HIV and tuberculosis in rural sub-Saharan Africa: a cohort study with two year follow-up.

The objective was to examine the effect of HIV seropositivity on outcomes in tuberculosis (TB) patients in a rural African setting, including rates of TB relapse, other morbid events and mortality. The study setting was a district level hospital in Mzuzu, Malawi. Adult TB patients presenting between November 1991 and May 1993 were included in the study. Treatment was given according to national guidelines. Patients with smear-positive TB received 8 months of rifampicin-containing short-course chemotherapy. Patients with smear-negative or extrapulmonary TB received 12 months of 'standard' treatment. Subjects were followed until they died or until the study concluded (December 1994). There were 225 eligible patients; 187 were tested for HIV and enrolled in the study (66.8% HIV seropositive). Ninety-four percent had complete follow-up information. The cure rate in smear-positive patients who survived to the end of treatment was over 90% and not significantly affected by HIV. Disorders of the gastrointestinal, neurological and dermatological systems were significantly more common in HIV-seropositive patients. HIV had a significant effect on the risk of relapse of TB (hazard ratio [HR] = 10.55 [95% CI 1.38, 80.93]) and on all-cause mortality (HR = 2.81 [95% CI 1.63, 4.64]). Despite high HIV prevalence, high rates of TB cure are achievable using the usual treatment protocols. However, excess TB relapse, other illnesses and mortality associated with HIV seropositivity have serious implications for TB control. There is an urgent need to identify effective intervention strategies aimed at prevention, early diagnosis and treatment of these illnesses.