We investigated whether early electroencephalographic features predicted intracranial pressure (ICP), cerebrovascular pressure reactivity, brain tissue oxygenation, and functional outcomes in patients with pediatric traumatic brain injury (TBI). This was a retrospective analysis of a prospective data set of 63 patients with pediatric TBI. Electroencephalographic features were collected in the first 24h of recording to predict values of ICP, pressure reactivity index (PRx), and brain tissue oxygenation (PbtO2) through the initial 7days of critical care monitoring, in addition to Glasgow Outcome Scale Extended-Pediatric Revision (GOSE-Peds) scores at 12months. Electroencephalographic features were averaged over all surface electrodes and included seizures, interictal epileptiform discharges, suppression percentage, complexity, the alpha/delta power ratio, and both absolute asymmetry indices and power in beta (13-20Hz), alpha (8-13Hz), theta (4-7Hz) and delta (0-4Hz) bands. Demographic data and injury severity scores, such as the Glasgow Coma Scale (GCS) and Pediatric Risk of Mortality III (PRISM III) scores, at presentation were also assessed. Univariate and multiple linear regression with guided stepwise variable selection was used to find combinations of risk factors that best explain variability in ICP, PRx, PbtO2, and GOSE-Peds values, and best fit models were applied to pediatric age strata. We hypothesized that suppression percentage and the alpha/delta power ratio in the first 24h of recording predict ICP, PRx, PbtO2, and GOSE-Peds values. Best subset model selection identified that increased suppression percentage and PRISM III scores predicted increased ICP (R2 = 79%, Akaike information criterion [AIC] = 332.30, root mean square error [RMSE] = 6.62), with suppression percentages < 5% (slope = - 5687.0, p = 0.0001) and ≥ 45% (slope = 9825.9, p = 0.0000) being predictive of dose of intracranial hypertension. When accounting for age and GCS score, increased suppression percentage predicted increased PRx values, suggestive of inefficient cerebrovascular pressure reactivity (R2 = 53%, AIC = 3.93, RMSE = 0.23), with suppression percentages ≥ 5% (p = 0.0033) and ≥ 45% (p = 0.0027) being predictive of median PRx values ≥ 0.3. Lower GCS scores, the presence of seizures, and increased suppression percentages each were independently associated with higher GOSE-Peds scores (R2 = 52%, AIC = 194.04, RMSE = 1.58), suggestive of unfavorable outcomes, with suppression percentages ≥ 5% (p = 0.0005) and ≥ 45% (p = 0.0000) being predictive of GOSE-Peds scores ≥ 5. At the univariate level, no electroencephalographic or clinical feature was associated with differences in PbtO2 values. Increased electroencephalographic suppression percentage on the initial day of monitoring may identify patients with pediatric TBI at risk of increased ICP, inefficient cerebrovascular pressure reactivity, and unfavorable outcomes.
Read full abstract