The case of diabetes in the Netherlands

Kath Bennett

19th Jan 2009

Data sources

• General sources of data for diabetes in EUROPE– Diabetes ATLAS– EUCID project– Global Burden of disease project 2000 (Wild et al ).

Table 3.1 gives table of prevalence data for all countries including Netherlands. (Mooy et al HOORN study, Diabetes care 1995;18- Old data)

• Incidence data less widely available according to GBD 2000.

EUCID Project

• The aim of the European Core Indicators in Diabetes (EUCID) project is to collect and compare data about risk factors for diabetes, complications and quality of care indicators in EU countries or future member states.

19 countries provided data for a list of indicators by age band which were representative at a regional or a national level for 2004, 2005 or 2006. The indicators for this project were designed during the EUropean Diabetes Indicators Project  - EUDIP.  Data were age-standardized for comparisons performed in the general population. Recently EUCID's final report was published at the DG SANCO website.

www.eucid.eu

The final list of collaborating institutes of the contract

Country OrganisationAustria Universitätsklinik für Kinder- und JugendheilkundeBelgium Scientific Institute of Public Health, WIV-ISP BrusselsCyprus Ministry of Health, Health Monitoring UnitDenmark Danish Diabetes DatabaseEngland Yorkshire and Humber Public Health Observatory, University of YorkFinland KTL, National Public Health InstituteFrance Institut de Veille SanitaireGermany Hospital GK HavelhoeheGreece Hippocrateion HospitalIreland The Adelaide & Meath HospitalItaly Associazione Medice DiabetologiLuxembourg Centre Hospitalier de LuxembourgPortugal Direccão-General da Saúde, Ministerio da SaudeRomania Institute of Diabetes “N. Paulescu”Scotland University of Dundee, Ninewells Hospital and Medical SchoolSpain Consejería Salud, Delegacíon Provincial de MálagaSweden NEPI FoundationNetherlands Dutch Institute of Health Care Improvement CBOTurkey Turkish Diabetes Foundation

Data selection

• Both incidence and prevalence from same data source

• Most recent data at the time (year 2003, but 2004 has since become available Dec 2008)

• As representative of the population as possible (national)

• Registries recording all patients

Particular problems with diabetes

• Under-reporting of diabetes as cause of death on death cerificates

• Those with diabetes often die of CVD

Mortality data - Diabetes

• Numbers of deaths caused by diabetes from Statistics NL database for 2006.

• Denominators (pop 2006) provided by Wilma (statline CBS NL).

• Calculated rates from the dividing number of deaths by population.

• Same as numbers of deaths from WHO HFA for 2006

Prevalence/incidence - Diabetes

Literature identified different sources of data:• HOORN study in elderly (50-74 yrs); 8.30% (Mooy et al)• Rotterdam study (55+ years); 11.30% (Stock et al)Different registries:• CMR Nijmegen (I/P, Nijmegen, available since 1971)• RNH Limburg (I/P, Limburg region, 1998 on)• Transition project (Incidence, multi-region 85-95)Often both types of diabetes combined as not possible to

distinguish. Sources from report by Hoogeveen et al RIVM March 2000,

Prevalence - diabetes

• Other studies – Zodiac-I

Using DISMOD

• Prevalence, incidence and cause-specific data entered into MS access (as for GBD2000) as input tables

• Import data into DISMOD II. Population data and all-cause mortality data imported into DISMOD II.

• Smooth input Incidence and prevalence data within DISMOD and then calculate output estimates.

• Apply age at one year intervals and also to saved tables from output.

Outputs from DISMOD

Mortality- DM calculated and reported

0

0.002

0.004

0.006

0.008

0.01

0.012

0.014

0.016

40-44

45-49

50-54

55-59

60-64

65-69

70-74

75-79

80-84

85-89

90-94

95+

Age

Rat

e DISMOD

WHO HFA

Different approaches

• GBD 2000 Smooth incidence and prevalence curves generated within DISMOD.

• Hoogeveen 2000: Smooths data first and interpolate over age before estimating model parameters. Uses penalty functions for smoothing with smoothing coefficients specified for each disease, in this case 0.5).

Suggestions

• Different smoothing approaches• Consistency with other data sources, existing

evidence (age patterns, gender differences)?• Approach/source similar for other member states?

– Can apply similar approach– WHO HFA combined source for mortality, not the case

for morbidity– ? Link with EUCID project for consistent approach?