metricas
Revista Española de Anestesiología y Reanimación (English Edition) Beyond compliance: Patient risk, ERAS adherence, and postoperative outcomes thro...
Journal Information
Vol. 73. Issue 1.
(January 2026)
Share
Download PDF
More article options
Visits
346
Vol. 73. Issue 1.
(January 2026)
Original article

Beyond compliance: Patient risk, ERAS adherence, and postoperative outcomes through explainable machine learning

Más allá del cumplimiento: riesgo del paciente, adherencia a ERAS y resultados postoperatorios mediante aprendizaje automatizado explicable
Visits
346
J. Ripollés-Melchora,b,c, Á.V. Espinosac,d, A. Abad-Motose,f,
Corresponding author
ripo542@gmail.com

Corresponding author.
, A. Abad-Gurumetaa,b, P. Galán-Menéndezg, A. Zorrilla-Vacah, R. Navarro-Pérezi, A. Ruiz-Escobara,b, J. Fernanz-Antónj, A. Suárez-de-la-Ricak, C. Aldecoal, the EuroPOWER Study Investigators Group 1
a Department of Anaesthesia and Critical Care, Infanta Leonor University Hospital, Madrid, Spain
b Complutense University of Madrid, Madrid, Spain
c Fluid Therapy and Hemodynamic Monitoring Working Group, Spanish Society of Anaesthesia and Critical Care, Madrid, Spain
d Department of Anaesthesia and Critical Care, Froedtert Hospital Medical, College of Wisconsin, Milwaukee, WI, United States
e Department of Anaesthesia and Critical Care, Donostia University Hospital, Donostia-San Sebastian, Spain
f Patient Blood Management Working Group, Spanish Society of Anaesthesia and Critical Care, Madrid, Spain
g Department of Anaesthesia and Critical Care, Vall d’Hebrón University Hospital, Barcelona, Spain
h Anesthesiology, Perioperative and Pain Medicine Department, Brigham and Womens Hospital, Boston, MA, United States
i Department of Anaesthesia and Critical Care, Clínico San Carlos University, Hospital, Madrid, Spain
j Department of Anaesthesia and Critical Care, Complex Hospitalari Universitari Moisès Broggi, Sant Joan Despí, Spain
k Department of Anaesthesia and Critical Care, La Princesa University Hospital, Madrid, Spain
l Department of Anaesthesia and Critical Care, Río Hortega University Hospital, Valladolid, Spain
Ver más
This item has received
Article information
Abstract
Full Text
Bibliography
Download PDF
Statistics
Abstract
Background

Enhanced Recovery After Surgery (ERAS) protocols improve outcomes after colorectal surgery, but adherence remains variable and may interact with patient risk. Traditional compliance scores lack granularity to explore these dynamics. We aimed to use interpretable machine learning to quantify the contribution of individual ERAS items and clinical features to postoperative complications, and to identify data-driven ERAS phenotypes.

Methods

This was a secondary analysis of the EuroPOWER cohort (NCT04889798), a prospective European study including 2841 adults undergoing elective colorectal surgery. Two Extreme Gradient Boosting models were trained to predict in-hospital complications: a complete model (clinical variables + 23 ERAS items) and an ERAS-only model. Both were interpreted using Shapley Additive Explanations (SHAP). In the complete model, SHAP matrices were clustered to derive phenotypes. Feature importance, adherence, and complication rates were compared descriptively.

Results

The complete model achieved an AUC of 0.627. SHAP analysis identified frailty, ASA class, BMI, and age as leading predictors, followed by early mobilisation, nutritional care, and thromboprophylaxis. Three phenotypes were identified, with complication rates of 17.7%, 27.1%, and 41.1%, corresponding to robust, intermediate, and frail profiles. The ERAS-only model showed similar discrimination (area under the curve 0.642), but reduced interpretability. SHAP redundancy analysis supported inclusion of all ERAS items.

Conclusions

The clinical effect of ERAS adherence appears to be modulated by baseline vulnerability and implementation patterns. SHAP-based models enable transparent risk attribution and phenotype identification, supporting more targeted ERAS strategies and future development of automated quality monitoring tools.

Keywords:
Colorectal surgery
Enhanced recovery after surgery
Machine learning
Postoperative complications
Frailty
Abbreviations:
ASA
AUC
EPCO
BMI
ERAS
ROC
SHAP
UMAP
XGBoost
Resumen
Antecedentes

Los protocolos ERAS (Enhanced Recovery After Surgery) mejoran los resultados tras la cirugía colorrectal, pero la adherencia sigue siendo variable y puede interactuar con el riesgo del paciente. Las puntuaciones tradicionales de cumplimiento carecen de granularidad para explorar dichas dinámicas. Nuestro objetivo fue utilizar el aprendizaje automatizado (machine learning) interpretable para cuantificar la contribución de los ítems individuales de ERAS y las características clínicas a las complicaciones postoperatorias, e identificar los fenotipos ERAS impulsados por los datos.

Métodos

Este es un análisis secundario de la cohorte EuroPOWER (NCT04889798), un estudio prospectivo europeo que incluye 2.841 adultos sometidos a cirugía colorrectal electiva. Se formaron dos modelos de potenciación extrema de gradiente (Extreme Gradient Boosting) para predecir las complicaciones intrahospitalarias: un modelo completo (variables clínicas + 23 ítems ERAS) y un modelo exclusivamente ERAS. Ambos fueron interpretados mediante SHAP (Shapley Additive Explanations). En el modelo completo, se agruparon las matrices SHAP para derivar fenotipos. Se compararon descriptivamente la importancia de la función, la adherencia a la misma, así como las tasas de complicaciones.

Resultados

El modelo completo logró un ABC de 0,627. El análisis SHAP identificó la fragilidad, clase ASA, IMC y edad como predictores principales, seguidos de la movilización temprana, la atención nutricional y la tromboprofilaxis. Se identificaron tres fenotipos, con tasas de complicaciones del 17,7, 27,1 y 41,1%, correspondientes a perfiles sólido, intermedio y frágil, respectivamente. El modelo exclusivamente ERAS reflejó una discriminación similar (área bajo la curva de 0,642), pero redujo la interpretabilidad. El análisis de redundancia SHAP respaldó la inclusión de todos los ítems ERAS.

Conclusiones

El efecto clínico de la adherencia a ERAS parece estar modulado por la vulnerabilidad basal y los patrones de implementación. Los modelos basados en SHAP permiten la atribución transparente del riesgo y la identificación de fenotipos, respaldando estrategias ERAS más dirigidas, así como el desarrollo futuro de herramientas automatizadas de monitorización de la calidad.

Palabras clave:
Cirugía colorrectal
Mejora de la recuperación tras la cirugía
Aprendizaje automatizado
Complicaciones postoperatorias
Fragilidad

Article

These are the options to access the full texts of the publication Revista Española de Anestesiología y Reanimación (English Edition)
Subscriber
Subscriber

If you already have your login data, please click here .

If you have forgotten your password you can you can recover it by clicking here and selecting the option “I have forgotten my password”
Purchase
Purchase article

Purchasing article the PDF version will be downloaded

Purchase now
Contact
Phone for subscriptions and reporting of errors
From Monday to Friday from 9 a.m. to 6 p.m. (GMT + 1) except for the months of July and August which will be from 9 a.m. to 3 p.m.
Calls from Spain
932 415 960
Calls from outside Spain
+34 932 415 960
E-mail
Article options
Tools
Supplemental materials