Emergency medical services interval and mortality in significant head injury: a retrospective cohort study

OBJECTIVES: Recent interest has focused on reorganizing emergency medical services (EMS) for English traumatic brain injury (TBI) patients, with bypass of nonspecialist hospitals and direct transportation to distant neuroscience centres. This may expedite specialist neurocritical care and neurosurgical interventions, but risks harms from prehospital deterioration and delayed resuscitation. We therefore aimed to investigate the effect of EMS interval on outcome following head injury. METHODS: We performed a multicentre cohort study examining adult patients with significant TBI (head region abbreviated injury scale ≥3) enrolled in the Trauma Audit and Research Network trauma registry between 2005 and 2011. The association between EMS interval and mortality in patients directly admitted to specialist neuroscience centres was explored using bivariate and multivariate logistic regression and propensity score matching analyses. RESULTS: In all, 7149 eligible patients presented directly to specialist neuroscience centres during the study period. Adjusted odds ratios for mortality showed no association between EMS interval and mortality, varying from 0.46 (95% confidence interval 0.1-2.6) for EMS intervals under 20 min to 0.67 (95% confidence interval 0.4-1.2) for EMS intervals more than 120 min (reference EMS interval 40-60 min). This lack of association was also observed following matching using propensity scores, with no significant difference apparent in mortality between EMS intervals less than 60 min and more than 60 min (17.85 vs. 17.0%, P=0.826). These results were unaffected in sensitivity analyses examining missing covariate data or unmeasured outcomes. CONCLUSION: The lack of observed association between EMS interval and mortality may not preclude bypass of significant TBI patients, with concomitantly prolonged primary transfers from the scene of injury to distant specialist centres. However, given the limitations of registry data, our results should be interpreted with caution. Language: en