Structural Differences between Health Insurance Funds and their Impact on Health Services Research: Results from the Bertelsmann Health-Care Monitor

Aim: Claims data of health insurance companies are an important database for health services research. We investigated if there are differences in baseline characteristics and prevalence of chronic diseases between members of several health insurance funds in Germany, and if so, whether adjusting for age and sex could explain these differences. Methods: 10 representative surveys (conducted between 2004 and 2008) of the 'Bertelsmann Health-Care Monitor' comprising 15 089 participants aged 18 to 79 years were analysed. Our main independent variable was membership in one of 8 health insurance funds. The prevalence of self-reported hypertension, diabetes, atopic diseases, coronary heart disease and heart failure was studied. We first estimated the crude prevalence of chronic diseases stratified by these funds. We further fitted logistic regression models and adjusted for age and sex as well as for further comorbidities and health related factors. Results: Most respondents were insured in the BKK (Betriebskrankenkassen; 20.1 %), the AOK (Allgemeinen Ortskrankenkassen; 19.2 %) and private health insurances (15.3 %). Substantial differences were found according to age, sex, educational level and prevalences of chronic diseases. Stratified by health insurance funds, prevalences ranged between 17.1-29.6 % for hypertension, between 3.9-11.4 % for diabetes, between 4.3-6.7 % for atopic diseases, between 3.4-6.7 % for coronary heart disease and between 2.6-5.7 % for heart failure. When adjusting for sex and age, estimates for all 5 diseases were significant higher in AOK members compared to privately insured persons (3 diseases within the BAMER and the DAK, accordingly). In total, this was the case for 17 out of 35 comparisons. Even after adjusting for further comorbidities and health related factors 6 out of 35 comparisons showed significant increased estimates compared to privately insured persons. Conclusion: We found considerable differences in the prevalence of chronic diseases between German health insurance funds that remained after controlling for age and sex, and even after adjustment for further health-related variables. Further methodological studies are urgently needed to assess strengths and weaknesses of German claim data.