Tuberculin skin testing in patients with HIV infection: limited benefit of reduced cutoff values

Clin Infect Dis. 2006 Sep 1;43(5):634-9. doi: 10.1086/506432. Epub 2006 Jul 12.


Background: When determining eligibility for isoniazid preventive therapy of human immunodeficiency virus (HIV)-infected patients, the cutoff value of the tuberculin skin test (TST) is often reduced from an induration of 10 mm in diameter to one of 5 mm in diameter to compensate for loss of sensitivity. The effectiveness of this reduction depends on the underlying mechanism: a gradual decrease in skin test responsiveness with decreasing immunocompetence or an all-or-nothing switch to complete anergy. No published studies have assessed this directly in patients with tuberculosis.

Methods: We performed a cross-sectional study of TST responses and HIV infection among patients with sputum smear-positive pulmonary tuberculosis in 6 hospitals in Tanzania. Skin test anergy was defined as a TST reaction < or =2 mm in diameter.

Results: Of 991 patients with complete results, 451 (45.5%) had HIV infection. Anergy was observed in 111 (24.6%) of 451 HIV-infected patients and 18 (3.3%) of 540 HIV-uninfected patients (P<.001). The reaction size distributions among nonanergic HIV-infected and uninfected patients showed a limited difference (mean diameter +/- standard deviation, 15.9 +/- 5.0 mm and 16.8 +/- 3.8 mm, respectively; P=.048). The sensitivity of the TST among HIV-uninfected patients was 91.1% at a cutoff value of 10 mm and 95.2% at a cutoff value of 5 mm. The sensitivity of the TST among HIV-infected patients was 64.3% at a cutoff value of 10 mm and 71.2% at a cutoff value of 5 mm; the sensitivity of the TST was 67.6% and 74.5%, respectively, after adjustment for tuberculosis-specific anergy.

Conclusion: In subjects with tuberculosis disease and HIV infection, loss of TST sensitivity is predominantly attributable to anergy (i.e., an all-or-nothing phenomenon). The decrease in the proportion of false-negative TST results obtained by reducing the cutoff value from 10 mm to 5 mm is limited.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adolescent
  • Adult
  • Cross-Sectional Studies
  • Female
  • HIV Infections / complications*
  • HIV Infections / epidemiology
  • Humans
  • Male
  • Middle Aged
  • Tanzania / epidemiology
  • Tuberculin Test / standards*
  • Tuberculosis / diagnosis*
  • Tuberculosis / epidemiology