Interpreter Use, Code-Switching, and Patient Outcomes in U.S. Clinics: A Narrative Review for Graduate Medical Education

Language discordance between clinicians and patients in U.S. clinics is a recognized driver of communication-sensitive adverse events, yet the synthesis linking specific linguistic practices to clinical outcomes remains incomplete, and graduate medical education programs lack a unified evidence base on which to build interpreter-mediated competencies. This narrative review evaluates whether interpreter use, translanguaging, and code-switching constitute an interactional pathway through which cultural and explanatory models shape outcomes for limited-English-proficient (LEP) patients, and derives recommendations for graduate medical education. Searches across MEDLINE/PubMed, Embase, CINAHL, PsycINFO, and Scopus covered January 2000–September 2025 and prioritized U.S. immigrant and LEP encounters that linked language practices to assessment, medication decisions, referrals, revisits, adherence, satisfaction, or safety; the Mixed Methods Appraisal Tool informed quality stratification. Across studies, professional interpretation reduced clinically consequential errors (12% vs 22% with ad hoc interpreters; 2% with ≥ 100 training hours). LEP adverse events showed more physical harm than English-proficient events (49.1% vs 29.5%). Verified interpreter coverage at admission and discharge shortened length of stay and lowered 30-day readmissions; chronic conditions reliant on transitional communication showed higher readmissions for LEP patients (heart failure RR 1.29; COPD RR 1.51), while protocolized conditions did not. In-person interpretation was most consistently associated with patient-reported satisfaction and rapport, with video and telephone modalities calibrated to encounter complexity. Reliable, qualified interpretation operates as a patient-safety lever; bundling verified coverage with closed-loop communication at transitions merits programmatic adoption in graduate medical education.