摘要:
Prognostic models for severe traumatic brain injury (sTBI) include initial Glasgow Coma Scale (GCS) score, pupillary response, and computerized tomography results in their computations; however, their predictive value is limited by confounding variables contributing to outcome and poor inter-rater reliability. Use of somatosensory evoked potentials (SSEPs) testing in the acute phase may provide additional objective data to improve prognostication in *** an Institutional Review Board-approved protocol, prospective clinical and outcome data were collected from all severe TBI patients (GCS = 8) admitted to UPMC Presbyterian Hospital over a 5-yr period. SSEPs were obtained on post-trauma day 5 in patients who were not following commands. N9, N13, P14, N20, and P30 latencies are measured; central conduction time was defined as the delay between the P14 peak recorded at the C-2 electrode and the N20 peak recorded at the somatosensory cortex. Peak-to-peak amplitudes measured were of N20/P30 using Fz reference. Criterion for pathologic SSEP event was defined as either the reduction of the N20 amplitude below 50% of baseline or the in-crease of latency between P14 and N20 by 10%. SSEPs were graded as bilaterally absent, unilateral absence, bilateral delay > 23 ms, unilateral delay with normal, and bilateral normal latency. Functional outcome was prospectively collected at 12-mo follow-up, and included mortality and neurologic status as assessed by *** total of 145 patients were available for analysis. Median GCS was 6 (IQR 2), while median radiographic injury was a Marshall score 3 (IQR 2). Mean age was 38 ± 16 yr. Univariable ordinal logistic regression analysis of SSEP demonstrated significant ordinal prediction of outcome according to GOS; favorable outcome correlated with higher SSEP grading [OR = 1.66, 95% CI (1.29-2.13), P = .001]. A dichotomized model grouping higher/lower SSEP grading significantly predicted favorable/unfavorable outcomes, resp