Prediction of short-term and long-term outcomes after cardiac arrest: a prospective multivariate approach combining biochemical, clinical, electrophysiological, and neuropsychological investigations.

OBJECTIVE

To determine the prognostic accuracy of biochemical, clinical, electrophysiological, and neuropsychological investigations in predicting outcomes after cardiac arrest.

DESIGN

Prospective study.

SETTING

Intensive care unit of the Hamburg-Eppendorf University Medical Center, Hamburg, Germany.

PATIENTS

A total of 80 patients (mean age, 63.79 +/- 15.85 yrs) after cardiopulmonary resuscitation.

INTERVENTIONS

Serial blood samples (days 2-4), clinical examinations (days 2 and 4), sensory-evoked potentials (day 4), and neuropsychological assessments (<or=1 and 6 months).

MEASUREMENTS AND MAIN RESULTS

We conducted a prospective study into the combined predictive efficacy of serum concentrations of neuron-specific enolase and protein S-100B, standardized clinical examinations, and short- and long-latency sensory-evoked potentials. For the prognostic validation, both the dichotomized 5-point Glasgow-Pittsburgh Cerebral Performance Categories (1-3, favorable outcome; 4-5, unfavorable outcome) and a comprehensive neuropsychological test battery were applied. A multivariate logistic-regression analysis resulted in a model in which 85% of the variance in the dichotomized Glasgow-Pittsburgh Cerebral Performance Categories was explained by neuron-specific enolase at day 4, clinical examination score at day 4, and age. This predictor index had a sensitivity of 92% and a specificity of 93%. In addition, 26 patients (out of 33) underwent neuropsychological testing at 6 months. Significant correlations were found with selected cognitive variables and S-100B at day 3, long-latency sensory-evoked potential at day 4, and neuropsychological bedside screening.

CONCLUSIONS

A multivariate assessment approach should be used to establish an early high-certainty prognosis after cardiac arrest. However, further prospective clinical studies are necessary to confirm this derived predictor index. In addition, an early recording of S-100B, long-latency sensory-evoked potential, and neuropsychological bedside screening reflect a cognitive long-term outcome.