Unit:
Κατεύθυνση ΒιοστατιστικήLibrary of the School of Health Sciences
Author:
Xanthakis Ioannis
Supervisors info:
Ευάγγελος Κριτσωτάκης, Αναπληρωτής Καθηγητής, Ιατρική Σχολή, Πανεπιστήμιο Κρήτης
Ευαγγελία Σαμόλη, Αναπληρώτρια Καθηγήτρια, Ιατρική Σχολή, ΕΚΠΑ
Μαρία Κάσδαγλη, Επιστημονική Συνεργάτιδα, Ιατρική Σχολή, ΕΚΠΑ
Original Title:
Αξιολόγηση της προγνωστικής ακρίβειας της κλίμακας εκτίμησης κινδύνου SAPS II για την πρόβλεψη της θνησιμότητας ασθενών μονάδων εντατικής θεραπείας: Συστηματική ανασκόπηση και μετα-ανάλυση
Translated title:
Validation of the prognostic accuracy of severity of illness score SAPS II in predicting ICU patients’ mortality: A Systematic Review and Meta-Analysis
Summary:
Introduction: The ideal management of severe diseases includes the fast and precise identification of critical care patients. A useful tool that can achieve that is the widely used prognostic model of in-hospital mortality, SAPS II. The objective of this study is to gather external validation studies of SAPS II from the international literature and to meta-analyze its predictive performace measurements.
Methods: A systematic search was conducted between October 16 and November 20, 2022. Studies were chosen if they were published between 2003 and 2022, and reviewed SAPS II on predicting ICU patients’ mortality. Data that were extracted included study and patients’ characteristics, predictive performance measurements and the risk of bias was assessed for each study. The discrimination measure that was used for the meta-analysis was the Area Under the ROC curve (AUROC). We synthesized the results with the use of a random effects model and researched the source of heterogeneity by subgroup analysis.
Results: A total of 728 articles were screened, 306(42%) of which had their full text assessed. Of those 200 were judged as suitable, but only those that were published in 2022 (25) were included in the study. Of the 25 studies, 23 (69,7%) researched in-hospital mortality, 2 (6%) ICU mortality, 8 (23,3%) 1-month mortality, while 11 (44%) were found to have low risk of bias, 11 (44%) unclear risk of bias and 3 (12%) high risk of bias. The pooled estimate of AUROC was 0,75 (95% CI: 0,72-0,78). A high degree of between study heterogeneity was noted (I2 = 98%), with a 95% prediction interval of AUROC between 0,58 and 0,87. Subgroup analysis depending on type of event and overall risk of bias showed no statistical difference between the groups.
Conclusions: The SAPS II model can achieve a decent discrimination as a tool of predicting ICU mortality. However, the absence of calibration measurements and the limited use of decision curve analysis deter the absolute and consistent review of the model and its establishment as a useful tool for daily clinical practice.
Main subject category:
Health Sciences
Keywords:
Μeta-analysis, SAPS II, Prognostic model, Mortality, ICU patients, AUROC
