759    

Background:  Little data exist on the incidence of new tuberculosis (TB) infection among persons with HIV. Tuberculin skin testing, despite deficiencies in sensitivity and specificity, remains the mainstay of screening for latent TB infection in the United States.

Methods:  The WIHS is a longitudinal study of women with and at risk for HIV in which the baseline prevalence of positive  tuberculin skin testing was 4.7% for HIV-infected women and 15.4% for uninfected women. We performed annual tuberculin skin testings by standard methodology from baseline (1994-1995) through March 31, 2002. Tuberculin skin testing conversion was defined as a new tuberculin skin testing >5-mm induration for HIV-infected and >10-mm or an increase of >10-mm induration for HIV-uninfected women with previous induration.

Results:  A total of 1526 women had >1 repeat tuberculin skin testing during follow-up, with a mean of 850 tests per year. The cumulative incidence of tuberculin skin testing conversion was 4.9% for 1242 HIV-infected women and 6.7% for 284 uninfected women over the 6.5 years of follow-up (RR 0.6; p = 0.04). Yearly incidence of TST conversion ranged from 0.7 to 2.2% for all women and from 0.5% to 1.9% for HIV-infected women. Unadjusted predictors of tuberculin skin testing conversion were African American (AA) ethnicity (5.5 vs 2.9%; RR 1.9; p = 0.02) and income <₁8,000/year (5.2 vs 2.7%; RR 1.9; p = 0.03) Highly active antiretroviral therapy (HAART) use was associated with a lower rate of TST conversion than less potent or no therapy (2.6 vs 5.1%; RR 0.51; p = 0.02). In the adjusted analysis, African American ethnicity (AOR 2.1, 95% CI:  1.1 to 4.2; p = 0.02) and HIV-uninfected status (AOR 2.2; CI:  1.2 to 4.3; p = 0.02) predicted conversion. Among HIV-infected women, tuberculin skin testing converters had higher median CD4⁺ counts than non-converters (397 [range 43 to 1173] vs 355 [0 to 1812]), although 25% of HIV-infected tuberculin skin testing converters had CD4 counts <227. HIV-infected tuberculin skin testing converters had experienced a median (mean) CD4 cell change of 4 (-5) cells/mm since the visit prior to conversion. In the mulitvariable analysis restricted to HIV-infected women, no factors independently predicted tuberculin skin testing conversion. Median (mean) size of a new positive tuberculin skin testing was 10(13) mm for HIV-infected women and 17(18) mm for uninfected women.

Conclusions:   HIV-infected women experienced tuberculin skin testing conversion at a mean rate of 1.0% per year. The demographics and HIV disease status of HIV-infected converters, including the negative association with HAART and lack of CD4 change preceding conversion, imply that many conversion were not secondary to immune restoration but reflect acquisition of TB infection.

Keywords: tuberculosis; skin testing