Performance of diagnostic procedures for bubonic plague in endemic settings in Madagascar: a prospective test accuracy sub-study within the IMASOY trial

Objectives We aimed to assess the diagnostic value of a battery of confirmatory tests included in the WHO guidelines for bubonic plague, combined and individually, including culture, PCR and serology, and the on-site and laboratory performance of the F1 antigen-based lateral flow rapid diagnostic test (F1RDT). 

Methods Bubo aspirates from patients (all ages) with suspected bubonic plague enrolled into the IMASOY trial (NCT04110340) underwent routine laboratory diagnosis complemented with serology to measure IgG F1 antibodies from blood samples taken on days 1, 11, and 21. The performance of the F1RDT done on site (on-site F1RDT) and at the Central Laboratory for Plague (reference laboratory F1RDT) was compared against two reference standards: RS1 (culture or PCR-positive), which is used in routine practice, and RS2 (RS1 and serology-positive), including all available confirmatory tests for plague. Results Of 438 suspected cases, 184 (42%) were confirmed by culture or PCR (RS1) and 211 (48%) by culture, PCR, or serology (RS2). PCR identified 179 cases (85%), culture 137 (65%), and serology 197 (93%). The combination of PCR and culture identified 87% of confirmed cases while the remaining 13% were identified by serology only. The sensitivity and specificity of on-site F1RDT were 94% (95% CI, 89.6–97.0) and 74% (95% CI, 68.2–79.3) against RS1 and 89.1% (95% CI, 84.1–93) and 77.5% (95% CI, 71.5–82.8) against RS2. The sensitivity and specificity of reference laboratory F1RDT were 91.8% (95% CI, 86.9–95.4), 97.6% (95% CI, 94.9–99.1) against RS1, and 82.0% (95% CI, 76.1–86.9) and 99.1% (95% CI, 96.9–99.9) against RS2. Conclusions PCR outperforms culture, the previous reference standard. While serology adds diagnostic value, it is impractical for routine use. The F1RDT done on site cannot be relied upon for clinical case management decisions. F1RDT performs better under laboratory conditions and could be implemented in peripheral laboratories for surveillance purposes in resource-constraint settings.