Epidemiology of pulmonary tuberculosis in France. Can the hospital discharge database be a reliable source of information?

Abstract

ObjectiveIn France, tuberculosis surveillance is based on mandatory notification (MN) of cases. However, the MN does not allow the full description of cases, and underreporting limits data interpretation. Aiming at better describing the cases of tuberculosis, the hospital record database (PMSI) was analyzed. Patients and methodIncident cases of active pulmonary tuberculosis identified in 2010 in France in the PMSI were included and their characteristics were compared with those of the cases identified through the MN. ResultsIn 2010, 5158 incident cases of pulmonary tuberculosis were identified in the PMSI. The mean duration of hospitalization was higher for cases considered contagious — at least one positive test result on pulmonary sample — (22 vs 13 days, P<0.001). Among all cases, 5% were infected by HIV. Death was reported for 4% of cases. The number of pulmonary TB cases reported in the MN was 3781 in 2010. PMSI data by sex, region of residence and month of diagnosis were similar with those of the MN but patients were older in the PMSI (52 vs 47 years, P<0.001). Considering the PMSI as exhaustive, sensitivity of the MN was estimated at 73.3% in 2010. ConclusionPMSI data were compatible with those of the MN and the estimation of the sensitivity was close to other French studies. PMSI can be considered as an interesting tool aiming at improving our knowledge about tuberculosis (TB) cases and strengthening awareness where the sensitivity of the MN is low.