Assessment of predictors of one-month outcome in head injury in a Nigerian tertiary hospital

Abstract

Background: Current prognostic models for head injury were developed using data mostly from high income countries. A more useful model may be obtained from predictors designed from data obtained from low and middle income countries where trauma is said to be a neglected endemic condition. Methods: This prospective study, after ethical and institutional clearance, examined 19 clinical and Computerized Tomographic (CT) scan parameters of all head injured patients who met the study's inclusion criteria. The significance of these parameters in predicting outcomes at 15 and 30 days post-injury using the Glasgow Outcome Scale (GOS) and mortality/survival was statistically determined using Univariate and Multivariate analysis. Results: One hundred and fifty patients were prospectively recruited. The age range was 0.33 to 82 years. After univariate analysis, the clinical parameters that were significantly predictive of outcomes were pupillary response (p < 0.0001, 95% CI = 0.0653–0.3589), Glasgow Coma Score (p < 0.0001, 95% CI = 15.3340–199.6713) and Age (p = 0.008, 95% CI = 0.0446–0.6142). Among the CT scan parameters, presence of multifocal contusions (p = 0.000), intracerebral haematoma (ICH) (p = 0.01), intraventricular haematoma (IVH) (p = 0.000), effacement of ventricles (p = 0.004) and cisterns (p = 0.000), midline shift (p = 0.012), and crowding of foramen magnum (p = 0.026) were found to be significantly associated with poor outcome. Multivariate analysis showed that GCS (p = 0.006, OR = 12.8171, 95% CI = 2.1013–78.1779), multifocal contusion (p = 0.01, OR = 0.1908, 95% CI = 0.0537–0.6773) and pupillary response (p = 0.046, OR = 0.2256, 95%CI = 0.0445–1.1453) were retained as independent predictors associated with overall outcome. Conclusions: This study suggests that GCS, pupillary response and multifocal contusions may independently predict outcome using GOS in head injury. A prediction model designed on these predictors needs to be considered and evaluated for this environment.