Migrants are often poorly represented in epidemiological studies which limits the generalizability of the results of population-based studies. This study aimed to assess whether a community-based sampling (CBS) of persons of Turkish origin leads to differences in the participants' characteristics compared to a register-based sampling (RBS). The two sampling strategies were used to recruit participants in three cities in Germany (CBS: n = 641; RBS: n = 578). We compared sociodemographic, migration- and health-related characteristics. Census data were used as an external reference. Lower German language skills and a lower acculturation status were more prevalent in the CBS than in the RBS. While age and sex adjusted obesity prevalence differed [CBS: 37.8 (33.6-42.4); RBS 30.0 (26.3-34.0); census data 19.1 (18.2-20.1)], most other health indicators were similar across the samples. In conclusion, the CBS approach led to a greater representation of persons of Turkish origin with lower language skills and lower acculturation status. Nevertheless, both recruitment strategies provided similar estimates of health status indicators.
