Setting: Self-assessment of tuberculin test results, if accurate, could enhance tuberculosis screening efforts by reducing the need for follow-up visits for skin test reading. We investigated tuberculin test self-assessment in a longitudinal study of tuberculosis infection among drug users.
Objective: To determine the accuracy of tuberculin reaction self-assessment by drug users at high risk for tuberculosis infection.
Design: Two readings were compared of the same skin test, performed 48-72 hours after placement: 1) self-assessment using a simple yes-no approach to induration, versus 2) trained examiner reading. Self-assessments were performed immediately prior to trained examiner readings.
Results: Participants were 137 human immunodeficiency virus (HIV) seropositive and 344 HIV-seronegative current and former drug users. Ten per cent (35/344) of reactions read by participants as 'flat' were read by trained examiners as > or =5 mm (54% of which were > or =10 mm). Twenty-three per cent (19/82) of reactions read by trained examiners as > or =10 mm and 32% (35/110) of reactions read by trained examiners as being > or =5 mm were self-read by participants as 'flat'. Sensitivity (0.68) and specificity (0.83) of self-read tuberculin reactions were sub-optimal. Inter-reader reliability was poorer between participants and trained examiners than between trained examiners.
Conclusion: Self-assessments of tuberculin skin test responses by drug users with or at risk for HIV infection are not reliable.