Background: The resurgence of tuberculosis in the United States is largely linked to the human immunodeficiency virus (HIV) epidemic. Despite this link, the epidemiology of tuberculosis and preventive strategies in patients infected with HIV are not completely understood.
Objectives: To determine the incidence and predictors of tuberculosis in HIV-infected persons.
Design: Prospective, multicenter cohort study.
Setting: Community-based cohort of persons with and without HIV infection at centers in the eastern, midwestern, and western United States.
Participants: 1130 HIV-seropositive patients without AIDS who were followed for a median of 53 months (814 homosexual men, 261 injection drug users, and 55 women who had acquired HIV through heterosexual contact).
Measurements: Delayed hypersensitivity response to purified protein derivative (PPD) tuberculin and mumps antigen, CD4 T-lymphocyte counts, and frequency of tuberculosis.
Results: 31 HIV-seropositive patients developed tuberculosis (0.7 cases per 100 person-years [95% CI, 0.5 to 1.0]). The most important demographic risk factor was location (adjusted risk ratio for eastern compared with midwestern and western United States, 4.1 [CI, 2.0 to 8.4]). Tuberculosis occurred more frequently in persons with CD4 counts of less than 200 cells/mm3 (1.2 cases per 100 person-years [CI, 0.7 to 1.9]) than in those with higher counts (0.5 cases per 100 person-years [CI, 0.3 to 0.8]). The rate of tuberculosis was highest among tuberculin converters (5.4 cases per 100 person-years [CI, 1.1 to 15.7]), lower among patients who were PPD positive at first testing (4.5 cases per 100 person-years [CI, 1.6 to 9.7]), and lowest among patients who remained PPD negative (0.4 cases per 100 person-years [CI, 0.2 to 0.7]). Tuberculosis was not reported among persons who had PPD reactions of 1 to 4 mm. Compared with that of patients who tested positive for mumps, the risk for tuberculosis of those who tested negative was increased about sevenfold if they were PPD positive (P < 0.03) and fourfold if they were PPD negative (P < 0.02).
Conclusions: Incidence of tuberculosis was higher in the eastern United States, in patients with CD4 counts of less than 200 cells/mm3, and in PPD-positive patients. Analysis of tuberculin reaction size supports the current interpretive criteria of the Centers for Disease Control and Prevention. Nonreactivity to mumps antigen indicated increased risk for tuberculosis independent of PPD response.