The temporal dynamics of relapse and reinfection tuberculosis after successful treatment: a retrospective cohort study.

ABSTRACT

BACKGROUND: There is increasing evidence from tuberculosis high-burden settings that exogenous reinfection contributes considerably to recurrent disease. However, large longitudinal studies of endogenous reactivation (relapse) and reinfection tuberculosis are lacking. We hypothesize a relationship between relapse vs reinfection and the time between treatment completion and recurrent disease. METHODS: Population-based retrospective cohort study on all smear-positive tuberculosis cases successfully treated between 1996 and 2008 in a suburban setting in Cape Town, South Africa. Inverse gaussian distributions were fitted to observed annual rates of relapse and reinfection, distinguished by DNA fingerprinting of Mycobacterium tuberculosis strains recultured from diagnostic samples. RESULTS: Paired DNA fingerprint data were available for 130 (64%) of 203 recurrent smear-positive tuberculosis cases in the 13-year study period. Reinfection accounted for 66 (51%) of 130 recurrent cases overall, 9 (20%) of 44 recurrent cases within the first year, and 57 (66%) of 86 thereafter (P < .001). The relapse rate peaked at 3.93% (95% confidence interval [CI], 2.35%-5.96%) per annum 0.35 (95% CI, .15-.45) years after treatment completion. The reinfection tuberculosis rate peaked at 1.58% (95% CI, .94%-2.46%) per annum 1.20 (95% CI, .55-1.70) years after completion. CONCLUSIONS: To our knowledge, this is the first study of sufficient size and duration using DNA fingerprinting to investigate tuberculosis relapse and reinfection over a lengthy period. Relapse occurred early after treatment completion, whereas reinfection dominated after 1 year and accounted for at least half of recurrent disease. This temporal relationship may explain the high variability in reinfection observed across smaller studies. We speculate that follow-up time in antituberculosis drug trials should take reinfection into account.