Prognostic value of SSEP and BAEP in a cohort of deeply sedated mechanically ventilated critically ill patients

Abstract

Somatosensory evoked potentials (SSEP) and brainstem auditory evoked potentials (BAEP) might be useful for predicting outcome in critically ill patients who require deep sedation and mechanical ventilation [1] . However, the impact on outcome of the assessment of brain dysfunction using SSEP and BAEP in this population remains unproven. In this pilot study, we investigated the prognostic value and clinical correlates of the lengthening of inter-peak latencies of SSEP and BAEP components, in a cohort of deeply sedated mechanically ventilated critically ill patients. Median nerve SSEP, BAEP and neurological examination were assessed at bedside, under deep sedation, before day 5 after admission, in intensive care unit (ICU). Interpeak latencies (IPL) of SSEP and BAEP were considered delayed when they were over the mean + 2.5 SD of the ones obtained from our cohort of healthy adults. Primary outcome was mortality in the ICU. Secondary outcome was post-sedation altered mental status defined as the occurrence of delirium or delayed awakening after discontinuation of sedation. Eighty six consecutive patients were enrolled, mean age 61 ± 19 years, and 65% with severe sepsis. Richmond Assessment Sedation Scale (RASS) score at inclusion was -5 (-5 to -4). Thirty seven patients (43%) died in the ICU. A prolonged subcortical-cortical conduction time (P14-N20 IPL SSEP component) i. e. > 5.3 ms, was found an independent predictor of in-ICU mortality after adjustment on total Sequential Organ Failure Assessment score (SOFA) [OR (95% CI) = 2.69 (1.05 to 6.85); P = 0.039]; and after adjustment on non-neurological SOFA [OR (95% CI) = 2.67 (1.05 to 6.81) ; P = 0.040]. Furthermore, delayed P14-N20 IPL SSEP component was associated with absent cough reflex. Finally, there was a trend for association between delayed III-V IPL BAEP component (ponto-mesencephalic conduction time) and post-sedation altered mental status (P = 0.053). These results support the hypothesis of brainstem dysfunction during critically illness [2] , [3] . Studies with larger sample are needed.