Diagnostic Accuracy for Salmonella enterica Serovar Typhi: A Critical Appraisal of Culture, Serological and Molecular Methods

Abstract

Typhoid fever remains a substantial cause of febrile illness and preventable death in settings with limited water and sanitation infrastructure, yet the laboratory confirmation of infection continues to rest on methods whose measured performance is unstable across populations, specimen types and analytical frameworks. This critical narrative review examines the comparative diagnostic accuracy of culture-based, serological and nucleic-acid-based methods for detecting Salmonella enterica serovar Typhi, and interrogates why estimates of accuracy differ so widely between studies that appear superficially similar. Literature was identified through searches of biomedical bibliographic databases, open scholarly indexes and institutional sources, supplemented by backward and forward citation searching, and was appraised for design adequacy, reference-standard handling, spectrum composition and transparency of reporting. Three findings dominate the synthesis. First, much of the apparent disagreement between studies is generated not by the index tests but by the reference standard: bacteriological isolation from blood recovers only a portion of true infections, so accuracy estimates anchored to it are systematically distorted, and composite reference standards and Bayesian latent class models redistribute rather than eliminate that distortion. Second, the three method classes fail in different and partly complementary ways, with culture limited by bacterial burden and specimen volume, antibody detection limited by background seroprevalence and cross-reactivity, and direct nucleic acid amplification limited by the very low concentration of circulating organisms. Third, the most promising recent gains come from defined-antigen serology, short-enrichment culture coupled to amplification, and resistance-marker detection, although few of these approaches have been evaluated prospectively at the point of care in the populations that would use them. Confidence in any single accuracy figure remains low, and comparative claims should be interpreted against the reference standard and spectrum from which they were derived rather than treated as transportable properties of the assay.