Impact de l’estimation de la population affiliée à l’assurance-maladie sur le calcul de la prévalence pharmacologique du diabète à l’échelle régionale et territoriale

BackgroundData derived from Health Insurance databases are very useful for health observation. These data are however still underused, particularly for small local areas. This may be partly explained by the lack of reliable data on the number of insured people. Recent simplification of the Répertoire national interrégimes de l’assurance-maladie (RNIAM) indicator (French register of health insurance) gives the opportunity to improve the usefulness of these databases. This indicator specifies the beneficiaries’ status towards the General Health Insurance Fund. This study aimed to select the population of beneficiaries, which could be most adequately used to calculate health indicators based on these data.MethodsData were collected from the outpatient database of the Southeastern France General Health Fund. We compared beneficiaries’ characteristics according to the RNIAM indicator, calculated the annual unadjusted and age-adjusted regional and local prevalence of diabetes mellitus in two different populations: the whole initial beneficiaries database, and the population of “effective” beneficiaries (persons whose reimbursements were effectively managed by the General Health Insurance).ResultsThe initial database included 4,817,871 beneficiaries. Almost 80% were in the “effective” population, 14% had left the General Health Insurance, or Southeastern France, and 4% were doubles. The annual unadjusted prevalence of diabetes mellitus was 3.31% in the initial database, and more than 20% higher when calculated among “effective” beneficiaries. Impact on aged-adjusted prevalence was less important (+9% at regional level), but the increase varied from 6 to 42% for the small local areas. Impact was much higher on age and gender specific rates.ConclusionWhen Health Insurance databases are used to calculate health indicators at various geographical levels, only “effective” beneficiaries should be selected. The methodology for determining health indicators might be improved by updating databases (e.g. the date of the RNIAM indicator last update should be added).