DESIGN: A long-term population cohort in Karonga District, northern Malawi.
METHODS: Patients who had completed treatment for laboratory-confirmed TB diagnosed since 1996 were visited annually to record vital status, ART use and screen for TB. Survival analysis estimated the effect of HIV/ART status at completion of treatment on mortality and recurrence. Analyses were stratified by time since treatment completion to estimate the effects on relapse (predominates during first year) and reinfection disease (predominates later).
RESULTS: Among 1133 index TB cases contributing 4353 person-years of follow-up, there were 307 deaths and 103 laboratory-confirmed recurrences (recurrence rate 4.6 per 100 person-years). Half the recurrences occurred in the first year since completing treatment. HIV infection increased the recurrence rate [rate ratio adjusted for age, sex, period and TB type 2.69, 95% confidence interval (CI) 1.69-4.26], but with less effect in the first year (adjusted rate ratio 1.71, 95% CI 0.87-3.35) than subsequently (adjusted rate ratio 4.2, 95% CI 2.16-8.15). Recurrence rates on ART were intermediate between those of HIV-negative individuals and HIV-positive individuals without ART. Compared with HIV-positive individuals without ART, the adjusted rate ratio was 0.74 (95% CI 0.27-2.06) in the first year, and 0.43 (95% CI 0.11-1.73) later.
CONCLUSION: The increased incidence of TB recurrence observed in HIV-positive patients appeared to be reduced by ART. The effects are mostly on later (likely reinfection) disease so the impact of ART on reducing recurrence will be highest in high TB incidence settings.