Approximating Molecular Sepsis Subtypes Using Bedside Data in Resource-Limited Settings: A Multicenter Analysis From Uganda.

IMPORTANCE

Biologically defined sepsis subtypes have been identified in low- and middle-income countries (LMICs), but limited access to molecular diagnostics constrains broader evaluation and implementation in resource-limited settings.

OBJECTIVES

To determine whether bedside-accessible variables could approximate molecular sepsis subtype assignments among Ugandan adults with sepsis.

DESIGN, SETTING, AND

PARTICIPANTS

Secondary analysis of two prospective observational sepsis cohorts conducted at Tororo General Hospital (Research in the Epidemiology of Severe and Emerging Infections in Uganda-2-Tororo [RESERVE-U-2-TOR]) and Entebbe Regional Referral Hospital (Research in the Epidemiology of Severe and Emerging Infections in Uganda-1-Entebbe [RESERVE-U-1-EBB]), Uganda. Participants were adults 18 years old or older hospitalized with sepsis who underwent transcriptomic (n = 355) and/or proteomic (n = 495) profiling.

MAIN OUTCOMES AND MEASURES: Prespecified clinical and clinico-microbiologic models were evaluated for discrimination and calibration against Uganda-derived transcriptomic and proteomic sepsis subtypes and, secondarily, high-income country (HIC)-derived sepsis subtypes and immune dysregulation frameworks.

RESULTS

In RESERVE-U-2-TOR, clinical models incorporating demographic and physiologic variables showed moderate discrimination for transcriptomic and proteomic subtypes (area under the receiver operating characteristic curve [AUROC], 0.75 [95% CI, 0.69-0.81] and 0.73 [95% CI, 0.66-0.80], respectively), with generally acceptable calibration. Adding rapid HIV, malaria, and tuberculosis test results did not meaningfully improve performance. In RESERVE-U-1-EBB, discrimination was more variable (AUROC range, 0.63-0.75), with generally acceptable calibration. Performance was similarly modest for HIC-derived sepsis subtypes and immune dysregulation axes.

CONCLUSIONS AND RELEVANCE: Bedside-accessible clinical variables, with or without rapid microbiologic testing, only partially approximated molecular sepsis frameworks in Uganda. Scalable molecular biomarker platforms are needed to advance precision medicine for sepsis in LMICs.