Acute Physiology and Chronic Health Evaluation (APACHE II) score and outcome in the surgical intensive care unit: an analysis of multiple intervention and outcome variables in 1,238 patients.

OBJECTIVE

To assess the statistical association of the Acute Physiology and Chronic Health Evaluation (APACHE II) score with multiple intervention and outcome variables in surgical ICU patients.

DESIGN

Continuous data collection on every patient admitted to the surgical ICU for a 21-month period.

MATERIALS AND

METHODS

For every admitted patient in the surgical ICU, APACHE II scores were calculated and the relationship between APACHE II score as an independent predictor of outcome was assessed with multiple outcome variables selected for study. The outcome and intervention variables tested included: treatment intervention measures such as days on ventilator; days with an arterial catheter, central venous catheter, triple lumen catheter, pulmonary artery catheter; days receiving total parenteral nutrition; days receiving tube feedings; number of transducers per days in the ICU; number of infusion pumps per ICU days, days in the hospital, number of complete blood counts; number of electrolyte determinations; number of blood gases; number of units of blood transfused; ICU and hospital mortality rates in the presence of complications, including: respiratory distress syndrome, renal failure, congestive heart failure, coma, requirement of cardiopulmonary resuscitation, and others.

RESULTS

The APACHE II score was statistically associated with each intervention and outcome variable tested. Unfortunately, the associations, although consistent, were weak with r2 values ranging from .03 to a maximum of .22 for Pearson's correlation coefficients.

CONCLUSION

The APACHE II score was statistically associated with all the variables examined in our surgical patients, but its predictive power for the individual surgical patient was limited. These findings suggest that the score may be useful for retrospective analyses of large cohorts of patients but should not be used as a triage tool or as a predictor of outcome for the individual patient. Triage decisions should continue to be based on the best available clinical judgment.