Improved outcome prediction in unconscious cardiac arrest survivors with sensory evoked potentials compared with...

To compare the prognostic ability of sensory evoked potentials in cardiac arrest survivors with the outcome predicted by a panel of experienced emergency physicians based on detailed prehospital, clinical, and laboratory data.

Inception cohort study.

Medical intensive care unit and department of emergency medicine at a university hospital.

A total of 162 unconscious, mechanically ventilated patients who survived > or =24 hrs after resuscitation from cardiac arrest.

Recording of sensory evoked potentials and outcome prediction after review of detailed clinical and laboratory data by emergency physicians within 24 hrs after cardiac arrest.

At 6 months, the outcome of 36 patients was classified as favorable and 126 patients were rated as poor. After review of prehospital data, emergency physicians predicted favorable vs. poor outcome with a sensitivity of 70% and a specificity of 65%. After additional assessment of data 1 hr after cardiac arrest, the sensitivity of emergency physician predictions increased to 80%, whereas the specificity decreased to 48%. Outcome prediction by emergency physicians was most accurate after obtaining detailed patient data 24 hrs after cardiac arrest (sensitivity, 81%; specificity, 58%). In 35 of 36 patients with favorable outcomes, the cortical evoked potential N70 peak was detected between 72 and 128 msec. Of 113 patients with an N70 peak latency >130 msec or an absent N70 peak, all except one had a poor outcome. By using a cutoff of 130 msec, the N70 peak latency alone had a sensitivity of 94% and a specificity of 97%. The predictive accuracy of the N70 peak latency was significantly higher than the clinical assessment 24 hrs after cardiac arrest (91% vs. 76%, p = .0003).

In unconscious cardiac arrest survivors, a recording of long-latency sensory evoked potentials is more accurate in predicting individual outcome than an emergency physician review of clinical data.