Accuracy of prospective space–time surveillance in detecting tuberculosis transmission

• We used prospective space–time surveillance to detect tuberculosis (TB) transmission.
• We compared TB genotypes in the space–time clusters to verify recent transmission.
• Many space–time clusters were not due to recent transmission, causing false alarms.
• Prospective space–time surveillance for TB is not useful in public health practice.

To improve detection of tuberculosis transmission, public health can supplement contact tracing with space–time surveillance. However, investigation of space–time clusters not due to transmission (false alarms), may lead to costly, unnecessary interventions. We measured the accuracy of prospective space–time surveillance in detecting tuberculosis transmission, assessing the number of clusters containing transmission and the false alarm rate. We simulated monthly prospective applications of a scan statistic using the home addresses and diagnosis dates of all 1566 culture-positive TB cases reported in Montreal during 1996–2007. We verified transmission within the space–time clusters by analyzing the TB genotype. Over 11.5 years, at 1.3 false alarms per month, we detected 89 transmission chains; at 0.05 false alarms per month we detected 5 transmission chains. We found evidence that prospective space–time surveillance for TB leads to a high false alarm rate, limiting its practical utility in settings with TB epidemiology similar to Montreal.