Background: HIV-infected children are at high risk of developing tuberculosis after infection by Mycobacterium tuberculosis. Emphasis is placed on tuberculin skin testing (TST) for diagnosing tuberculosis in children; however, its value in HIV-infected children is controversial.
Objectives: To determine whether concurrent antigen testing and/or CD4+ lymphocyte counts help in the interpretation of the TST in children with tuberculosis.
Methods: Children eligible for the study were diagnosed as having tuberculosis on clinical criteria. CD4+ lymphocyte counts and delayed-type hypersensitivity (DTH) test, using the CMI Multitest were performed when tuberculosis was diagnosed.
Results: One hundred thirty children were enrolled. Tuberculin reactivity was lower in HIV-infected children at all cutoff levels than in HIV-uninfected children (P < 0.0001). The positive predictive value of normal CD4+ lymphocyte counts in predicting tuberculin reactions of > or =5 mm (in HIV-1-infected) and > or =10 mm (in HIV-uninfected patients) were 50 and 80.3%, respectively (P < 0.0001). An intact DTH reaction to the CMI Multitest in predicting reactions of > or =5 mm and > or =10 mm to tuberculin in HIV-infected and -uninfected children were 55 and 76%, respectively (P < 0.001). Kwashiorkor was responsible for 53.3% of false-negative TST in HIV-uninfected children with normal CD4+ lymphocyte counts.
Conclusion: TST is of limited value as an adjunct in diagnosing tuberculosis in HIV-infected children. CD4+ lymphocyte counts and concurrent DTH testing are not useful for predicting tuberculin reactivity in HIV-infected patients with tuberculosis.