Abstract
BACKGROUND: The importance of remote infection with M.tuberculosis as a cause of tuberculosis disease (TB) is unclear, with limited evidence of impact on TB rates beyond 10 years. Our objective was to assess rates of tuberculosis over 30 years by M.tuberculosis infection status at baseline in Karonga District, Northern Malawi. MATERIALS AND METHODS: Population-based surveys of tuberculin skin testing (TST) from the 1980s were linked with follow-up and TB surveillance in Karonga district. We compared rates of microbiologically-confirmed TB by baseline TST induration <5mm (no evidence of M.tuberculosis infection) and those with baseline TST >17mm (evidence of M.tuberculosis infection), using hazard ratios by time since baseline and attributable risk percent. The attributable risk percent was calculated to estimate the proportion of TB in those infected that can be attributed to that prior infection. We analysed whole genome sequences of M.tuberculosis strains to identify recent transmission. RESULTS: Over 412,959 person-years, 208 incident TB episodes were recorded. Compared to the small induration group, rates of TB were much higher in the first two years in the large induration group, and remained higher to 20 years: age, sex and area-adjusted hazard ratios (HR) 2-9 years post-TST 4.27 (95%CI 2.56-7.11); 10-19 years after TST 2.15 (1.10-4.21); ≥20 years post-TST 1.88 (0.76-4.65). The attributable risk percent of remote infection was 76.6% (60.9-85.9) 2-9 years post-TST, and 53.5% (9.1-76.2) 10-19 years post-TST. Individuals with large TST indurations had higher rates of unique-strain TB (HR adjusted for age, sex and area = HR 6.56 (95% CI 1.96-22.99)), suggesting disease following remote infection, but not of linked-strain TB (recent transmission). CONCLUSIONS: M.tuberculosis infection can increase the risk of TB far beyond 10 years, accounting for a substantial proportion of TB occurring among those remotely infected.