPN-ERA/EDTA Registry Analysis, 54th ERA-EDTA congress, Madrid, Spain

2017 0.5

Presentations (poster) Disparities in Treatment Rates of Paediatric End-Stage Renal Disease across Europe: Insights from the ESPN/ERA-EDTA Registry, 51th ERA-EDTA congress, Amsterdam, The Netherlands

2014 0.5

Mortality Risk Disparities in Children with End-Stage Renal Disease across Europe - An ESPN-ERA/EDTA Registry Analysis, 53rd ERA-EDTA congress, Vienna, Austria

2016 0.5

Mortality Risk Disparities in Children with End-Stage Renal Disease across Europe - An ESPN-ERA/EDTA Registry Analysis, IPNA 17th congress, Iguacu, Brazil

2016 0.5

The Association of Donor and Recipient Age with Graft Survival in Pediatric Renal Transplant Recipients - an ESPN/ERA-EDTA Registry Study, 54th ERA-EDTA congress, Madrid, Spain

2017 0.5

(Inter)national conferences 50th ERA-EDTA congress, Istanbul, Turkey 2013 1.0 51th ERA-EDTA congress, Amsterdam, The Netherlands 2014 1.0 ESPN/ERA-EDTA 47th congress, Porto, Portugal 2014 1.0 52nd ERA-EDTA congress, London, United Kingdom 2015 1.0 ESPN/ERA-EDTA 48th congress, Brussel, Belgium 2015 1.0 53rd ERA-EDTA congress, Vienna, Austria 2016 1.0 Seventeenth Congress of the International Pediatric Nephrology Association, Iguacu, Brazil

2016 1.0

EURenOmics Project and General Assembly Meeting, Paris, France 2016 1.0 54th ERA-EDTA congress, Madrid, Spain 2017 1.0 Other Journal club 2013-

2017 2.0

2. Teaching Lecturing CME: Introductory Course on Epidemiology, Kopenhagen, Denmark 2016 1.5 CME: Introductory Course on Epidemiology, Nicosia, Cyprus 2017 1.5 Supervising Gulfidan Yasar. Survival in children with end-stage renal disease. 2014-

2015 1.0

183

LIST OF PUBLICATIONS Chesnaye NC, van Stralen KJ, Bonthuis M, et al. Survival in children requiring chronic renal

replacement therapy. Pediatr Nephrol 2017 May; Epub ahead of print

Chesnaye NC, van Stralen KJ, Bonthuis M, et al. The association of donor and recipient age with graft

survival in paediatric renal transplant recipients: an ESPN/ERA-EDTA Registry study. Nephrol Dial

Transplant 2017 Jul; Accepted

Chesnaye NC, Schaefer F, Bonthuis M, et al. Mortality risk disparities in children receiving chronic

renal replacement therapy for the treatment of end-stage renal disease across Europe - An ESPN-

ERA/EDTA Registry analysis. The Lancet 2017 May; 389(10084): 2128-2137

Vidal E, van Stralen KJ, Chesnaye NC, et al (2016) Infants Requiring Maintenance Dialysis: Outcomes

of Hemodialysis and Peritoneal Dialysis. Am J Kidney Dis. 2017 May; 69(5): 617-625

Chesnaye NC, Schaefer F, Groothoff JW, et al. Mortality risk in European children with end-stage

renal disease on dialysis: Results from the ESPN/ERA-EDTA Registry. Kidney Int 2016 Jun; 89(6): 1355–

1362

Chesnaye NC, Schaefer F, Groothoff JW, et al. Disparities in treatment rates of paediatric end-stage

renal disease across Europe: insights from the ESPN/ERA-EDTA Registry. Nephrol Dial Transplant 2015

Aug; 30(8): 1377–1385

Hafiz I, Berhan M, Keller A, Haq R, Chesnaye NC, et al. School-based mass distributions of

mebendazole to control soil-transmitted helminthiasis in the Munshiganj and Lakshmipur districts of

Bangladesh: An evaluation of the treatment monitoring process and knowledge, attitudes, and practices of

the population. Acta Trop. 2015;141(Part B):385–90

Chesnaye NC, Bonthuis M, Schaefer F, et al. Demographics of paediatric renal replacement therapy in

Europe: a report of the ESPN/ERA-EDTA Registry. Pediatr Nephrol 2014 Dec; 29(12): 2403–2410

Chesnaye NC, Sinuon M, Socheat D, Koporc K, Mathieu E. Treatment coverage survey after a

school-based mass distribution of mebendazole: Kampot Province, Cambodia. Acta Trop. 2011

Apr;118(1):21-6

184

185

REFERENCES

1. Tjaden LA, Vogelzang J, Jager KJ, et al (2014) Long-term quality of life and social outcome of childhood end-

stage renal disease. J Pediatr 165:336–342.

2. Tonelli M (2006) Chronic Kidney Disease and Mortality Risk: A Systematic Review. J Am Soc Nephrol

17:2034–2047.

3. Liyanage T, Ninomiya T, Jha V, et al (2015) Worldwide access to treatment for end-stage kidney disease: A

systematic review. Lancet 385:1975–1982.

4. Warady BA, Chadha V (2007) Chronic kidney disease in children: the global perspective. Pediatr Nephrol

22:1999–2009.

5. Harambat J, van Stralen KJ, Kim JJ, Tizard EJ (2012) Epidemiology of chronic kidney disease in children.

Pediatr Nephrol 27:363–373.

6. Gillen DL, Stehman-Breen CO, Smith JM, et al (2008) Survival advantage of pediatric recipients of a first

kidney transplant among children awaiting kidney transplantation. Am J Transplant 8:2600–6.

7. Mendley SR, Zelko FA (1999) Improvement in specific aspects of neurocognitive performance in children

after renal transplantation. Kidney Int 56:318–23.

8. McDonald SP, Craig JC (2004) Long-term survival of children with end-stage renal disease. N Engl J Med

350:2654–2662.

9. Goldstein SL, Graham N, Burwinkle T, et al (2006) Health-related quality of life in pediatric patients with

ESRD. Pediatr Nephrol 21:846–850.

10. Van Arendonk KJ, Boyarsky BJ, Orandi BJ, et al (2014) National trends over 25 years in pediatric kidney

transplant outcomes. Pediatrics 133:594–601.

11. NAPRTCS Collaborative Studies NAPRTCS 2014 Annual Transplant Report.

12. Neuhaus TJ, Goetschel P LE (1998) Small patients—high costs: economic aspects of medical treatment of

children with early onset of chronic renal failure. Praxis (Bern 1994) 1593–1599.

13. Bunchman TE (1996) The ethics of infant dialysis. Perit Dial Int 16 Suppl 1:S505-8.

14. WHO Health 2020 policy framework and strategy.

15. Dhoul N, de Lusignan S, Dmitrieva O, et al (2012) Quality achievement and disease prevalence in primary

care predicts regional variation in renal replacement therapy (RRT) incidence: an ecological study. Nephrol

Dial Transplant 27:739–46.

16. Visser A, Noordzij M, Gansevoort RT, et al (2012) Exploration of the difference in incidence of renal

replacement therapy between Flanders and the Netherlands--investigation of explanatory variables. Nephrol

Dial Transplant 27:803–9.

17. Castledine CI, Gilg JA, Rogers C, et al (2012) How much of the regional variation in RRT incidence rates

within the UK is explained by the health needs of the general population? Nephrol Dial Transplant 27:3943–

50.

18. Judge A, Caskey FJ, Welton NJ, et al (2012) Inequalities in rates of renal replacement therapy in England:

does it matter who you are or where you live? Nephrol Dial Transplant 27:1598–607.

19. Caskey FJ, Kramer A, Elliott RF, et al (2011) Global variation in renal replacement therapy for end-stage

renal disease. Nephrol Dial Transplant 26:2604–10.

20. Kramer A, Stel VS, Caskey FJ, et al (2012) Exploring the association between macroeconomic indicators and

186

dialysis mortality. Clin J Am Soc Nephrol 7:1655–63.

21. Robinson BM, Zhang J, Morgenstern H, et al (2014) Worldwide, mortality risk is high soon after initiation of

hemodialysis. Kidney Int 85:158–65.

22. van Dijk P, Zwinderman A, Dekker F, et al (2007) Effect of general population mortality on the north-south

mortality gradient in patients on replacement therapy in Europe. Kidney Int 71:53–9.

23. Schaefer F, Borzych-Duzalka D, Azocar M, et al (2012) Impact of global economic disparities on practices

and outcomes of chronic peritoneal dialysis in children: insights from the International Pediatric Peritoneal

Dialysis Network Registry. Perit Dial Int 32:399–409.

24. Wühl E, van Stralen KJ, Wanner C, et al (2014) Renal replacement therapy for rare diseases affecting the

kidney: an analysis of the ERA-EDTA Registry. Nephrol Dial Transplant 29 Suppl 4:iv1-iv8.

25. White S (2008) How can we achieve global equity in provision of renal replacement therapy? Bull World

Health Organ 86:229–237.

26. Shroff R, Ledermann S (2009) Long-term outcome of chronic dialysis in children. Pediatr Nephrol 24:463–

74.

27. van de Luijtgaarden MWM, Noordzij M, Stel VS, et al (2011) Effects of comorbid and demographic factors

on dialysis modality choice and related patient survival in Europe. Nephrol Dial Transplant 26:2940–2947.

28. Weinhandl ED, Foley RN, Gilbertson DT, et al (2010) Propensity-matched mortality comparison of incident

hemodialysis and peritoneal dialysis patients. J Am Soc Nephrol 21:499–506.

29. Lukowsky LR, Mehrotra R, Kheifets L, et al (2013) Comparing mortality of peritoneal and hemodialysis

patients in the first 2 years of dialysis therapy: a marginal structural model analysis. Clin J Am Soc Nephrol

8:619–628.

30. Mehrotra R, Chiu Y-W, Kalantar-Zadeh K, et al (2011) Similar outcomes with hemodialysis and peritoneal

dialysis in patients with end-stage renal disease. Arch Intern Med 171:110–118.

31. Stokes JB (2012) Peritoneal dialysis is not a superior therapy to hemodialysis: A comparison. Blood Purif

33:160–164.

32. Noordzij M, Jager KJ (2014) Patient survival on dialysis in Korea: a different story? Kidney Int 86:877–880.

33. Wong CS, Hingorani S, Gillen DL, et al (2002) Hypoalbuminemia and risk of death in pediatric patients with

end-stage renal disease. Kidney Int 61:630–637.

34. Mitsnefes MM, Laskin BL, Dahhou M, et al (2013) Mortality Risk Among Children Initially Treated With

Dialysis for End-Stage Kidney Disease, 1990–2010. JAMA 309:1921–1929.

35. Lin HH, Tsai C-W, Lin PH, et al (2012) Survival analysis of pediatric dialysis patients in Taiwan. Nephrology

(Carlton) 17:621–627.

36. Foster BJ, Dahhou M, Zhang X, et al (2011) Change in mortality risk over time in young kidney transplant

recipients. Am J Transplant 11:2432–2442.

37. Cecka JM (2008) Kidney transplantation in the United States. Clin Transpl 1–18.

38. Pape L, Ehrich JHH, Offner G (2007) Young for young! Mandatory age-matched exchange of paediatric

kidneys. Pediatr Nephrol 22:477–479.

39. Harmon WE, Alexander S, Tejani A, Stablein D (1992) The effect of Donor Age on Graft Survival in

Pediatric Cadaver Renal Transplant Recipients - A Report Of The North American Pediatric Renal

Transplant Cooperative Study. Transplantation 54(2):232--7.

40. Hayes JM, Novick AC, Streem SB, et al (1988) The use of single pediatric cadaver kidneys for

187

transplantation. Transplantation 45:106–10.

41. Pape L, Offner G, Ehrich JHH, et al (2004) Renal Allograft Function in Matched Pediatric and Adult Recipient

Pairs of the Same Donor. Transplantation 77:1191–1194.

42. Harambat J, Van Stralen KJ, Schaefer F, et al (2013) Disparities in Policies , Practices and Rates of Pediatric

Kidney Transplantation in Europe. Am J Transplant 13:2066–2074.

43. Dubourg L, Cochat P, Hadj-Aïssa A, et al (2002) Better long-term functional adaptation to the child’s size

with pediatric compared to adult kidney donors. Kidney Int 62:1454–1460.

44. Pape L, Hoppe J, Becker T, et al (2006) Superior long-term graft function and better growth of grafts in

children receiving kidneys from paediatric compared with adult donors. Nephrol Dial Transplant 21:2596–

2600.

45. Bresnahan BA, McBride MA, Cherikh WS, Hariharan S (2001) Risk factors for renal allograft survival from

pediatric cadaver donors: an analysis of united network for organ sharing data. Transplantation 72:256–61.

46. Singh A, Stablein D, Tejani A (1997) Risk factors for vascular thrombosis in pediatric renal transplantation: a

special report of the North American Pediatric Renal Transplant Cooperative Study. Transplantation

63:1263–7.

47. Schurman SJ, McEnery PT (1997) Factors influencing short-term and long-term pediatric renal transplant

survival. J Pediatr 130:455–62.

48. Dharnidharka VR, Fiorina P, Harmon WE (2014) Kidney Transplantation in Children. N Engl J Med 371:549–

558.

49. Pape L, Ehrich JHH, Zivicnjak M, Offner G (2005) Growth in children after kidney transplantation with living

related donor graft or cadaveric graft. Lancet 366:151–153.

50. Tizard EJ, Verrina E, van Stralen KJ, Jager KJ (2009) Progress with the European Society for Paediatric

Nephrology (ESPN)/ERA-EDTA Registry for children with established renal failure (ERF). Nephrol Dial

Transplant 24:2615–7.

51. ESPN/ERA-EDTA (2013) http://www.espn-reg.org/.

52. Jager KJ, Zoccali C, Kramar R, Dekker FW (2007) Measuring disease occurrence. Kidney Int 72:412–5.

53. European Commission (2013) http://epp.eurostat.ec.europa.eu/.

54. ERA-EDTA Registry (2015) ERA-EDTA Registry Annual Report 2013.

55. Schwartz GJ, Muñoz A, Schneider MF, et al (2009) New equations to estimate GFR in children with CKD. J

Am Soc Nephrol 20:629–37.

56. National Kidney Foundation (2006) KDOQI Clinical Practice Guidelines and Clinical Practice

Recommendations for 2006 Updates: Hemodialysis Adequacy, Peritoneal Dialysis Adequacy and Vascular

Access. Am J Kidney Dis 48:S1–S322.

57. Noordzij M, Leffondré K, van Stralen KJ, et al (2013) When do we need competing risks methods for

survival analysis in nephrology? Nephrol Dial Transplant 1–8.

58. Pocock SJ, Clayton TC, Altman DG (2002) Survival plots of time-to-event outcomes in clinical trials: good

practice and pitfalls. 359:1686–1689.

59. van Dijk PC, Jager KJ, de Charro F, et al (2001) Renal replacement therapy in Europe: the results of a

collaborative effort by the ERA-EDTA registry and six national or regional registries. Nephrol Dial


60. European Mortality Database (2013) http://data.euro.who.int/.

188

61. The Malaysian Dialysis & Transplant Registry (2013) 20th Report of the Malaysian Dialysis & Transplant

Registry.

62. National Institutes of Health National Institute of Diabetes and Digestive and Kidney Diseases. (2012)

United States Renal Data System, USRDS 2012 Annual Data Report: Atlas of Chronic Kidney Disease and

End-Stage Renal Disease in the United States.

63. Australia and New Zealand Dialysis and Transplant Registry ANZDATA Registry Report 2012.

64. National Institutes of Health National Institute of Diabetes and Digestive and Kidney Diseases (2011) U.S.

Renal Data System, USRDS 2014 Annual Data Report: Atlas of Chronic Kidney Disease and End-Stage Renal

Disease in the United States.

65. van Stralen KJ, Tizard EJ, Jager KJ, et al (2010) Determinants of eGFR at start of renal replacement therapy

in paediatric patients. Nephrol Dial Transplant 25:3325–32.

66. Samuel SM, Tonelli MA, Foster BJ, et al (2011) Survival in pediatric dialysis and transplant patients. Clin J Am

Soc Nephrol 6:1094–1099.

67. Roberts MA, Polkinghorne KR, McDonald SP, Ierino FL (2011) Secular trends in cardiovascular mortality

rates of patients receiving dialysis compared with the general population. Am J Kidney Dis 58:64–72.

68. van der Heijden BJ, van Dijk PCW, Verrier-Jones K, et al (2004) Renal replacement therapy in children: data

from 12 registries in Europe. Pediatr Nephrol 19:213–221.

69. Vogelzang JL, van Stralen KJ, Jager KJ, Groothoff JW (2013) Trend from cardiovascular to non-cardiovascular

late mortality in patients with renal replacement therapy since childhood. NDT 28:2082–2089.

70. Groothoff JW, Gruppen MP, Offringa M, et al (2002) Mortality and causes of death of end-stage renal

disease in children: a Dutch cohort study. Kidney Int 61:621–629.

71. Wolfe I, Thompson M, Gill P, et al (2013) Health services for children in western Europe. Lancet 381:1224–

34.

72. ESRI ArcGIS Desktop 10.2. 11.

73. Ralph P, Coop G (2013) The geography of recent genetic ancestry across Europe. PLoS Biol 11:e1001555.

74. Benjamini, Y Hochberg Y (2000) On the Adaptive Control of the False Discovery Rate in Multiple Testing

With Independent Statistics. J Educ Behav Stat 25:60–83.

75. World Bank Database (2013) http://data.worldbank.org/.

76. Wiesel A, Queisser-Luft A, Clementi M, et al (2005) Prenatal detection of congenital renal malformations by

fetal ultrasonographic examination: an analysis of 709,030 births in 12 European countries. Eur J Med Genet

48:131–44.

77. Tattersall J, Dekker F, Heimbürger O, et al (2011) When to start dialysis: updated guidance following

publication of the Initiating Dialysis Early and Late (IDEAL) study. Nephrol Dial Transplant 26:2082–6.

78. Statistical Methodology and Applications Branch Surveillance Research Program National Cancer Institute.

Joinpoint Regression Program, Version 4.0.4.

79. Chesnaye N, Bonthuis M, Schaefer F, et al (2014) Demographics of paediatric renal replacement therapy in

Europe: a report of the ESPN/ERA-EDTA registry. Pediatr Nephrol 29(12):2403–2410.

80. WHO (2013) Health, health systems and economic crisis in Europe Impact and policy implications.

81. Ehrich JHH, El Gendi AA, Drukker A, et al (2005) Demography of paediatric renal care in Europe:

organization and delivery. Nephrol Dial Transplant 20:297–305.

82. Wisk LE, Witt WP (2012) Predictors of delayed or forgone needed health care for families with children.

189

Pediatrics 130:1027–37.

83. Schell CO, Reilly M, Rosling H, et al (2007) Socioeconomic determinants of infant mortality: a worldwide

study of 152 low-, middle-, and high-income countries. Scand J Public Health 35:288–97.

84. Europeristat (2010) European Perinatal Health Report: Health and Care of Pregnant Women and Babies in

Europe in 2010.

85. Richardus JH, Graafmans WC, Verloove-Vanhorick SP, Mackenbach JP (2003) Differences in perinatal

mortality and suboptimal care between 10 European regions: results of an international audit. BJOG 110:97–

105.

86. van Stralen KJ, Borzych-Dużalka D, Hataya H, et al (2014) Survival and clinical outcomes of children starting

renal replacement therapy in the neonatal period. Kidney Int 86:168–174.

87. Damen-Elias HAM, De Jong TPVM, Stigter RH, et al (2005) Congenital renal tract anomalies: outcome and

follow-up of 402 cases detected antenatally between 1986 and 2001. Ultrasound Obstet Gynecol 25:134–43.

88. Scott JES (2002) Fetal, perinatal, and infant death with congenital renal anomaly. Arch Dis Child 87:114–7.

89. Joseph KS, Kinniburgh B, Hutcheon JA, et al (2013) Determinants of increases in stillbirth rates from 2000 to

2010. Cmaj 185:345–351.

90. Khoshnood B, Greenlees R, Loane M, Dolk H (2011) Paper 2: EUROCAT public health indicators for

congenital anomalies in Europe. Birth Defects Res A Clin Mol Teratol 91 Suppl 1:S16-22.

91. Hanlon M, Burstein R, Masters SH, Zhang R (2012) Exploring the relationship between population density

and maternal health coverage. BMC Health Serv Res 12:416.

92. S. R. Alexander MH (1993) Continuous peritoneal dialysis for children: a decade of worldwide growth and

development. Kidney Int Suppl 40:S65–S74.

93. Chesnaye NC, Schaefer F, Groothoff JW, et al (2015) Disparities in treatment rates of paediatric end-stage

renal disease across Europe: Insights from the ESPN/ERA-EDTA registry. Nephrol Dial Transplant 30:1377–

1385.

94. Spiegelhalter DJ (2005) Funnel plots for comparing institutional performance. Stat Med 24:1185–1202.

95. Merlo J, Chaix B, Yang M, et al (2005) A brief conceptual tutorial of multilevel analysis in social epidemiology:

linking the statistical concept of clustering to the idea of contextual phenomenon. J Epidemiol Community

Health 59:443–449.

96. Aelvoet W, Terryn N, Blommaert A, et al (2016) Community-acquired pneumonia (CAP) hospitalizations

and deaths: is there a role for quality improvement through inter-hospital comparisons? Int J Qual Health

Care 28:22–32.

97. Legrand C, Duchateau L, Sylvester R, et al (2006) Heterogeneity in disease free survival between centers:

Lessons learned from an EORTC breast cancer trial. Clin Trials 3:10–18.

98. Merlo J, Yang M, Chaix B, et al (2005) A brief conceptual tutorial on multilevel analysis in social

epidemiology: investigating contextual phenomena in different groups of people. J Epidemiol Community

Health 59:729–736.

99. Jager K, Zoccali C, MacLeod A, Dekker F (2008) Confounding: What it is and how to deal with it. Kidney Int

73:256–260.

100. OECD (2012) Health at a Glance: Europe 2012. OECD

101. Covic A, Schiller A (2010) Burden of disease - prevalence and incidence of ESRD in selected European

regions and populations. Clin Nephrol 74 Suppl 1:S23-7.

190

102. Rutkowski B (2006) Highlights of the epidemiology of renal replacement therapy in Central and Eastern

Europe. Nephrol Dial Transplant 21:4–10.

103. Dussault G, Fronteira I, Cabral J (2009) Migration of health personnel in the WHO European Region.

104. Buchan J (2006) Migration of health workers in Europe: policy problem or policy solution?

105. Buchan J, Perfilieva G (2006) Health Worker Migration in the European Region: Country Case Studies and

Policy Implications.

106. European Society for Paediatric Nephrology. http://espn-online.org/training.

107. Chavers BM, Molony JT, Solid CA, et al (2015) One-Year Mortality Rates in US Children with End-Stage

Renal Disease. Am J Nephrol 41:121–128.

108. Groothoff JW (2005) Long-term outcomes of children with end-stage renal disease. Pediatr Nephrol

20:849–853.

109. Chesney RW (2002) The development of pediatric nephrology. Pediatr Res 52:770–778.

110. Carey WA, Martz KL, Warady BA (2015) Outcome of Patients Initiating Chronic Peritoneal Dialysis During

the First Year of Life. Pediatrics 136:e615–e622.

111. Laskin BL, Mitsnefes MM, Dahhou M, et al (2014) The mortality risk with graft function has decreased among

children receiving a first kidney transplant in the United States. Kidney Int 87:1–9.

112. Hattori M, Sako M, Kaneko T, et al (2015) End-stage renal disease in Japanese children: a nationwide survey

during 2006-2011. Clin Exp Nephrol 19:933–938.

113. National Institutes of Health National Institute of Diabetes and Digestive and Kidney Diseases. (2015)

United States Renal Data System. 2015 USRDS annual data report: Epidemiology of kidney disease in the

United States.

114. NAPRTCS (2011) NAPRTCS 2011 Annual Report.

115. Alexander RT, Foster BJ, Tonelli MA, et al (2012) Survival and transplantation outcomes of children less than

2 years of age with end-stage renal disease. Pediatr Nephrol 27:1975–1983.

116. Bunchman TE (1995) Chronic dialysis in the infant less than 1 year of age. Pediatr Nephrol S18-22.

117. Mekahli D, Shaw V, Ledermann SE, Rees L (2010) Long-term outcome of infants with severe chronic kidney

disease. Clin J Am Soc Nephrol 5:10–17.

118. Kari JA, Gonzalez C, Ledermann SE, et al (2000) Outcome and growth of infants with severe chronic renal

failure. Kidney Int 57:1681–1687.

119. Shooter M, Watson A (2000) The ethics of withholding and withdrawing dialysis therapy in infants. Pediatr

Nephrol 14:347–351.

120. Zurowska AM, Fischbach M, Watson AR, et al (2013) Clinical practice recommendations for the care of

infants with stage 5 chronic kidney disease (CKD5). Pediatr Nephrol 28:1739–1748.

121. Pollack S, Eisenstein I, Tarabeih M, et al (2015) Long-term hemodialysis therapy in neonates and infants with

end-stage renal disease: a 16-year experience and outcome. Pediatr Nephrol 31:305–313.

122. Hogan J, Couchoud C, Bonthuis M, et al (2016) Gender Disparities in Access to Pediatric Renal

Transplantation in Europe: Data from the ESPN/ERA-EDTA Registry. Am J Transplant 16:2097–2105.

123. Laskin BL, Mitsnefes MM, Dahhou M, et al (2014) The mortality risk with graft function has decreased among

children receiving a first kidney transplant in the United States. Kidney Int 87:1–9.

124. Laster M, Soohoo M, Hall C, et al (2016) Racial–ethnic disparities in mortality and kidney transplant

outcomes among pediatric dialysis patients. Pediatr. Nephrol.

191

125. Parekh RS, Carroll CE, Wolfe RA, Port FK (2002) Cardiovascular mortality in children and young adults with

end-stage kidney disease. J Pediatr 141:191–197.

126. Chavers BM, Li S, Collins AJ, Herzog CA (2002) Cardiovascular disease in pediatric chronic dialysis patients.

Kidney Int 62:648–653.

127. Mitsnefes M, Stablein D (2005) Hypertension in pediatric patients on long-term dialysis: a report of the

North American Pediatric Renal Transplant Cooperative Study (NAPRTCS). Am J Kidney Dis 45:309–315.

128. Flynn JT, Mitsnefes M, Pierce C, et al (2008) Blood pressure in children with chronic kidney disease: a report

from the Chronic Kidney Disease in Children study. Hypertension 52:631–637.

129. Tjaden LA, Noordzij M, van Stralen KJ, et al (2015) Racial Disparities in Access to and Outcomes of Kidney

Transplantation in Children, Adolescents, and Young Adults: Results From the ESPN/ERA-EDTA. Am J

Kidney Dis 67:293–301.


Renal Data System, USRDS 2008 Annual Data Report: Atlas of End-Stage Renal Disease in the United States.

131. Sule S, Fivush B, Neu A, Furth S (2011) Increased risk of death in pediatric and adult patients with ESRD

secondary to lupus. Pediatr Nephrol 26:93–98.

132. Harambat J, van Stralen KJ, Espinosa L, et al (2012) Characteristics and outcomes of children with primary

oxalosis requiring renal replacement therapy. Clin J Am Soc Nephrol 7:458–465.

133. Wong CS, Gipson DS, Gillen DL, et al (2000) Anthropometric measures and risk of death in children with

end-stage renal disease. Am J Kidney Dis 36:811–819.

134. Ku E, Glidden D, Hsu CY, et al (2016) Association of Body Mass Index with Patient-Centered Outcomes in

Children with ESRD. J Am Soc Nephrol 27:551–558.

135. Furth SL, Hwang W, Yang C, et al (2002) Growth failure, risk of hospitalization and death for children with

end-stage renal disease. Pediatr Nephrol 17:450–455.

136. Furth SL, Stablein D, Fine RN, et al (2002) Adverse Clinical Outcomes Associated With Short Stature at

Dialysis Initiation: A Report of the North American Pediatric Renal Transplant Cooperative Study. Pediatrics

109:909–913.

137. Ku E, Fine RN, Hsu C, et al (2016) Height at First RRT and Mortality in Children. Clin J Am Soc Nephrol

11:832–839.

138. Pruthi R, O’Brien C, Casula A, et al (2013) UK Renal Registry 15th annual report: Chapter 4 demography of

the UK paediatric renal replacement therapy population in 2011. Nephron Clin Pract 123 Suppl:81–92.

139. Tsai H, Yang L, Chin T, et al (2010) Outcome and risk factors for mortality in pediatric peritoneal dialysis.

Perit Dial Int 30:233–239.

140. Neu AM, Sander A, Borzych-Duzalka D, et al (2012) Comorbidities in chronic pediatric peritoneal dialysis

patients: a report of the International Pediatric Peritoneal Dialysis Network. Perit Dial Int 32:410–418.

141. Shroff R, Rees L, Trompeter R, et al (2006) Long-term outcome of chronic dialysis in children. Pediatr

Nephrol 21:257–264.

142. Wood EG, Hand M, Briscoe DM, et al (2001) Risk factors for mortality in infants and young children on

dialysis. Am J Kidney Dis 37:573–579.

143. Ledermann SE, Scanes ME, Fernando ON, et al (2000) Long-term outcome of peritoneal dialysis in infants. J

Pediatr 136:24–29.

144. Ellis EN, Pearson D, Champion B, Wood EG (1995) Outcome of infants on chronic peritoneal dialysis. Adv

192

Perit Dial 11:266–269.

145. Hijazi R, Abitbol CL, Chandar J, et al (2009) Twenty-five Years of Infant Dialysis: A Single Center

Experience. J Pediatr 155:111–117.

146. Amaral S, Sayed BA, Kutner N, Patzer RE (2016) Preemptive kidney transplantation is associated with

survival benefits among pediatric patients with end-stage renal disease. Kidney Int 90:1100–1108.

147. Korevaar JC, Feith GW, Dekker FW, et al (2003) Effect of starting with hemodialysis compared with

peritoneal dialysis in patients new on dialysis treatment�: A randomized controlled trial. Kidney Int

64:2222–2228.

148. Chesnaye NC, Schaefer F, Groothoff JW, et al (2016) Mortality risk in European children with end-stage

renal disease on dialysis. Kidney Int 89:1355–1362.

149. Vidal E, van Stralen KJ, Chesnaye NC, et al (2016) Infants Requiring Maintenance Dialysis: Outcomes of

Hemodialysis and Peritoneal Dialysis. Am. J. Kidney Dis. epub:

150. Chang HJ, Han KH, Cho MH, et al (2014) Outcomes of chronic dialysis in Korean children with respect to

survival rates and causes of death. Korean J Pediatr 57:135–139.

151. Van De Luijtgaarden MWM, Jager KJ, Segelmark M, et al (2016) Trends in dialysis modality choice and

related patient survival in the ERA-EDTA Registry over a 20-year period. Nephrol Dial Transplant 31:120–

128.


Renal Data System, USRDS 2011 Annual Data Report: Atlas of Chronic Kidney Disease and End-Stage Renal

Disease in the United States.

153. Bargman JM, Thorpe KE, Churchill DN (2001) Relative contribution of residual renal function and peritoneal

clearance to adequacy of dialysis: a reanalysis of the CANUSA study. J Am Soc Nephrol 12:2158–2162.

154. Liao C-T, Chen Y-M, Shiao C-C, et al (2009) Rate of decline of residual renal function is associated with all-

cause mortality and technique failure in patients on long-term peritoneal dialysis. Nephrol Dial Transplant

24:2909–2914.

155. Guzzo I, Mancini E, Kengne Wafo S, et al (2009) Residual renal function and nutrition in young patients on

chronic hemodialysis. Pediatr Nephrol 24:1391–1397.

156. Chadha V, Blowey DL, Warady BA (2001) Is growth a valid outcome measure of dialysis clearance in

children undergoing peritoneal dialysis? Perit Dial Int 21 Suppl 3:S179-184.

157. Borzych D, Rees L, Ha IS, et al (2010) The bone and mineral disorder of children undergoing chronic

peritoneal dialysis. Kidney Int 78:1295–1304.

158. Wright S, Klausner D, Baird B, et al (2010) Timing of Dialysis Initiation and Survival in ESRD. Clin J Am Soc

Nephrol 5:1828–1835.

159. Korevaar JC, Jansen MA, Dekker FW, et al (2001) When to initiate dialysis: effect of proposed US guidelines

on survival. Lancet 358:1046–1050.

160. Lee J, An JN, Hwang JH, et al (2014) Effect of Dialysis Initiation Timing on Clinical Outcomes: A Propensity-

Matched Analysis of a Prospective Cohort Study in Korea. PLoS One 9:e105532.

161. Atkinson MA, Oberai PC, Neu AM, et al (2010) Predictors and consequences of higher estimated

glomerular filtration rate at dialysis initiation. Pediatr Nephrol 25:1153–61.

162. Cooper BA, Branley P, Bulfone L, et al (2010) A Randomized, Controlled Trial of Early versus Late Initiation

of Dialysis. N Engl J Med 363:609–619.

193

163. Rama I, Grinyó JM (2010) Malignancy after renal transplantation: the role of immunosuppression. Nat Rev

Nephrol 6:511–519.

164. Bartosh SM, Leverson G, Robillard D, Sollinger HW (2003) Long-term outcomes in pediatric renal

transplant recipients who survive into adulthood. Transplantation 76:1195–1200.

165. Smith J, Martz K, Blydt-Hansen TD (2013) Pediatric kidney transplant practice patterns and outcome

benchmarks , 1987 – 2010�: A report of the North American Pediatric Renal Trials and Collaborative

Studies. Pediatr Transplant 17:149–157.

166. Webster AC, Craig JC, Simpson JM, et al (2007) Identifying high risk groups and quantifying absolute risk of

cancer after kidney transplantation: A cohort study of 15 183 recipients. Am J Transplant 7:2140–2151.

167. Ploos van Amstel S, Vogelzang JL, Starink M V, et al (2015) Long-Term Risk of Cancer in Survivors of

Pediatric ESRD. Clin J Am Soc Nephrol 10:2198–2204.

168. Luyckx VA, Naicker S, McKee M (2013) Equity and economics of kidney disease in sub-Saharan Africa.

Lancet 382:103–104.

169. Sinha A, Bagga A (2014) Maintenance dialysis in developing countries. Pediatr Nephrol 211–219.

170. Harambat J, Ekulu PM (2016) Inequalities in access to pediatric ESRD care: a global health challenge. Pediatr

Nephrol 31(3):353–358.

171. Miller MEY, Williams JA (2009) Chronic Renal Failure in Jamaican Children – an Update (2001 – 2006). West

Indian Med J 58:231–234.

172. Asinobi AO, Ademola AD, Ogunkunle OO, Mott SA (2014) Paediatric end-stage renal disease in a tertiary

hospital in South West Nigeria. BMC Nephrol 15:25.

173. Mong Hiep TT, Janssen F, Ismaili K, et al (2008) Etiology and outcome of chronic renal failure in hospitalized

children in Ho Chi Minh City, Vietnam. Pediatr Nephrol 23:965–970.

174. Gulati S, Mittal S, Sharma RK, Gupta A (1999) Etiology and outcome of chronic renal failure in Indian

children. Pediatr Nephrol 13:594–596.

175. Nahas AM El, Bello AK (2010) Chronic kidney disease: the global challenge. Lancet 365:331–340.

176. Kramer A, Stel VS, Tizard J, et al (2009) Characteristics and survival of young adults who started renal

replacement therapy during childhood. Nephrol Dial Transplant 24:926–933.

177. Vogelzang JL, Heestermans LWAA, van Stralen KJ, et al (2013) Simultaneous reversal of risk factors for

cardiac death and intensi fi ed therapy in long-term survivors of paediatric end-stage renal disease over the

last 10 years.

178. Dobrowolski LC, van Huis M, van der Lee JH, et al (2016) Epidemiology and management of hypertension in

paediatric and young adult kidney transplant recipients in The Netherlands. Nephrol Dial Transplant

31:1947–1956.

179. Groothoff JW, Gruppen MP, Offringa M, et al (2002) Increased arterial stiffness in young adults with end-

stage renal disease since childhood. J Am Soc Nephrol 13:2953–2961.

180. Gruppen MP, Groothoff JW, Prins M, et al (2003) Cardiac disease in young adult patients with end-stage

renal disease since childhood: A Dutch cohort study. Kidney Int 63:1058–1065.

181. Rianthavorn P, Ettenger RB, Malekzadeh M, et al (2004) Noncompliance with immunosuppressive

medications in pediatric and adolescent patients receiving solid-organ transplants. Transplantation 77:778–

782.

182. Van Arendonk KJ, James NT, Boyarsky BJ, et al (2013) Age at graft loss after pediatric kidney

194

transplantation: exploring the high-risk age window. Clin J Am Soc Nephrol 8:1019–1026.

183. Watson AR (2000) Non-compliance and transfer from paediatric to adult transplant unit. Pediatr Nephrol

14:469–472.

184. IPNA (2017) IPNA website. https://www.ipna-online.org/. Accessed 6 Feb 2017

185. Liu FX, Rutherford P, Smoyer-Tomic K, et al (2015) A global overview of renal registries: a systematic

review. BMC Nephrol 16:1–10.

186. Verrina E, Edefonti A, Gianoglio B, et al (2004) A multicenter experience on patient and technique survival

in children on chronic dialysis. Pediatr Nephrol 19:82–90.

187. Mandrekar JN Cutpoint Determination Methods in Survival Analysis using SAS.

188. Vonesh EF (2004) The differential impact of risk factors on mortality in hemodialysis and peritoneal dialysis.

66:2389–2401.

189. Heaf JG, Løkkegaard H, Madsen M (2002) Initial survival advantage of peritoneal dialysis relative to

haemodialysis. Nephrol Dial Transpl 17:112–117.

190. Perl J, Wald R, McFarlane P, et al (2011) Hemodialysis vascular access modifies the association between

dialysis modality and survival. J Am Soc Nephrol 22:1113–21.

191. Termorshuizen F (2003) Hemodialysis and Peritoneal Dialysis: Comparison of Adjusted Mortality Rates

According to the Duration of Dialysis: Analysis of the Netherlands Cooperative Study on the Adequacy of

Dialysis 2. J Am Soc Nephrol 14:2851–2860.

192. Vonesh EF, Snyder JJ, Foley RN, Collins AJ (2006) Mortality studies comparing peritoneal dialysis and

hemodialysis: what do they tell us? Kidney Int Suppl 70:S3-11.

193. Jansen MAM, Hart AAM, Korevaar JC, et al (2002) Predictors of the rate of decline of residual renal function

in incident dialysis patients. Kidney Int 62:1046–53.

194. Boehm M, Winkelmayer WC, Arbeiter K, et al (2010) Late referral to paediatric renal failure service impairs

access to pre-emptive kidney transplantation in children. Arch Dis Child 95:634–638.

195. Jander A, Nowick M, Tkaczyk M, et al (2006) Does a late referral to a nephrologist constitute a problem in

children starting renal replacement therapy in Poland? - A nationwide study. Nephrol Dial Transplant

21:957–961.

196. Kennedy SE, Bailey R, Kainer G (2012) Causes and outcome of late referral of children who develop end-

stage kidney disease. J Paediatr Child Health 48:253–8.

197. Warady B a., Abraham AG, Schwartz GJ, et al (2015) Predictors of Rapid Progression of Glomerular and

Nonglomerular Kidney Disease in Children and Adolescents: The Chronic Kidney Disease in Children

(CKiD) Cohort. Am. J. Kidney Dis.

198. Ivarsen P, Povlsen J V (2014) Can peritoneal dialysis be applied for unplanned initiation of chronic dialysis?

Nephrol Dial Transplant 29:2201–6.

199. Foley RN, Chen S-C, Solid CA, et al (2014) Early mortality in patients starting dialysis appears to go

unregistered. Kidney Int 86:1–7.

200. Watson AR, Hayes WN, Vondrak K, et al (2013) Factors influencing choice of renal replacement therapy in

European paediatric nephrology units. Pediatr Nephrol 28:2361–8.

201. Geary DF (1998) Attitudes of pediatric nephrologists to management of end-stage renal disease in infants. J

Pediatr 154–156.

202. Laakkonen H, Holtta T, Lonnqvist T, et al (2008) Peritoneal dialysis in children under two years of age.

195

Nephrol Dial Transplant 23:1747–1753.

203. Vidal E, Edefonti A, Murer L, et al (2012) Peritoneal dialysis in infants: the experience of the Italian Registry

of Paediatric Chronic Dialysis. Nephrol Dial Transplant 27:388–395.

204. Wedekin M, Ehrich JHH, Offner G, Pape L (2010) Renal replacement therapy in infants with chronic renal

failure in the first year of life. Clin J Am Soc Nephrol 5:18–23.

205. Teh JC, Frieling ML, Sienna JL, Geary DF (2011) Attitudes of caregivers to management of end-stage renal

disease in infants. Perit Dial Int 31:459–465.

206. Feber J, Schärer K, Schaefer F, et al (1994) Residual renal function in children on haemodialysis and

peritoneal dialysis therapy. Pediatr Nephrol 8:579–583.

207. Kiliś-Pstrusińska K, Wasilewska A, Medyńska A, et al (2013) Psychosocial aspects of children and families of

children treated with automated peritoneal dialysis. Pediatr Nephrol 28:2157–2167.

208. Donckerwolcke RA, Bunchman TE (1994) Hemodialysis in infants and small children. Pediatr Nephrol 8:103–

6.

209. Shroff R, Wright E, Ledermann S, et al (2003) Chronic hemodialysis in infants and children under 2 years of

age. Pediatr Nephrol 18:378–83.

210. Feinstein S, Rinat C, Becker-Cohen R, et al (2008) The outcome of chronic dialysis in infants and toddlers--

advantages and drawbacks of haemodialysis. Nephrol Dial Transplant 23:1336–45.

211. Quinlan C, Bates M, Sheils A, et al (2013) Chronic hemodialysis in children weighing less than 10 kg. Pediatr

Nephrol 28:803–809.

212. Bonthuis M, van Stralen KJ, Verrina E, et al (2012) Use of national and international growth charts for

studying height in European children: development of up-to-date European height-for-age charts. PLoS One

7:e42506.

213. Ehrich JH, Rizzoni G, Brunner FP, et al (1992) Renal replacement therapy for end-stage renal failure before 2

years of age. Nephrol Dial Transplant 7:1171–7.

214. Goldstein SL, Macierowski CT, Jabs K (1997) Hemodialysis catheter survival and complications in children

and adolescents. Pediatr Nephrol 11:74–7.

215. Chalem Y, Ryckelynck J-P, Tuppin P, et al (2005) Access to, and outcome of, renal transplantation according

to treatment modality of end-stage renal disease in France. Kidney Int 67:2448–53.

216. Fuggle S V, Allen JE, Johnson RJ, et al (2010) Factors affecting graft and patient survival after live donor

kidney transplantation in the UK. Transplantation 89:694–701.

217. Gill JS, Gill J, Rose C, et al (2006) The older living kidney donor: Part of the solution to the organ shortage.

Transplantation 82:1662–6.

218. Øien CM, Reisaeter AV, Leivestad T, et al (2007) Living donor kidney transplantation: the effects of donor

age and gender on short- and long-term outcomes. Transplantation 83:600–6.

219. Van Arendonk KJ, King E a., Orandi BJ, et al (2015) Loss of Pediatric Kidney Grafts During the “High-Risk

Age Window”: Insights From Pediatric Liver and Simultaneous Liver-Kidney Recipients. Am J Transplant

15:445–452.

220. Smith JM, Ho PLM, McDonald RA (2002) Renal transplant outcomes in adolescents: a report of the North

American Pediatric Renal Transplant Cooperative Study. Pediatr Transplant 6:493–499.

221. Foster BJ, Pai ALH (2014) Adherence in Adolescent and Young Adult Kidney Transplant Recipients. Open

Urol Nephrol J 7:133–143.

196

222. Arbus GS, Rochon J, Thompson D (1991) Survival of cadaveric renal transplant grafts from young donors

and in young recipients. Pediatr Nephrol 5:152–157.

223. Briscoe DM, Kim MS, Lillehei C, et al (1992) Outcome of renal transplantation in children less than two

years of age. Kidney Int 42:657–662.

224. Postlethwaite RJ, Johnson RJ, Armstrong S, et al (2002) The outcome of pediatric cadaveric renal

transplantation in the UK and Eire. Pediatr Transplant 6:367–77.

225. Jahnukainen T, Bjerre A, Larsson M, et al (2016) The second report of the Nordic Pediatric Renal

Transplantation Registry 1997-2012: More infant recipients and improved graft survivals. Pediatr Transplant

20:364–371.

226. Opelz G, Collaborative Transplant Study (1988) Influence of recipient and donor age in pediatric renal

transplantation. Transpl Int 1:95–98.

227. Dale-Shall AW, Smith JM, McBride MA, et al (2009) The relationship of donor source and age on short- and

long-term allograft survival in pediatric renal transplantation. Pediatr Transplant 13:711–718.

228. Opelz G, Döhler B (2010) Pediatric kidney transplantation: analysis of donor age, HLA match, and

posttransplant non-Hodgkin lymphoma: a collaborative transplant study report. Transplantation 90:292–297.

229. Dharnidharka VR, Stevens G, Howard RJ (2005) En-bloc kidney transplantation in the United States: An

analysis of United Network of Organ Sharing (UNOS) data from 1987 to 2003. Am J Transplant 5:1513–

1517.

230. Sharma A, Fisher RA, Cotterell AH, et al (2011) En Bloc Kidney Transplantation From Pediatric Donors:

Comparable Outcomes With Living Donor Kidney Transplantation. Transplantation 92:564–569.

231. Dion M, Rowe N, Shum J, et al (2014) Donation after Cardiac Death Pediatric En Bloc Renal

Transplantation. J Urol 193:1–5.

232. Winnicki E, Dharmar M, Tancredi D, Butani L (2016) Comparable Survival of En Bloc versus Standard

Donor Kidney Transplants in Children. J Pediatr 173:1–6.

233. Cransberg K, Van Gool JD, Davin JC, et al (2000) Pediatric renal transplantations in the Netherlands. Pediatr


234. Eurotransplant Foundation (2013) Eurotransplant Manual – version 3.0. Eurotransplant Foundation, Leiden

235. Foster BJ, Dahhou M, Zhang X, et al (2013) Relative importance of HLA mismatch and donor age to graft

survival in young kidney transplant recipients. Transplantation 96:469–75.

236. Simpson CM, McTaggart SJ, Sterne JAC, et al (2005) Grandparent donors in paediatric renal transplantation.

Pediatr Nephrol 20:1636–1641.

237. Papachristou F, Stabouli S, Printza N, et al (2016) Long-term outcome of pediatric kidney transplantation: A

single-center experience from Greece. Pediatr Transplant 500–506.

238. WHO (2011) Rio Political Declaration on Social Determinants of Health.

239. Caskey FJ, Schober-Halstenberg H-J, Roderick PJ, et al (2006) Exploring the differences in epidemiology of

treated ESRD between Germany and England and Wales. Am J Kidney Dis 47:445–54.

240. Couchoud C, Guihenneuc C, Bayer F, et al (2012) Medical practice patterns and socio-economic factors may

explain geographical variation of end-stage renal disease incidence. Nephrol Dial Transplant 27:2312–22.

241. Hommel K, Rasmussen S, Kamper A-L, Madsen M (2010) Regional and social inequalities in chronic renal

replacement therapy in Denmark. Nephrol Dial Transplant 25:2624–32.

242. Böger CA, Gorski M, Li M, et al (2011) Association of eGFR-related loci identified by GWAS with incident

197

CKD and ESRD. PLoS Genet 7:3–10.

243. Köttgen A, Glazer NL, Dehghan A, et al (2011) Multiple Novel Loci are Associated with Indices of Renal

Function and Chronic Kidney Disease. 41:712–717.

244. Stiller CA, Parkin DM (1990) International variations in the incidence of childhood lymphomas. Paediatr

Perinat Epidemiol 4:303–324.

245. Stiller CA (2004) Epidemiology and genetics of childhood cancer. Oncogene 23:6429–44.

246. Hölttä T, Bonthuis M, Van Stralen KJ, et al (2016) Timing of renal replacement therapy does not influence

survival and growth in children with congenital nephrotic syndrome caused by mutations in NPHS1: data

from the ESPN/ERA-EDTA Registry. Pediatr Nephrol 2317–2325.

247. Bodenheimer T (2013) High and Rising Health Care Costs. Part 2: Technologic Innovation.

248. Sorenson C, Drummond M, Bhuiyan Khan B (2013) Medical technology as a key driver of rising health

expenditure: disentangling the relationship. Clinicoecon Outcomes Res 5:223–34.

249. Bech M (2010) The influence of economic incentives and regulatory factors on the adoption of treatment

technologies: a case study of technologies used to treat heart attacks. 18:1114–1132.

250. Vecchi AF De, Dratwa M, Wiedemann ME (1999) Nephrology Dialysis Transplantation Healthcare systems

and end-stage renal disease ( ESRD ) therapies — an international review : costs and reimbursement /

funding of ESRD therapies. 14:31–41.

251. Dor A, Pauly M V, Eichleay M a, Held PJ (2007) End-stage renal disease and economic incentives: the

International Study of Health Care Organization and Financing (ISHCOF).

252. Kollée LAA, Cuttini M, Delmas D, et al (2009) Obstetric interventions for babies born before 28 weeks of

gestation in Europe: results of the MOSAIC study. BJOG 116:1481–91.

253. Fan ZJ, Lackland DT, Lipsitz SR, et al (2007) Geographical patterns of end-stage renal disease incidence and

risk factors in rural and urban areas of South Carolina. Health Place 13:179–87.

254. Couchoud C, Guihenneuc C, Bayer F, Stengel B (2010) The timing of dialysis initiation affects the incidence

of renal replacement therapy. Nephrol Dial Transplant 25:1576–8.

255. Grassmann A, Gioberge S, Moeller S, Brown G (2005) ESRD patients in 2004: global overview of patient

numbers, treatment modalities and associated trends. Nephrol Dial Transplant 20:2587–93.

256. Boyle PJ, Kudlac H, Williams AJ (1996) Geographical variation in the referral of patients with chronic end

stage renal failure for renal replacement therapy. QJM 89:151–7.

257. Mircescu G, Capsa D, Covic M, et al (2004) Nephrology and renal replacement therapy in Romania--

transition still continues (Cinderella story revisited). Nephrol Dial Transplant 19:2971–80.

258. Rodwell C, Aymé S (2015) Rare disease policies to improve care for patients in Europe. Biochim Biophys

Acta - Mol Basis Dis 1852:2329–2335.

259. Union E (2011) European Union Committee of Experts on Rare Diseases: recommendations on quality

criteria for centres of expertise for rare diseases in Member States.

260. Yeates K, Ghosh S, Kilonzo K (2013) Developing nephrology programs in very low-resource settings:

challenges in sustainability. Kidney Int Suppl 3:202–205.

261. Jiang Y, Shen Y, Lau KKH (2014) Survey of chronic haemodialysis in children between 2007 and 2012 in

China. Nephrology 19:375–378.

262. Wühl E, Trivelli A, Picca S, et al (2009) Strict blood-pressure control and progression of renal failure in

children. N Engl J Med 361:1639–50.

198

263. Chaudhary K, Sangha H, Khanna R (2011) Peritoneal dialysis first: rationale. Clin J Am Soc Nephrol 6:447–

56.

264. Juergensen E, Wuerth D, Finkelstein SH, et al (2006) Hemodialysis and peritoneal dialysis: patients’

assessment of their satisfaction with therapy and the impact of the therapy on their lives. Clin J Am Soc

Nephrol 1:1191–1196.

265. Rubin HR, Fink NE, Plantinga LC, et al (2004) Patient ratings of dialysis care with peritoneal dialysis vs

hemodialysis. JAMA 291:697–703.

266. Karahalios A, Baglietto L, Carlin JB, et al (2012) A review of the reporting and handling of missing data in

cohort studies with repeated assessment of exposure measures. BMC Med. Res. Methodol. 12:

267. Ehrich JHH, Kerbl R, Pettoello-Mantovani M, Lenton S (2015) Opening the Debate on Pediatric

Subspecialties and Specialist Centers: Opportunities for Better Care or Risks of Care Fragmentation? J

Pediatr 167:1177–1178.

268. Ethier J, Mendelssohn DC, Elder SJ, et al (2008) Vascular access use and outcomes: An international

perspective from the dialysis outcomes and practice patterns study. Nephrol Dial Transplant 23:3219–3226.

Documents

UvA-DARE (Digital Academic Repository) European ... · of paediatric RRT programs in the 1960s, ESRD in children was a death sentence. Since then, substantial advances in renal medicine