Setting: Enhanced tuberculosis (TB) case finding using detection rats in Tanzania.
Objectives: To assess the diagnostic accuracy of detection rats compared with culture and Xpert® MTB/RIF, and to compare enhanced case-finding algorithms using rats in smear-negative presumptive TB patients.
Design: A fully paired diagnostic accuracy study in which sputum of new adult presumptive TB patients in Tanzania was tested using smear microscopy, 11 detection rats, culture and Xpert.
Results: Of 771 eligible participants, 345 (45%) were culture-positive for Mycobacterium tuberculosis, and 264 (34%) were human immunodeficiency virus (HIV) positive. The sensitivity of the detection rats was up to 75.1% (95%CI 70.1-79.5) when compared with culture, and up to 81.8% (95%CI 76.0-86.5) when compared with Xpert, which was statistically significantly higher than the sensitivity of smear microscopy. Corresponding specificity was 40.6% (95%CI 35.9-45.5) compared with culture. The accuracy of rat detection was independent of HIV status. Using rats for triage, followed by Xpert, would result in a statistically higher yield than rats followed by light-emitting diode fluorescence microscopy, whereas the number of false-positives would be significantly lower than when using Xpert alone.
Conclusion: Although detection rats did not meet the accuracy criteria as standalone diagnostic or triage testing for presumptive TB, they have additive value as a triage test for enhanced case finding among smear-negative TB patients if more advanced diagnostics are not available.