[Participation of migrants in health surveys conducted by telephone: potential and limits]

Migrants living in Germany are a both large and vulnerable population subgroup. They are not easily induced to participate in health surveys, Hence, achieving high participation rates of migrants in health surveys and avoiding selection bias is a difficult task. In this study, we report on the participation of migrants in the German National Health Telephone Survey 2003 (GSTel03), the first comprehensive national health survey conducted by telephone in Germany. Three migrant groups were identified: individuals with non-German citizenship (foreigners), naturalized migrants, and ethnic German immigrants (Spätaussiedler). The aim of this study is to evaluate the degree to which the GSTel03 subsample of foreigners is representative for foreigners living in Germany. We compare the prevalence of sociodemographic characteristics and selected health indicators of foreigners in the GNTel03 subsample with prevalences from national statistics and from a large national household survey ("Mikrozensus 2003"). The proportion of participants with non-German nationality in the overall GSTel03 sample was significantly lower than the proportion of foreigners in the residential population in Germany (3.7 % vs. 8.9 %). While there was no evidence of selection bias with regard to age and sex distribution, we found significant differences with regard to other factors, including nationality, length of stay in Germany, unemployment rate and education. The comparison of health indicators showed only moderate differences between GSTel03 sample and "Mikrozensus" results. However, these differences did not consistently point to a better or worse health status in the GSTel03 sample of foreigners and should therefore not be generalised in respect of other health indicators. Our study emphasises the importance of a continuous effort to improve migrant participation in health studies and of a thorough analysis of selection bias when interpreting results.