Family Medicine

LLM-assisted CDS safe but no reduction in 14-day treatment failure in Kenyan primary care

In a pragmatic cluster-randomized trial across 16 Kenyan primary care facilities with 9,691 patients, LLM-assisted CDS did not reduce 14-day treatment failure (intervention 2.2% vs control 2.0%). Adjusted odds ratio for the primary outcome was 0.77 (95% CI 0.55–1.08; P = 0.13), indicating no statistically significant benefit. No serious adverse events were judged related to the LLM and independent review found no safety signal. Any clinical benefit, if present, is probably modest and context-dependent. 1

Geographic and socioeconomic variables rarely reported in US cancer RCT publications

Systematic review of 441 US randomized phase II/III cancer trials (2020–2025) found near-universal age and sex reporting but absent reporting of rurality, income, and area deprivation. Race and ethnicity reporting rose over time, with 82.8% reporting multiple race categories and Hispanic ethnicity in 31.8% of trials. No trials reported rurality, income, or area-deprivation; only two reported education and one reported insurance. Limited geographic and socioeconomic reporting constrains assessment of trial generalizability to underserved populations. 2

High interest but limited capacity for caregiver-focused trials in NCORP community practices

Cross-sectional survey of 126 NCORP practices found 84.9% expressed interest in a caregiver-focused trial but only 37.3% met all predefined capacity criteria. Capacity (meeting accrual and implementer criteria) was the sole predictor of interest (adjusted OR 5.79; 95% CI 1.23–27.13). Common barriers included staffing limitations and recruitment challenges related to dyad eligibility. Stakeholders recommended clear dyad eligibility, workflow training, recruitment materials, and resource planning to improve participation. 3