Abstract
Background:
The predictive potential of scoring systems used in intensive care units (ICUs) to distinguish patients’ mortality is heavily investigated but scarcely pooled.

Objectives:
To statistically analyze the accuracy of three ICU generic scores, including Acute Physiology and Chronic Health Evaluation (APACHE II), Simplified Acute Physiology Score (SAPS II), and Sequential Organ Failure Assessment (SOFA), to predict mortality.

Methods:
A meta-analysis was conducted involving prospective studies published between January 2000 and February 2019 to analyze the performance of APACHE II, SAPS II, and SOFA to predict patients’ mortality through pooling their discriminative indicators, such as sensitivity specificity, and the area under the hierarchical summary receiver operating characteristic curve (HSROC). The inconsistency index (I²) was used to assess heterogeneity in sensitivity and specificity, while meta-regression analysis was performed to detect the potential sources of heterogeneity. Deek’s funnel plots were used to assess the potential publication bias.

Results:
In a total of 37 studies (7612 patients, 63.58% males, 75.7% assessed in-hospital mortality), 2170 observed deaths were reported. APACHE II, SAPS II, and SOFA scoring systems showed good mortality predictive performances, where the pooled sensitivities were 0.81, 0.76, and 0.80, respectively, specificities were 0.78, 0.89, and 0.79, respectively, and pooled HSROCs were 0.87, 0.85, and 0.88, respectively. For APACHE II, SAPS II, and SOFA, there were significant in-between study heterogeneities in sensitivity (I²=85.21%, 84.31%, and 71.67%, respectively) and specificity (I²=93.74%, 92.65%, and 89.41%, respectively) while no publication bias was detected (P=0.689, 0.465, and 0.181, respectively). There was significant heterogeneity among studies that investigated APACHE II with a cut-off score of > 20 (P = 0.04) and those investigating SAPS II for > 24 months (P < .001).

Conclusion:
Within the limitations of the study, ICU scores showed good prognostic performance. Future studies should be conducted using fixed-time endpoints of mortality, involving multiple countries and employ a combination of generic ICU scores.

Key words: APACHE II, SAPS II, SOFA, critical illness, Mortality