- Poster presentation
- Open Access
Using outcome prediction tools in the ICU: performance of APACHE II and SAPS 2 scores in clinical patients
© BioMed Central Ltd 2006
- Published: 21 March 2006
- Prospective Cohort
- Positive Predictive Value
- Prospective Cohort Study
- Negative Predictive Value
- Outcome Prediction
Outcome prediction has become increasingly important over time, either to evaluate quality of care or to make therapeutic decisions. The APACHE II and SAPS 2 scores are widely used tools to assess prognosis among ICU patients. Nevertheless, there are still conflicting data concerning the best instrument to be used in specific groups of patients.
To compare the performance of APACHE II and SAPS 2 in a cohort of clinical patients.
A prospective cohort study in a 19-bed ICU in a private hospital.
All clinical patients admitted to the ICU during the period of the study.
There were 527 patients admitted between September and November 2005. Seventy-nine (15%) of them were admitted for clinical reasons, and 56 (70.8%) had simultaneous measurement of the APACHE II and the SAPS 2 scores. The mean APACHE II and SAPS 2 were 20.87 ± 10.27 and 39.09 ± 14.59, respectively. Nine patients died (3.7%). The mean APACHE II and SAPS 2 in the survivors and nonsurvivors were 18.09 ± 7.94 and 35.44 ± 8.86 (P < 0.0001), and 35.36 ± 12.33 and 58.56 ± 9.08 (P < 0.0001), respectively. The area under the ROC curve (AUC) was 0.960 (95% CI 0.910–1.010) for the APACHE II score, and 0.934 (95% CI 0.869–0.998) for the SAPS 2 score. Using a cutoff value of 51 points for the SAPS 2, sensitivity was 100%, specificity 88.4%, positive predictive value (PPV) 64.3% and negative predictive value (NPV) was 100%. Using a cutoff value of 25 points for the APACHE II, sensitivity was 100%, specificity 91.5%, PPV 69.2% and NPV 100%. The OR for an APACHE score >25 was 3.25 (95% CI 1.44–7.34), and for SAPS 2 >51 was 2.8 (95% CI 1.39–5.66). The association of both scores increases the specificity to 97.9% and PPV to 90%, maintaining the sensitivity and NPV of isolated scores, increasing the OR to 10 (95% CI 1.59–64.20).
In this cohort of clinical patients, the association of SAPS 2 and APACHE II scores is a better predictor of mortality than the isolated measurements.