OBJECTIVE: As Mycobacterium tuberculosis isolation rates in tuberculous effusions are relatively low, several biochemical and immunological markers have been proposed to diagnose tuberculous pleurisy including adenosine deaminase (ADA) and interferon-gamma (IFN-gamma). Here we summarise the literature on ADA and IFN-gamma as predictors of tuberculous pleurisy. METHODS: After a systematic review of English language studies, we used summary receiver operating characteristic curve (SROC) analysis to determine the cumulative diagnostic accuracy of both markers and Bayes' theorem to calculate post-test probability of disease in settings with different prevalences of tuberculous pleurisy, assessed and reported the quality of primary studies. RESULTS: From 1978 to November 2000, studies containing sufficient data for the determination of both sensitivity and specificity were 31 on ADA, including 4738 patients, and 13 on IFN-gamma, including 1189 patients. SROC curve yielded a maximum joint sensitivity and specificity of 93% for ADA and 96% for IFN-gamma. In the setting of tuberculous effusion prevalence of 5%, 25% and 85%, post-test probability of a negative ADA test were 0.4%, 2.4% and 24%, and 0.22%, 1.2% and 17% for a negative IFN-gamma test. CONCLUSION: With the caveat that limitations in the design of the studies summarised here may distort estimates of test performance, ADA and IFN-gamma appear to be reasonably accurate at detecting TB pleurisy.
OBJECTIVE: As Mycobacterium tuberculosis isolation rates in tuberculous effusions are relatively low, several biochemical and immunological markers have been proposed to diagnose tuberculous pleurisy including adenosine deaminase (ADA) and interferon-gamma (IFN-gamma). Here we summarise the literature on ADA and IFN-gamma as predictors of tuberculous pleurisy. METHODS: After a systematic review of English language studies, we used summary receiver operating characteristic curve (SROC) analysis to determine the cumulative diagnostic accuracy of both markers and Bayes' theorem to calculate post-test probability of disease in settings with different prevalences of tuberculous pleurisy, assessed and reported the quality of primary studies. RESULTS: From 1978 to November 2000, studies containing sufficient data for the determination of both sensitivity and specificity were 31 on ADA, including 4738 patients, and 13 on IFN-gamma, including 1189 patients. SROC curve yielded a maximum joint sensitivity and specificity of 93% for ADA and 96% for IFN-gamma. In the setting of tuberculous effusion prevalence of 5%, 25% and 85%, post-test probability of a negative ADA test were 0.4%, 2.4% and 24%, and 0.22%, 1.2% and 17% for a negative IFN-gamma test. CONCLUSION: With the caveat that limitations in the design of the studies summarised here may distort estimates of test performance, ADA and IFN-gamma appear to be reasonably accurate at detecting TB pleurisy.