Abstract
Background: Intensive Care Units (ICUs) are vital for critically ill patients, yet face high mortality rates (27.6% in Indonesia) and significant costs, necessitating efficient resource allocation. To improve patient care and resource utilization, scoring systems such as the Acute Physiology and Chronic Health Evaluation II (APACHE II), Sequential Organ Failure Assessment (SOFA), Simplified Acute Physiology Score II (SAPS II), Early Warning Scores (EWS), and Modified SOFA (MSOFA) are used to assess illness severity and predict outcomes. This research at Wahidin Sudirohusodo Hospital aimed to validate the predictive power of these scores for mortality and length of stay in intensive care unit (ICU) patients, ultimately enhancing critical care quality, and informing admission criteria.
Methods: This retrospective study employed a diagnostic and longitudinal cross-sectional design, analyzing data from 210 ICU patients admitted to Dr. Wahidin Sudirohusodo Hospital in Makassar between January 2023 and January 2024. The dataset included EWS, APACHE II, SAPS II, SOFA, and MSOFA scores. Numerical variables were summarized as mean±standard deviation. Receiver operating characteristic (ROC) curves were used to assess diagnostic accuracy (AUC, cut-off values, sensitivity, specificity), while Spearman’s Rho was employed for correlation analysis, and the Kruskal-Wallis test for comparative analysis.
Results: While APACHE II, SAPS II, SOFA, EWS, and MSOFA scores generally predicted mortality and ICU length of stay, EWS was less accurate, particularly for long-term predictions, though combining it with other scores improved the results. APACHE II excelled in predicting long-term mortality, SAPS II for 2-7 day mortality, and SOFA was strong across all time frames. MSOFA was the most accurate for predicting very short-term mortality (≤1 day) and showed the strongest correlation with length of hospital stay.
Conclusion: Various scoring systems, including APACHE II, SAPS II, SOFA, EWS, and MSOFA, have been shown to be valid in predicting patient mortality and ICU length of stay. While EWS has lower predictive ability, particularly for medium and long-term mortality, combining it with other scores could improve accuracy. Specifically, MSOFA demonstrated the best performance in predicting short-term mortality (≤1 day) and has the strongest correlation with length of stay. SOFA was also an excellent predictor across all mortality timeframes, while APACHE II was most accurate for long-term.
Fuad Widianto, Ari Santri Palinrungi, Faisal Muchtar, Syamsu Hilal Salam, Muhammad Rum, Andi Adil
