Predicting death after CPR. Experience at a nonteaching community hospital with a full-time critical care staff

To identify a series of variables which predict death after in-hospital cardiopulmonary resuscitation (CPR).

Retrospective observational study.

A nonteaching community hospital with 24-hr on-site critical care specialists.

Consecutive adults undergoing CPR between August 1989 and July 1991.

None.

Two hundred forty-two patients suffered a total of 289 cardiopulmonary arrests. Forty patients (16.5%) survived to discharge. Thirty-nine (16%) patients had more than one cardiopulmonary arrest. Survival of second CPR was 18%. Acute physiology and chronic health evaluation (APACHE) II scores within 24 h of admission and CPR (APACHE[a] and APACHE[b]) were measured. APACHE(a) and (b) scores more than 20 had a 96% predictive value positive and were associated with a five-fold decrease in survival. Besides APACHE, cardiopulmonary arrests on medical floors and after day 4 of hospitalization, duration of CPR more than 15 min, and asystole assumed significance at multivariate levels for predicting death. Ventilatory assistance and Glasgow coma score of less than 9 at 24 h after CPR predicted death for initial survivors at multivariate levels. Survival on telemetry units were similar to the ICU (17 vs 21%) but twice that of the medical floors.

The CPR outcome can be predicted early during hospital course, which may assist physicians to formulate a do-not-resuscitate order. Patients surviving a CPR should be considered candidates for another resuscitation if clinically warranted. Low-risk patients can safely be admitted to telemetry units instead of to more costly ICUs.