Management and outcome of patients with traumatic brain injury treated in three major Nordic intensive care units: a comparative cohort study

BACKGROUND: Traumatic brain injury (TBI) is a major cause of death and disability, with large variations in management strategies and outcomes across countries. Nordic countries are similar in demographics, but differences in treatment strategies likely exist and outcome differences have been reported. We aimed to assess temporal differences in treatments and outcomes in three large Nordic centers, with a hypothesis that outcomes, after adjustment for injury severity, are generally similar and have improved over time. MATERIALS AND METHODS: We performed an international, multicenter, retrospective, observational registry study including consecutive intensive care unit (ICU) managed TBI patients from three of the largest university hospitals in the Nordics (Helsinki (HUS), Finland [2005-2020]; Stockholm (KUH), Sweden [2005-2022]; Oslo (OUH), Norway [2015-2022]). Our primary outcome was 6-12-month unfavorable Glasgow Outcome Scale (GOS, 1-3) and secondary outcome was 30-day mortality. We compared outcomes after adjusting for TBI outcome predictors: age, pupil responsiveness and admission Glasgow Coma Scale (GCS) over time. Other known outcome predictors like computerized tomography (CT) scores (Marshall CT classification and Rotterdam CT score) were collected if available, and intracranial pressure monitoring (ICP) for GCS 3-8 patients was specifically analyzed. RESULTS: In total, 5,970 patients were included (HUS = 2,769, KUH = 1,390, OUH = 1,811). Six-months crude unfavorable outcome was 43% and predicted unfavorable outcome was 44%. Crude 30-day mortality was 17% and predicted 30-day mortality was 18%. Of the included patients, 72% had a moderate-to-severe TBI (GCS 3-12) and 54% underwent some form of neurosurgical intervention. Model application diminished differences between centers, across both outcome measures. Minimal variation in outcomes between hospitals was observed, with a hospital-level variance after applying the model of 0.02 (SD 0.15) for unfavorable outcome and 0.06 (SD 0.24) for 30-day mortality. Overall, no significant temporal trend in predicted unfavorable outcome was observed (OR per year 0.95, 95% CI 0.88-1.03, p = 0.20). Variations in ICP monitoring, but also caseload, were observed. CONCLUSION: Differences in patient case-mix and caseload between hospitals were associated with variation in treatment strategies and crude outcomes. However, after adjusting for TBI outcome predictors, no meaningful between-hospital variation in outcomes was observed, and no temporal trends in outcome found over the study period.