Patient selection: hypothermia following cardiac arrest
Outcome: good vs poor, with poor defined as Cerebral Performance categories 3 to 5 (severe disability, coma or death):
Independent predictors for a poor outcome:
(1) incomplete brainstem reflexes (papillary, oculocephalic, corneal) and myoclonus in normothermia (abnormal clinical exam)
(2) hypothermic electroencephalopathy nonreactive (absence of EEG reactivity)
(3) peak serum neuron specific enolase > 33 micrograms per L (usually at 24-48 hours, reference range not given, analysis Brahms NSE Kryptor immunoassay)
Absent hypothermic EEG reactivity had the highest coefficienct in ordinal logistic regression (4.4).
Bilateral absence of cortical somatosensory-evoked potentials (SEEP) was not an independent predictor for poor outcome.
Interpretation:
• The discussion on page 1343 includes SEEP in the cutoff for good vs poor outcome.
• It appears that the presence of all 3 predictors identifies a group that will have a poor outcome.
• Patients with <= 2 of the predictors may have a good or poor outcome.
Performance:
• The combination of the 3 independent predictors had an area under the ROC curve of 0.88 for poor outcome and 0.89 for mortality.
• The predictors can identify a group at high risk for poor outcome but others at high risk overlap with those having a good outcome.
• The performance of NSE assays varies so a different cutoff would be needed if a different assay is used.
