Early discharges in severely injured patients by ISS score: exploring injury patterns and coding practices

Purpose This study examines the prevalence of early discharge (ED) among patients classified as severely injured in the Dutch National Trauma Registry (DNTR) and evaluates whether identifiable patient-, injury-, and system-related factors are associated with the occurrence of ED in this population. Methods This retrospective cohort study analyzed DNTR data from 2015 to 2022, focusing on severely injured patients (Injury Severity Score [ISS] >= 16). Patients were grouped as early discharge (ED; discharged within 48 h without in-hospital mortality) or non-ED (NED). Patients' characteristics were compared using descriptive statistics, and a mixed-effect model identified factors associated with ED. Results From 2015 to 2022, a total of 37,626 severely injured patients were registered including 1,454 (3.9%) ED patients. This proportion increased from 3.2% in 2015 to 4.8% in 2022. ED was more common in younger patients, males, general practitioner (GP) or self-referred cases, those with severe head injury (Abbreviated Injury Scale [AIS] >= 3) and fewer registered injury codes. In the mixed-effects model, younger age (OR up to 2.0), increasing head injury severity (OR: 1.16 per AIS point), less than 3 registered AIS codes (OR: 0.60), referral by a general practitioner (OR 1.68) or self-referral (OR 1.92), and admission year (OR: 1.08 per year) independently predicted early discharge. Subdural hematoma (small-medium) was the most frequent severe injury code among ED patients. Conclusion The prevalence of early discharge among patients classified as severely injured has increased in the Netherlands overtime. Early discharge was more frequently observed in patients with severe head injuries, those who were self-or general practitioner-referred, and varied across regions. These patterns may reflect differences in clinical practice, trauma system organization, or limitations in AIS/ISS-based injury classification rather than a single cause. Greater attention to coding specificity and injury classification may improve the interpretation of registry-based outcome evaluations. Clinical trial number Not applicable.