ABSTRACT
An advanced and integrative information technology (IT)-landscape is needed for optimal support of future processes in health-care, including health services research. Most researches in the primary care sector are based on data collected for reimbursement. The aim of this study is to show the limits and options of secondary analysis based on data that was exported via the "Behandlungsdatentransfer" (treatment data transport) BDT-interface in the software systems of German general practitioners and afterwards prepared for further research in SPSS. From the middle of 2005 to the end of 2007 all 168 teaching practices of the Hannover Medical School (MHH) were invited to join the study. Finally routine data from 28 practices could be collected successfully. The data from 139 other practices which had been collected for the project "Health Care in Practice" ("Medizinische Versorgung in der Praxis" - MedViP) was also added to the pool. The process of data preparation included a complete cycle from data collection, merging the data in a relational database system, via statistics and analysis to publishing and generating a feedback report for the participating practices. During the whole study the limits and options of this method were systematically identified. Of the 168 practices, 68 (40.5%) were interested to participate. From 28 (16.7%) physicians the data could be exported from their software systems. In 15 (8.9%) cases no collection was possible due to technical and in 26 (15.5%) to administrative reasons. The method of data extraction varied, as the BDT-interface was differently implemented by the software companies. Together with the MedViP data, the database at the MHH now consists of 167 practices with 974 304 patients and 12 555 943 treatments. For 44.1% of the 11 497 899 prescription entries an anatomic therapeutic chemical (ATC) code could be applied, by matching the entries to the master data from the Scientific Institute of Local Health-Care Funds ("Wissenschaftliches Instituts der Ortskrankenkassen" - WIdO). Periodically consistent sets of SPSS files could successfully be created for further research and feedback reports for the participating practices were generated as portable document format (PDF) files. The BDT-interface seems quite out of date, but can still reveal interesting information, especially on data about medical treatments and findings. Much of the data is contained in fields based on free text, which makes analysis difficult. Coded information, like agents, as ATC, could partially be extracted from the data, which afterwards was easy to prepare for further research. Quality and content of the data depend mainly on the data enterer, the physicians and their practice staff. Future research could be improved by more classified and coded data, which would better be transported through an interface more advanced than BDT.