Health monitoring at the Robert Koch Institute – Effects of a change in study design on sample composition and prevalence estimates due to the start of the ‘Health in Germany’ Panel

Authors:
  • EMElvira Mauz
    Elvira Mauz
  • FVFelicitas Vogelgesang
    Felicitas Vogelgesang
  • SDStefan Damerow
    Stefan Damerow
  • ASAnja Schienkiewitz
    Anja Schienkiewitz
  • NMNiels Michalski
    Niels Michalski
  • BGBeate Gaertner
    Beate Gaertner
  • JAJennifer Allen
    Jennifer Allen
  • JBJens Baumert
    Jens Baumert
  • YDYong Du
    Yong Du
  • RKRonny Kuhnert
    Ronny Kuhnert
  • JLJohannes Lemcke
    Johannes Lemcke
  • RSRamona Scheufele
    Ramona Scheufele

Abstract
The Robert Koch Institute (RKI) continuously monitors key health indicators in the general population by collecting data repeatedly. As changes in survey design can affect prevalence estimates and thus make interpreting trends difficult, the launch of the RKI Panel ‘Health in Germany’ was accompanied by a methodological study. The RKI Panel is based on a random sample drawn from population registers. The survey is self-administered in written format (online or paper). The composition of the sample, prevalence estimates and response behaviour were then compared with data collected in parallel in the GEDA 2024 telephone survey. Data from previous surveys were included in the modelling to quantify method-related differences in prevalence estimates. The RKI Panel 2024 was more successful in representing young adults, the elderly and individuals with low levels of education. The prevalence estimates differ significantly from GEDA 2024 in some cases, particularly for mental health indicators and their associated factors. The RKI Panel includes more older adults with limited physical health, while in young adults more participants with poorer mental health are present. Despite method-related differences in prevalence, modelling can usually be used to assess trends. The RKI Panel provides a more realistic representation of the German population than previous telephone surveys. The differences in prevalence are due to effects of the survey mode, questionnaire design, and changes in sample composition.

This publication has 33 references indexed in Scilit: