Laboratory-based evaluation of 19 commercially available rapid diagnostic tests for tuberculosis

Overview

Accurate and prompt tuberculosis (TB) diagnosis is critical to disease control. Simple user-friendly and affordable detection tools could save lives and reduce overall costs borne by patients and health systems. Rapid, accessible serologic tests for tuberculosis are on the market, largely in developing countries, but little reliable information about their content and performance is available. Therefore, TB case detection remains dependent upon sputum smear microscopy, radiography and clinical symptomatology. In recent years, remarkable efforts have been made globally to improve access to, and the quality of, tuberculosis diagnostic services and to identify promising new diagnostic tools. Global case notification rates have increased and more than 15 diagnostic candidates are in the pipeline. However, still less than 20% of TB patients receive a microbiologically confirmed diagnosis. To this end, in cooperation with rapid TB test manufacturers, WHO/TDR sponsored an evaluation of commercially available rapid TB tests to assess their performance, reproducibility and operational characteristics and to identify promising candidates.

Using 355 well-characterized archived serum samples, 19 rapid TB tests were evaluated at the Prince Leopold Institute of Tropical Medicine Mycobacteriology Unit. The sensitivity of these rapid tests ranged from 0.97% to 59.7%; specificity ranged from 53% to 98.7%, compared against a combined reference standard of mycobacterial culture and clinical follow-up. In general, tests with high specificity (>95%) had very low sensitivity (0.97–21%). Test performance was poorer in patients with sputum smear-negative TB (sensitivity & specificity: p=0.0006) and in HIV-positive patients (sensitivity: p=<0.0001, specificity: p=0.44). The average difference in test sensitivity between the HIV-negative (n=198) and the HIV-positive population (n=157) was +22%; the maximum difference was +43%. Several tests showed high reliability; the average inter-reader variability kappa was 0.77 and the overall lot–to–lot and run–to–run variability ranges were 0–25% and 0–26%, respectively. Twelve of the tests (63%) were rated as very easy to use and therefore appropriate for use in primary health care settings in developing countries. None of the assays performed well enough to replace microscopy. However, smear microscopy combined with most rapid tests improved overall diagnostic sensitivity from 75% (smear alone) up to 89% (smear plus rapid test). This gain is equivalent to the detection of 57% (29/51) of the smear negative, culture positive TB cases but concomitantly yielded an unacceptable overall false positive rate of 42% (63/149).