Impact of pulmonary tuberculosis on survival of HIV-infected adults: a prospective epidemiologic study in Uganda

Background: Retrospective cohort studies of tuberculosis suggest that activ e tuberculosis accelerates the progression of HIV infection. The validity o f these findings has been questioned because of their retrospective design, -diverse study populations, variable compliance with anti-tuberculous ther apy and use of anti-retroviral medication. To assess the impact of tubercul osis on survival in HIV infection we performed a prospective study among HI V-infected Ugandan adults with and without tuberculosis. Methods: In a prospective cohort study, 230 patients with HIV-associated tu berculosis and 442 HIV-infected subjects without tuberculosis were followed for a mean duration of 19 months for survival. To assess changes in viral load over 1 year, 20 pairs of tuberculosis cases and controls were selected and matched according to baseline CD4 lymphocyte count, age, sex and tuber culin skin test status. Results: During the follow-up period, 63 out of of 230 tuberculosis cases ( 28%) died compared with 85 out of 442 controls (19%), with a crude risk rat io of 1.4 [95% confidence interval (CI), 1.07-1.87]. Most deaths occurred i n patients with CD4 lymphocyte counts < 200 x 10(6) cells/l at baseline (n = 99) and occurred with similar frequency in the tuberculosis cases (46%) a nd the controls (44%). When the CD4 lymphocyte count was > 200 x 10(6) /l, however, the relative risk of death in HIV-associated tuberculosis was 2.1 (195% CI, 1.27-3.62) compared with subjects without tuberculosis. For subje cts with a CD4 lymphocyte count > 200 X 10(6)/l, the 1-year survival propor tion was slightly lower in the cases than in the controls (0.91 versus 0.96 ), but by 2 years the survival proportion was significantly lower in the ca ses than in the controls (0.84 versus 0.91; P<0.02; log-rank test). For sub jects with a CD4 lymphocyte count of 200 x 10(6) cells/l or fewer, the surv ival proportion at 1 year for the controls was lower than cases (0.59 versu s 0.64), but this difference was not statistically significant (P = 0.53; l ogrank test). After adjusting for age, sex, tuberculin skin test status, CD 4 lymphocyte count, and history of HIV-related infections, the overall rela tive hazard for death associated with tuberculosis was 1.81 (95% CI, 1.24-2 .65). In a nested Cox regression model, the relative hazard for death was 3 .0 (95% CI, 1.62-5.63) for subjects with CD4 lymphocyte counts > 200 x 10(6 )/l and 1.5 (95% CI, 0.99-2.40) for subjects with a CD4 lymphocyte count of 200 x 10(6)/l or fewer. Conclusion: The findings from this prospective study indicate that active t uberculosis exerts its greatest effect on survival in the early stages of H IV infection, when there is a reserve capacity of the host immune response. These observations provide a theoretical basis for the treatment of latent tuberculous infection in HIV-infected persons. (C) 2000 Lippincott William s & Wilkins .