Work

Development and validation of a nursing-oriented triage decision tool for prehospital trauma patients in China: a retrospective cohort study

Year: 2026

Type: article

Abstract: In China, prehospital emergency medical services (EMS) frequently employ nurses as frontline triage personnel, a practice that differs from EMS systems in many Western countries where paramedics typically fulfill this role. However, existing triage tools such as ATLS and CRAMS were not specifically designed for nursing-assessable variables in the Chinese prehospital context. This study aimed to develop and internally validate a nursing-oriented triage tool for prehospital trauma patients in a Chinese metropolitan setting. This retrospective cohort study was conducted at 12 EMS stations affiliated with three tertiary hospitals in Sichuan Province, China (January 2020–December 2022). The study followed STROBE and TRIPOD-AI reporting guidelines. A total of 2,864 patients were randomly split into development ( n = 1,909, 66.7%) and validation ( n = 955, 33.3%) sets using stratified random sampling. Variable selection employed LASSO (Least Absolute Shrinkage and Selection Operator) penalized regression. A Trauma Triage Nursing Score (TTNS) and a decision tree (rpart) were developed. The primary outcome was defined as 24-hour mortality, Injury Severity Score (ISS) ≥ 16, or need for emergent surgery — all surrogates measurable within the first 24 h of hospital arrival to avoid incorporation bias from post-arrival data. Internal validation assessed discrimination (AUROC, AUCPR), calibration (Brier score, calibration plot), and clinical utility. Head-to-head comparisons with ATLS and CRAMS were performed on the same validation dataset. The TTNS included five variables: GCS score (0–4 points), SBP < 90 mmHg (3 points), active hemorrhage (3 points), fatal trauma type (2 points), and age ≥ 65 years (1 point), with a total score range of 0–13. In the validation set, the TTNS demonstrated good discriminative ability (AUROC = 0.883, 95% CI: 0.857–0.909) and acceptable calibration (Brier score = 0.089). The AUCPR was 0.612. At the optimal cutoff (≥ 5 points), sensitivity was 90.4%, specificity 83.7%, PPV 82.5%, and NPV 92.0%. In head-to-head comparisons on the same validation dataset, the TTNS outperformed both ATLS (AUROC = 0.824, P < 0.001) and CRAMS (AUROC = 0.801, P < 0.001). The decision tree achieved sensitivity of 91.2% and specificity of 83.7%. The TTNS and decision tree showed good predictive performance for identifying trauma patients requiring Level I trauma center transport in a Chinese prehospital EMS setting. These findings are limited to internal validation in a single metropolitan region with a nurse-centric EMS model. Prospective external validation is warranted before clinical implementation. Not applicable.

Source: BMC Emergency Medicine

DOI: 10.1186/s12873-026-01762-x

Authors: Yi Chen, Seidu A. Richard, Zhigang Lan

Institutions: Sichuan University, West China Hospital of Sichuan University, University of Technology and Applied Sciences

Language: English

Topic: Trauma and Emergency Care Studies

Subfield: Emergency Medicine

Field: Medicine

Domain: Health Sciences

SDG: Good health and well-being

Open Access status: gold

Estimated APC paid: $2,790