Predicting referral need for febrile children in low-resource community settings in South and Southeast Asia.

Summary

Across five Asian countries, simple clinical models for febrile children outperformed WHO danger signs for predicting severe disease within 2 days. Adding pulse oximetry or sTREM1 further increased sensitivity to ~89% while reducing referral rates threefold, with favorable cost-effectiveness.

Key Findings

• A simple clinical-parameter model achieved sensitivity 74.7% and specificity 99.1%, outperforming WHO danger signs (sensitivity 55.5%, specificity 82.6%).
• Adding pulse oximetry or sTREM1 increased sensitivity to 88.9% and 89.2%, respectively, with a threefold reduction in referral rates.
• Cost-effectiveness favored pulse oximetry (ICER $26.28) and sTREM1 (ICER $196.46), supporting scalable implementation.

Clinical Implications

Implementing pulse oximetry–enhanced triage algorithms can reduce unnecessary referrals while improving early identification of children needing organ support, informing community-level sepsis pathways.

Limitations

• Hospital-based enrollment may limit generalizability to true community settings
• Biomarker availability (sTREM1) and implementation logistics require feasibility testing

Future Directions

Prospective community trials to assess real-world impact on mortality, referral workflows, and antimicrobial stewardship; head-to-head implementation comparing pulse oximetry versus biomarker-augmented pathways.