Setting: Sotiria Chest Diseases Hospital (SCDH), a referral hospital in Athens, Greece, 2012.
Objective: To assess the completeness of the mandatory notification system for tuberculosis (TB) at the SCDH, and compare the observed and estimated annual incidence rates.
Design: Record linkage and the capture-recapture method were applied. Data sources were the registers from the national mandatory notification register (Hellenic Centre for Disease Control and Prevention [HCDCP]), the National Reference Laboratory for Mycobacteria (NRLM) and SCDH records. The log-linear model with the lowest Akaike information criterion was selected as the most valid statistical model.
Results: The observed and estimated TB under-reporting rates at the national level were respectively 55% (95%CI 49-60) and 75% (95%CI 71-78). The observed completeness of the HCDCP, NRLM and SCDH registers were respectively 45% (95%CI 40-51), 66% (95%CI 61-71) and 36.5% (95%CI 31-42). The estimated TB incidence rate was 15 cases per 100 000 (range 13-19/100 000), compared to the 4.9/100 000 rate officially notified.
Conclusion: Adult TB incidence has been largely underestimated, and the TB burden is likely to be much higher than officially notified in our setting. A thorough review of the notification system should be carried out. The implementation of a network-based notification system and retraining of all relevant personnel is advised.