Description:
Health services research, Epidemiology, Cardiovascular diseases, Health insurance data, administrative data, Germany.
Routine data of statutory health insurance providers (SHI) are increasingly used in health services research. The size, the longitudinal character of the data, the largely unselected populations as well as the possibility to evaluate routine care render SHI routine data an attractive data source for health services research. Since the data is generated for reimbursement processes, SHI routine data is limited with regard to the range and depth of information. Furthermore, the validity of the data for research purposes often is unclear. The aim of this thesis is to discuss the methodological challenges regarding the use of SHI routine data in health services research arising from the limitations inherent in the data and to further outline approaches to overcome these limitations. One of the biggest challenges regarding the use of SHI routine data in health services research is the fact that information on disease severity which may represent a confounder in analytical health services research is largely lacking. In this context, advantages and disadvantages of different propensity score methods regarding their potential for confounder control are discussed in the thesis. A further challenge is related to the fact that studies are often based on data from a single or few SHIs. Since insurants covered by the various SHIs differ with regard to sociodemographic characteristics which, in turn, can be related to the morbidity risk or the probability of receiving certain health services, the external validity of study results might be limited. Concerning this matter, weighting approaches to account for these differences are discussed. To evaluate the validity of SHI routine data with an external gold standard or to enrich SHI routine data with external data due to the limited range of information, a patient-individual record linkage with another data source is required. However, such a record linkage often cannot be realized, since processing person-identifying information requires meeting very high standards of data protection. In this thesis, methods allowing a record linkage of SHI routine data with another data source based on person-identifying information are discussed. Finally, there is an outlook regarding the availability of SHI routine data for research purposes.