Enhancing meningitis diagnosis: a diagnostic stewardship workflow in a resource-limited setting

Central nervous system (CNS) infections cause high morbidity and mortality, especially in low- and middle-income countries, where the burden is unevenly distributed. This study aims to develop a diagnostic stewardship workflow (DSW) to improve diagnostic efficiency and increase pathogen detection in resource-limited settings. We conducted a prospective cross-sectional study in adults with suspected meningitis in Lima, Peru. Cerebrospinal fluid specimens were processed in parallel by the hospital laboratory (standard of care [SoC]) and an independent reference laboratory executing the DSW, enabling within-patient head-to-head comparison. The DSW was compared with the SoC using McNemar's and Cochran's Q tests with Holm-Bonferroni correction. Among 103 adults (median age: 39 years, 36.9% HIV-positive), CNS infection was confirmed in 44 (42.7%), and a non-infectious etiology was diagnosed in 20 (19.4%). The DSW achieved significantly higher pathogen detection rates (97.7%; 43/44 confirmed cases). Incremental yield over SoC was etiology-specific: 0 pp for cryptococcosis, +36 pp for tuberculous meningitis, +67 pp for bacterial meningitis, and +100 pp for viral encephalitis. A deterministic cost simulation showed that applying the DSW with sequential stopping rules would reduce per-cohort diagnostic expenditure by approximately 61% compared with the all-tests-to-all strategy applied in this study. A DSW substantially increases pathogen detection compared to standard diagnostic procedures, with the largest incremental gains for viral and difficult-to-culture bacterial pathogens, while reducing unnecessary testing and diagnostic costs and providing practical guidance to clinicians to streamline the diagnostic process.

We developed a meningitis diagnostic stewardship workflow using clinical information and initial test results to guide decision-making and streamline pathogen detection. This approach increased diagnostic yield, has the potential to reduce unnecessary testing, and may support more efficient diagnostic care in resource-limited settings.