The utility of predictive scores for intraoperative or postoperative morbi-mortality in visceral surgery - 10/12/25
, Julie Veziant b, c, Afak Nsiri d, e, Arnaud Alves fHighlights |
• | Validation of a predictive score for morbi-mortality is carried out in accordance with a well-defined process: internal validation, external validation, calculation of discriminant and calibration capacity, and analysis of its clinical impact. |
• | Few of the predictive scores put forward in the literature have been subjected to the full validation process with regard to scheduled visceral surgery. |
• | In spite of its insufficiently discriminative performances, the ASA-PS score is the most widely used in current practice. |
• | The SORT score is more effectively predictive than the POSSUM scores. |
• | Up until now, no specific organ-based score had received formal external validation. |
• | The role of artificial intelligence in the form of machine learning, otherwise known as automatic learning, remains to be defined. |
Summary |
The objective of this narrative review of the literature was to discuss the original concept and ongoing validation of predictive scores for intraoperative or postoperative morbi-mortality, and their overall utility in scheduled visceral surgery. The development and validation of a predictive score for morbi-mortality is carried out by means of a well-defined process encompassing internal and external validation according to temporality, geographical location or domain (patient population and/or type of surgery), and the clinical impact of application of the score. Bibliographic analysis has shown that very few existing studies possess the required methodological qualities. Detailed individual analysis of the different scores, not including nutritional scores, showed that the ASA-PS score is easy to use but yields only moderate performance, that the SORT score (six variables) is more powerfully discriminatory than the POSSUM score, that the POSPOM score has not been validated in visceral surgery, that the ACS-NSQIP score cannot be utilized in Europe and, finally, that the Charlson index is not sufficiently discriminatory in the short term. If specific types of surgery are taken into consideration, none of these scores have received formal external validation. In conclusion, the SORT score has been the most validated, and the ASA-PS score is widely used in current practice. The role of artificial intelligence in the form of machine learning remains to be defined.
Le texte complet de cet article est disponible en PDF.Keywords : Predictive tests, Prediction, Morbidity, Mortality, Visceral surgery
Plan
Bienvenue sur EM-consulte, la référence des professionnels de santé.
L’accès au texte intégral de cet article nécessite un abonnement.
Déjà abonné à cette revue ?