[Evaluation of the efficiency of care in the ICU]

To evaluate the efficiency of care in the ICU using a predictive model.

A prospective, observational cohort study

Seventeen Spanish polyvalent ICUs.

A total of 1956 patients were initially considered (cohort A). Posteriorly, and at 6-year intervals, we documented cohorts B (n=453), C (n=2567) and D (n=711) in one of the studied ICUs.

Five standard severity indices were calculated for all cohorts, and with these the standardized mortality ratios (observed/calculated) for each cohort were compared. Multiple regression analysis was used to develop a predictive model of length of stay in the ICU (ICU-LOS). This model was used for calculation of the standardized LOS ratios for each cohort. We analyzed the organizational changes in the studied ICU during these periods in relation to the results obtained.

The calculated probability of in-hospital death was 15.4%, versus 14.7% as calculated 24 hours after admission. Actual in-hospital mortality was 20.3%. A final multiple regression model was constructed. Standardized LOS and mortality ratios were 1.8 and 1.2 (cohort B), 0.97 and 1.07 (cohort C), and 0.63 and 1.07 (cohort D), respectively. The progressive improvement in the results observed was related to the introduced organizational and structural changes.

The model developed in this study was a good predictor of actual ICU-LOS, and both LOS and mortality analysis could be a good tool for ICU care evaluation.