Language-technology interventions increase interpreter-service utilization and process measures without demonstrated improvement in patient outcomes, and the utilization gains are not shown to persist

Narrative synthesis #

Language-technology interventions reliably move process measures — a mobile app raised weekly interpreter calls (4.3→12.8), a wheeled-tablet initiative raised audio and video interpreter volume from near-zero, and a multilingual chatbot achieved equal engagement — but patient-outcome benefits are not demonstrated and the process gains are not shown to persist. Narang is the only study that looked after the intervention stopped, and the elevated call frequency was not sustained. The chatbot's readmission signal (0% vs 8.3%) is confounded (historical control) and its ED/reoperation results were null; J 2025's outcome nulls sit in a setting where the intervention was barely delivered (interpreter use documented for 62%, concordant documentation for 31%), making them weak evidence of no effect rather than evidence of no effect.

Polarity flag for the maintainer: the fewer-readmissions EVD is wired here as support for the overall pattern, but if the claim title is read as "no outcome benefit," that EVD becomes contradicting evidence — this is a decision to make, not a silent call. Merge candidates (proposal only — do not touch): C-0235ClaimC-0235Initial AI draftA mobile application providing one-touch access to telephone interpreters increases clinician use of interpreter servicesGiving clinicians a mobile application that places over-the-phone interpreter calls at the push of a button increases their use of professional interpreter services for LEP patients, though the increase may not persist o…, C-0061ClaimC-0061Initial AI draftA multilingual chatbot engages LEP patients as effectively as English-primary patientsPatients with limited English proficiency engage with a multilingual SMS chatbot to the same degree as patients whose primary language is English, indicating that a language-concordant automated messaging platform does n…, C-0062ClaimC-0062Initial AI draftEnrolling LEP arthroplasty patients in a multilingual chatbot is associated with fewer hospital readmissionsEnrolling limited-English-proficiency arthroplasty patients in a multilingual perioperative chatbot is associated with fewer hospital readmissions after total joint arthroplasty, relative to comparable LEP patients not e…, C-0063ClaimC-0063Initial AI draftEnrolling LEP arthroplasty patients in a multilingual chatbot reduces emergency department visitsEnrolling limited-English-proficiency arthroplasty patients in a multilingual perioperative chatbot reduces postoperative emergency department visits, relative to comparable LEP patients not enrolled. The single study be…, C-0064ClaimC-0064Initial AI draftEnrolling LEP arthroplasty patients in a multilingual chatbot reduces reoperationsEnrolling limited-English-proficiency arthroplasty patients in a multilingual perioperative chatbot reduces reoperations, relative to comparable LEP patients not enrolled. The one study bearing on this claim (Rainey 2023…, C-0028ClaimC-0028Initial AI draftProfessional interpretation at admission and discharge shortens length of stay for LEP inpatientsWithin a single LEP inpatient population, receiving a professional interpreter at admission (and especially at both admission and discharge) is associated with a significantly shorter hospital stay after adjusting for ag…, C-0048ClaimC-0048Initial AI draftLEP status, especially with inconsistent professional interpreter use, is a risk factor for poorer perioperative care and outcomesCLMs this review thesis generalizes / informs (→ CLM—informs→CLM by syncrelations.py). - C-0030 - C-0034 - C-0040 Luan-Erfe et al.'s PRISMA review of 10 high-quality (Newcastle–Ottawa) studies grades the body of evidence….