Association of time to craniectomy with survival in patients with severe combat-related brain injury

Abstract

OBJECTIVE In combat and austere environments, evacuation to a location with neurosurgery capability is challenging. A planning target in terms of time to neurosurgery is paramount to inform prepositioning of neurosurgical and transport resources to support a population at risk. This study sought to examine the association of wait time to craniectomy with mortality in patients with severe combat-related brain injury who received decompressive craniectomy. METHODS Patients with combat-related brain injury sustained between 2005 and 2015 who underwent craniectomy at deployed surgical facilities were identified from the Department of Defense Trauma Registry and Joint Trauma System Role 2 Registry. Eligible patients survived transport to a hospital capable of diagnosing the need for craniectomy and performing surgery. Statistical analyses included unadjusted comparisons of postoperative mortality by elapsed time from injury to start of craniectomy, and Cox proportional hazards modeling adjusting for potential confounders. Time from injury to craniectomy was divided into quintiles, and explored in Cox models as a binary variable comparing early versus delayed craniectomy with cutoffs determined by the maximum value of each quintile (quintile 1 vs 2-5, quintiles 1-2 vs 3-5, etc.). Covariates included location of the facility at which the craniectomy was performed (limited-resource role 2 facility vs neurosurgically capable role 3 facility), use of head CT scan, US military status, age, head Abbreviated Injury Scale score, Injury Severity Score, and injury year. To reduce immortal time bias, time from injury to hospital arrival was included as a covariate, entry into the survival analysis cohort was defined as hospital arrival time, and early versus delayed craniectomy was modeled as a time-dependent covariate. Follow-up for survival ended at death, hospital discharge, or hospital day 16, whichever occurred first. RESULTS Of 486 patients identified as having undergone craniectomy, 213 (44%) had complete date/time values. Unadjusted postoperative mortality was 23% for quintile 1 (n = 43, time from injury to start of craniectomy 30-152 minutes); 7% for quintile 2 (n = 42, 154-210 minutes); 7% for quintile 3 (n = 43, 212-320 minutes); 19% for quintile 4 (n = 42, 325-639 minutes); and 14% for quintile 5 (n = 43, 665-3885 minutes). In Cox models adjusted for potential confounders and immortal time bias, postoperative mortality was significantly lower when time to craniectomy was within 5.33 hours of injury (quintiles 1-3) relative to longer delays (quintiles 4-5), with an adjusted hazard ratio of 0.28, 95% CI 0.10-0.76 (p = 0.012). CONCLUSIONS Postoperative mortality was significantly lower when craniectomy was initiated within 5.33 hours of injury. Further research to optimize craniectomy timing and mitigate delays is needed. Functional outcomes should also be evaluated.