[Feasibility and practical value of statistical matching of a general practice database and a health insurance database applied to diabetes and hypertension]

Abstract

Objectives: Public Health actors in France are striving to improve the use of national databases for public health and research. The main objective of this project was to develop a research tool in ambulatory care by matching medical data and reimbursement data.

Methods: Data sources were the health insurance database (SNIIRAM) and the General Practice Observatory (OMG) database. The SNIIRAM is a national medical and administrative database comprising data used in healthcare reimbursement. The OMG is a medical database on ambulatory care recording presenting complaints called "Results of Consultation" (RC). Based on data for patients who consulted one of the 30 general practitioners selected in 2008, we performed a probabilistic matching of the two databases.

Results: The linkage procedure allowed matching of 89,211 consultations or doctor visits and 29,088 patients. Comparison of long-term diseases (ALD) and RC showed that 94% of patients with diabetes as ALD had at least one RC coded as diabetes during the year, but only 65% of patients with one RC coded as diabetes were reported as ALD for this disease. Matching of the databases identified 12% of diabetic patients without antidiabetic treatment and without ALD for this affection; these patients were therefore not identifiable in the SNIIRAM database.

Conclusion: This study describes an innovative database matching methodology. It also illustrates the contribution of this model of matched data in terms of targeting populations at risk. Other approaches to analysis of comorbidities, medical practices and care pathways could be proposed.