Post-approval studies (PAS) constitute an important tool in medical devices (MD) assessment usually supported by registries. However, registries are often poorly designed or incomplete. The French health insurance databases are organized since 2003 into a digital data warehouse, the Système national d'information inter-régime de l'assurance maladie (SNIIR-AM), and is the main source of information on patients, hospital activity and associated expenditure. The aim of the study was to determine if these medico-administrative data can be sufficiently relevant to guide a renewal of MD reimbursement in the context of registry failure.

The initial PAS aimed to assess the impact of the guidelines on practice (characteristics of patients, type of stenosis, indications, use of cerebral protection system, surgical procedure) and to determine the 30-day cumulative morbidity and mortality rate of endovascular procedure associated with stenting. Medico-administrative databases provide information on age, sex, symptomatic or asymptomatic stenosis in-hospital mortality and long-term mortality (with a linkage to epidemiological data) and morbidity estimated by ischemic stroke.

The database allowed selection of a cohort of 2,071 patients in whom carotid stenting was performed in 161 centers (40 percent of stents were implanted in 14 centers) with a follow-up of 1 year. Carotid stents were mainly implanted in asymptomatic patients (81.6 percent). Morbi-mortality in symptomatic patients at 30 days (9.2 percent: 5.7 percent stroke and 3.4 percent mortality) was similar with results observed in a French comparative study EVA-3S (9.6 percent: 2.8 percent stroke and 8.8 percent mortality). These data allow the concerned HAS (French Health Authority) committee to renew the reimbursement proposal of these stents.

Medico-administrative database collecting robust criteria can be used to support reimbursement renewal of high risk implantable medical devices. The implementation of other criteria including the disease etiology and the complications imputability may allow to consider the use of these data for non-invasive MD.