Prospective comparison of three risk score models at three different surgical intensive care units

Although risk score models are of great value, their use is restricted because of the additional effort involved. The aim of this study was to compare three different score systems. Each of these requires a different degree of effort by the medical staff. One of the score systems is solely based on routine laboratory parameters. Data were collected on three different ICUs units, with each showing a large variety in patients' health conditions.

Prospective data of 588 surgical patients were collected by means of Acute Physiology and Chronic Health Evaluation II (APACHE II), Dense Laboratory Whole Blood Applied Risk Estimation (DELAWARE), and Simplified Acute Physiology Score II (SAPS II) score systems. These patients were admitted to three different intensive care units over a period of 12 months. On the day of admission, predicted hospital survival and mortality were evaluated.

With a cutoff value of 0.6, the sensitivity of the APACHE II, DELAWARE, and SAPS II was at 0.19, 0.24, and 0.21; the specificity was at 0.98, 0.92, and 0.98; and the correct classification rate at 0.86, 0.83, and 0.86. The r(2) value was 0.35 for the APACHE II, 0.12 for the DELAWARE, and 0.21 for the SAPS II. The hospital mortality rate was overestimated in all three score systems.

The results of this first multicenter study comparing three risk score systems indicate that it is possible to establish a general risk score for surgical intensive care patients on admission date. Such a risk score is solely based on quality-controlled, low-cost routine laboratory parameters.