Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
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
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
European demographics
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
European demographics
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.
European demographics
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.
European demographics
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
European demographics
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.
European demographics
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.
European demographics
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) -
European demographics
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
European demographics
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
ESPN/ERA-EDTA Registry
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.
European disparities in RRT incidence
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.
European disparities in RRT incidence
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].
European disparities in RRT incidence
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.
European disparities in RRT incidence
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.
European disparities in RRT incidence
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.
European disparities in RRT incidence
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)
European disparities in RRT incidence
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
European disparities in RRT incidence
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,
European disparities in RRT incidence
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.
European disparities in RRT incidence
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
European disparities in RRT incidence
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.
European disparities in RRT incidence
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
European disparities in RRT incidence
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.
European disparities in RRT incidence
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.
European disparities in RRT incidence
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
countries paediatric (patient) population and the countries influence on the regression line. Country
abbreviations were expressed by ISO2 country codes.
European disparities in RRT incidence
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
countries paediatric (patient) population and the countries influence on the regression line. Country
abbreviations were expressed by ISO2 country codes.
European disparities in RRT incidence
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
abbreviations were expressed by ISO2 country codes.
European disparities in RRT incidence
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
abbreviations were expressed by ISO2 country codes.
European disparities in RRT incidence
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.
European disparities in RRT incidence
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
European disparities in RRT incidence
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.
European disparities in RRT incidence
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.
European disparities in RRT incidence
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.
Mortality risk disparities
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.
Mortality risk disparities
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
Mortality risk disparities
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
Mortality risk disparities
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
Mortality risk disparities
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
Mortality risk disparities
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
Mortality risk disparities
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
95%-95% control limits
95%-99% control limits
>99% control limits
Mortality rate
European average
Mortality risk disparities
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%
Mortality risk disparities
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
Mortality risk disparities
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.
Mortality risk disparities
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
Mortality risk disparities
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
Mortality risk disparities
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
Mortality risk disparities
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
Mortality risk disparities
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
Mortality risk disparities
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.
Mortality risk disparities
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.
Mortality risk disparities
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.
Mortality risk disparities
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
Mortality risk disparities
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%-95% control limits
95%-95% control limits
95%-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].
Survival in paediatric RRT
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]
Survival in paediatric RRT
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%
Survival in paediatric RRT
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
Survival in paediatric RRT
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].
Survival in paediatric RRT
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
Survival in paediatric RRT
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].
Survival in paediatric RRT
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].
Survival in paediatric RRT
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].
Survival in paediatric RRT
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
Survival in paediatric RRT
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.
Survival in paediatric RRT
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.
Survival in paediatric RRT
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.
Survival in paediatric RRT
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
ESPN/ERA-EDTA Registry
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.
Mortality risk on dialysis
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].
Statistical analysis
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-
Mortality risk on dialysis
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
Mortality risk on dialysis
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
Mortality risk on dialysis
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
Primary renal disease
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
Time under treatment of a nephrologist†
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.
Mortality risk on dialysis
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).
Mortality risk on dialysis
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.
Mortality risk on dialysis
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
Time under treatment of a nephrologist†
1 - 5 months 14 / 155 13 / 357 13 / 139 5 / 160
> 5 months 13 / 771 28 / 1255 12 / 644 13 / 669
Mortality risk on dialysis
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)
Primary renal disease
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)
Time under treatment of a nephrologist†
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)
Mortality risk on dialysis
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.
Mortality risk on dialysis
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
Mortality risk on dialysis
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
Mortality risk on dialysis
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
Mortality risk on dialysis
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.
Mortality risk on dialysis
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.
Mortality risk on dialysis
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.
Mortality risk on dialysis
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
Mortality risk on dialysis
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
Mortality risk on dialysis
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.
Mortality risk on dialysis
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.
Statistical analysis
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)
European demographics
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.
Statistical analysis
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 12-19 DD 294 92.2 (89.0-95.6) Reference Reference
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 12-19 DD 653 83.8 (80.4-87.3) Reference Reference
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
Transplant 16:1120–9.
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.
130. National Institutes of Health National Institute of Diabetes and Digestive and Kidney Diseases (2008) U.S.
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.
152. National Institutes of Health National Institute of Diabetes and Digestive and Kidney Diseases (2011) U.S.
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
Transplant 4:72–81.
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 qual- ity
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.