Effectiveness of a digital clinical decision support algorithm for guiding antibiotic prescribing in pediatric outpatient care in Rwanda: A pragmatic cluster non-randomized controlled trial

Background

Poor adherence to clinical guidelines and diagnostic uncertainty are key contributors to antibiotic overprescription, accelerating antimicrobial resistance—a major global health threat. We developed ePOCT+, a digital clinical decision support algorithm designed to assist primary care clinicians in the diagnostic process and therapeutic management of acutely ill children under 15 years of age. ePOCT+ integrates oxygen saturation and hemoglobin measurements to help detect severe illnesses, and the C-reactive protein rapid test to help distinguish between bacterial and viral infections. The goal of implementing ePOCT+ was to reduce antibiotic prescriptions without compromising clinical outcomes.

Methods and findings

We tested the effectiveness of ePOCT+ in a pragmatic, open-label, two-arm, parallel-group, cluster non-randomized controlled superiority trial in 32 Rwandan health centers. Sixteen sites implemented ePOCT+, while the remainder provided standard care, enabling the intervention-control comparison. After five months, the control group transitioned to the intervention, enabling a before–after comparison. The initial intervention group continued to use ePOCT+, enabling a longitudinal assessment. A total of 59,921 outpatient consultations were enrolled between 1 December 2021 and 30 April 2023; 47,822 new consultations were analyzed, after excluding referrals and re-attendance visits. The uptake of the intervention (percentage of registered cases managed using ePOCT+) was 75% on average.

In the per-protocol (PP) analysis, ePOCT+ use was associated with a significant reduction in antibiotic prescription rates from 70.5% to 24.5% in the intervention–control comparison (absolute difference −46.0, 95% confidence interval (CI) −52.5 to −39.5) and to 27.5% in the before–after comparison (−43.0, 95% CI: −52.5 to −39.5). The corresponding reductions were smaller in the intention-to-treat (ITT) analysis: −36.7 (95% CI: −42.1 to −31.3) in the intervention-control comparison and −26.7 (95% CI: −32.2 to −21.4) in the before-after comparison. Once reduced, prescription rates stayed low (25-40%) throughout the intervention period. Nonetheless, approximately 25% of prescribed antibiotics were not recommended by ePOCT+, indicating that with additional training and mentorship of clinicians, a bigger impact could be achieved on reducing unnecessary antibiotic prescriptions. Notably, only 6% of patients took additional medications not prescribed during the initial consultation, equivalent in both study arms.

Reduced antibiotic prescription rates did not increase the risk of clinical failure seven days after the initial consultation. In the PP analysis, adjusted relative risk [aRR] was 1.07, 95% CI 0.97–1.18 in the intervention-control comparison and 1.00, 0.92–1.10 in the before-after comparison. In the ITT analysis, aRRs were slightly elevated in both comparisons but remained within the predefined non-inferiority margin (upper 95% CI < 1.3): 1.14 (95% CI: 1.02–1.26) in the intervention-control comparison and 1.09 (95% CI: 1.02–1.18 in the before-after comparison.

Conclusions

In this pragmatic trial, the implementation of ePOCT+ substantially reduced antibiotic prescribing in pediatric outpatient care in Rwanda, without compromising clinical recovery. To reach meaningful scale, integrating into Rwanda’s national electronic medical record platform and ensuring that all clinical and reporting functions can be performed through a unified system, is a critical next step. Such integration could streamline service delivery, improve data quality, and promote more consistent, evidence-based care at scale.

Trial registration

Clinicaltrials.gov NCT05108831