Systematic review and meta-analysis of diagnostic tests for diagnosis of melioidosis

Melioidosis is an emerging tropical disease caused by B. pseudomallei that can rapidly prove fatal and require prompt and appropriate antibiotic treatment. Diagnosis currently relies on culture; however, this delays appropriate antibiotics and contributes to mortality as results can take up to one week or more. Several non-culture based diagnostic tests are available; however, their role remains a point of contention. This review was performed to characterise the accuracy of various diagnostic tests.

Medline/Pubmed, CINAHL, Informit, Scopus, The Cochrane Library, Web of Science and Embase databases were searched from inception to April 2020. Clinical trials investigating diagnostic tests capable of providing results in ≤48 hours using samples from suspected human cases; with bacterial culture as the reference test, and reporting sensitivity and specificity were eligible for inclusion. Data was pooled using bivariate analysis for diagnostic tests reported in ≥4 studies.

22 publications comprising 10963 individual tests were included. Meta-analysis was able to be performed for immunofluorescence assay (sensitivity 63.8% [95% CI, 45.6-78.7%]; specificity 99.4% [95% CI, 97.2-99.9%]), polymerase chain reaction (sensitivity 77.1 [95% CI, 20.8%-97.8%]; specificity 99.8 [95% CI, 91.6%-100.0%]) and lateral flow immunoassay (sensitivity 58.2% [95% CI, 34.1%-78.9%]; specificity 95.0% [95% CI, 93.3%-96.3%]). Measures of sensitivity were overall similar although specificity of immunofluorescence assay was statistically superior to lateral flow immunoassay. There was a trend for reduced sensitivity of direct detection methods applied to blood samples compared to other sample types, although statistically insignificant. Serological methods were unable to be meta-analysed due to an insufficient number of studies, but their sensitivities were generally higher than direct detection methods (median 84% [IQR 71.5-88%] vs 51% [IQR 39-79%]), however they lacked specificity compared to direct detection (median 82% [IQR 70-86%] vs 98% [IQR 95-100%]).

Overall, no method showed sensitivity and specificity which would allow it to substitute culture. Serological tests may play a role in ruling out infection in endemic regions given their higher sensitivity, with direct detection methods being used for diagnostic confirmation. Further research into cost-effectiveness and implementation studies are required before diagnostic tests can be introduced clinically in the detection of melioidosis.