An evaluation of the actual incidence of tuberculosis in French Guiana using a capture-recapture model

In order to estimate the level of under-reporting and to improve estimates of the incidence of tuberculosis (TB) in the vicinity of Cayenne, French Guiana, we performed capture-recapture analysis from 1996 through 2003. We cross-linked data from the Institut Pasteur, the Département d'Information Médicale of Cayenne Hospital, and the Service de Lutte Anti-Tuberculeuse. The estimate of 381 TB cases obtained after matching those three sources was revised to 425 (95% confidence interval: 407, 453) using the capture-recapture model based on sample coverage. The corresponding average annual incidence was 63.1 TB cases per 100,000 population. The evaluated sensitivity of the compulsory notification system was 35.3%, indicating wide under-notification of TB in the vicinity of Cayenne. The estimated coverage reported by the three sources was fairly accurate (i.e. 85.9%), but not sufficient to evaluate the risk of transmission of TB in the Ile-de-Cayenne (Cayenne and its suburbs).