Ratenbildung bei KV-Daten mit GKV-Versicherten auf Kreisebene – ein empirisches Schätzmodell auf der Basis des Mikrozensus

Abstract

In Germany, data of the statutory health insurance system are used, amongst others, in health monitoring and health care research at the district level. For the calculation of exact ratios, the number of those covered by statutory health insurance is needed as denominator. For some federal states, however, this number is not available on a district level. Therefore, ratios based on statutory health care data are calculated using a surrogate defined in terms of visits to the doctor. This leads to uncertainties that limit small area comparisons. Therefore, the aim of the present study was to develop a superior estimation model for the number of those covered by statutory health insurance on a district level. The proportion of those covered by statutory health insurance in the Bavarian districts is estimated by a multiple linear regression model. The model relates data on determinants of the insurance status (income, proportions of civil servants and of self-employed persons) available on district level to data on the number of those covered by statutory health insurance obtained from microcensus on a regional level. The proportion of those covered by statutory health insurance estimated by this model is compared to the surrogate. As an example for practical application, small area estimations for diabetes prevalence are compared to data provided by the Bavarian Association of Statutory Health Insurance Physicians. The proportion of those covered by the statutory health insurance in the Bavarian districts as estimated by the regression model varies between 74.7 and 91.6%. The difference to the currently used surrogate reaches up to 18.6 percentage points. This is also reflected in treatment prevalence, shown here using the example of diabetes mellitus. The present analysis shows the uncertainties of ratios and consequences for small area comparisons based on statutory healthcare data. Providing valid data for the denominator in accordance with the data transparency regulation in the Social Insurance Code (SGB) V should be attempted.