Simplification of the SAPS by selecting independent variables

Area of Science:

• Critical Care Medicine
• Health Informatics
• Biostatistics

Background:

• The Simplified Acute Physiology Score (SAPS) is a widely used severity scoring system in intensive care units (ICUs).
• Optimizing the number of variables in scoring systems is crucial for efficient clinical application and data management.

Purpose of the Study:

• To develop and prospectively validate simplified versions of the SAPS, namely SAPS.R and SAPS.E.
• To assess if incorporating additional biological variables improves the accuracy of SAPS-derived scores.
• To evaluate the discriminating power of the original SAPS, SAPS.R, and SAPS.E.

Main Methods:

• Retrospective creation of SAPS.R (14 variables) and SAPS.E (22 variables) using logistic regression on 351 ICU admissions.
• Prospective validation of SAPS, SAPS.R, and SAPS.E using receiver operating characteristic (ROC) curve analysis on 446 ICU admissions.
• Comparison of the areas under the ROC curves to assess the discriminating power of each score.

Main Results:

• SAPS.R was derived using 5 variables, and SAPS.E using 7 variables from the original SAPS and additional biological data.
• No significant differences were found in the overall accuracy between SAPS, SAPS.R, and SAPS.E.
• The addition of biological variables did not enhance the predictive accuracy of the scoring systems.

Conclusions:

• The number of variables in the original SAPS can be significantly reduced without compromising accuracy.
• Simplified scoring systems like SAPS.R and SAPS.E offer comparable performance to the original SAPS.
• Streamlined ICU scoring systems can facilitate more efficient patient assessment and data analysis.