Modification and empirical characterisation of the Lund Outcome Set for Evaluation of Triage

Background Triage systems are used globally to prioritise patients in Emergency Departments (ED), yet no universally accepted standard exists that defines what patients should receive high priority. Each triage system has therefore been assessed in isolation using different methodologies. The Lund Outcome Set for Evaluation of Triage (LOSET), comprising 49 time-critical outcomes, was recently developed through expert consensus to address this knowledge gap, but requires systematic refinement and empirical characterization. Objective The aim of this study was to modify LOSET using definitions allowing it to be implementable using standard clinical data sources available across diverse healthcare systems. Methods We systematically modified LOSET from 49 to 36 outcomes by eliminating redundancies and ensuring implementability with standard coding practices. Implementation was tested using 624,909 ED visits in 2017 and 2018 at eight Swedish EDs. To examine whether the modified LOSET (mLOSET) outcomes captured genuine ED-level time-criticality, we assessed the association between mLOSET status and short-term death or ICU admission. We then characterized associations between mLOSET and commonly used triage outcomes including 7-day and 30-day mortality, hospital admission, and resource utilization using risk ratios. Results mLOSET excluding death/ICU admission strongly predicted death/ICU admission (risk ratio 34.2). mLOSET identified seven times more time-critical patients than death/ICU admission alone (54.7 versus 7.5 per 1,000 visits). mLOSET showed strong alignment with previously used outcomes: mLOSET-positive patients (defined as ≥ 1 mLOSET outcome) had a risk ratio of 40.4 for 7-day mortality, of 13.4 for 30-day mortality, of 3.8 for hospital admission, and 8-fold higher median costs. Diagnostic and intervention codes combined with laboratory values captured 93.5% of mLOSET-positive cases. Conclusion Systematic modification of the expert-derived LOSET yielded an implementable 36-outcome composite that captured time critical conditions using routinely available clinical data. The modified LOSET demonstrated strong alignment with commonly used triage outcomes while markedly expanding case identification, supporting its clinical applicability for triage evaluation.