Predictive models for expansive intracranial hematomas occurrence and surgical evacuation outcomes in traumatic brain injury patients at Mulago Hospital.

Abstract

Background Expansive intracranial hematoma (EIH) following traumatic brain injury (TBI) is common problem in Uganda and often leads to prolonged disability and death. Aim:Study assessed the hospital-based prevalence of EIH and developed predictive models for EIH occurrence and surgical evacuation outcomes in TBI patients at Mulago National Referral Hospital. Methods:A prospective cohort study was conducted on adult TBI patients with intracranial hematoma and undergoing surgical evacuation during a period of 18 months. A total of 324 TBI participants were grouped into two arms based on presence or absence of EIH and followed up for 6 months after surgery. Seven variables were captured and categorized as demographical information, clinical assessment, neuroimaging patterns, progression of hematoma over time, type, timing and site of surgery. Data was analyzed using regression models, Kaplan–Meier survival curve and log-rank test at p < 0.05 and a 95% CI. Results:Findings show that 59.3% developed EIH vs 41.7 % had no EIH. Mean (SD) age was 42.3 (17.9) for EIH arm and 30.5 (14.0) years for non EIH arm, respectively. Majority of participant were male (153 (79.7%) in EIH arm and 108 (81.8%) in non EIH, respectively. Patients who had early surgery within 24 hours of accident (17.0%) had a median time to mortality of 2 days and surgery more than 24 hours (83%) had a median time to mortality of 4 days (p=0.004). Patients who had early surgery had a median hospital stay (LOS) similar to late surgery of 2 days (p=0.278). Respective inpatient postoperative prevalences were 10.2% for mortality, 58.0 % for overall complications (i.e., early PTS, brain edema, infection, etc.), 26.2% for unfavorable functional outcome, 46.6% for 18 months overall survival. Models demonstrated good discrimination with AUC for EIH occurrence (0.833) while the average accuracy was 73.4%, intraoperative complications (0.75), mortality (0.77), 6 months unfavorable quality of life (0.73), and unfavorable functional outcomes after neurosurgical evacuation in TBI patients (0.85). Conclusion:EIH is common among TBI patients in Uganda with a prevalence of 59.3%. More than half of EIH patients experience complications, less than half with poor survival outcome postoperatively with 1 in 10 patients dying. To avoid complications and death, EIH patients need to be identified early and monitored closely in this study. Findings from the study still do not offer a better option whether to operate early or late. We developed 5 mathematical models able to predict earlier individualized estimates of EIH occurrence and surgical evacuation outcomes based on preoperative conditions and potentially can be used as an adjunct for decision making and improve resource allocation in the remote setting where several challenges are common.