An evaluation of the completeness and accuracy of active tuberculosis reporting in the United States military

SETTING: Despite the low incidence of tuberculosis (TB) in the United States military, there is uncertainty in the overall reporting and estimates of incidence. OBJECTIVE: To assess TB reporting in the active component US military. DESIGN: TB notification in the US military was compared with three other data sources: laboratory, hospitalization and pharmacy records. Sensitivity and positive predictive value were estimated for all data sources using a gold standard of either a reportable medical event (RME) reported as confirmed or a positive laboratory result for Mycobacterium tuberculosis. Uncorrected and capture-recapture (CR) methods were used to estimate underreporting and completeness of data sources. RESULTS: Completeness of reporting of pulmonary TB cases was estimated as 72.4% uncorrected or 58.3% with CR. Even after correction for possible underreporting, the incidence of active pulmonary TB was only 0.87 per 100000 person-years between 2004 and 2006. CONCLUSION: The rate of active TB in the US military is low. Like civilian surveillance, US military RME surveillance may substantially underreport TB incidence rates. Expanding surveillance to include data sources such as hospitalizations and pharmacy records will increase the number of TB diagnoses at the cost of including many false-positives.