[Problems of evaluating regional integrated care using the example of "Gesundes Kinzigtal Integrated Care"]

"Gesundes Kinzigtal Integrated Care" (GKIC) is one of the few population-based integrated care systems in Germany. The external evaluation of GKIC has a modular design and is coordinated by the Department of Medical Sociology of Freiburg University. In this article we will outline the different evaluation modules and address typical problems arising in the evaluation of regional integrated health care systems. One of these problems is the small size of the intervention population in a recently established pilot care system which makes the use of epidemiologic measures (such as incidence figures) difficult. Total population surveys could alleviate this problem, but when the same intervention population is questioned repeatedly for different evaluation modules, the danger of 'over-surveying' arises. This may lead to decreasing participation in further studies. These problems may be managed by using short survey instruments, by conducting surveys successively (not simultaneously) using time shifts, and by analysing claims data from health insurers. Another problem in the evaluation of comparably small systems of care is how to attribute outcomes (effects) to certain health promotion or disease management programmes: Since participants usually take part in more than one programme, the intervention effect is contaminated by multiple programme effects. These will have to be adjusted for in the final evaluation of the data. Finally, randomised controlled study (RCT) designs frequently cannot be used in the evaluation of a region-wide pilot intervention, so the evidence generated by such a (non-randomised) study is weaker than the evidence generated by an RCT.