Related Experiment Video
Updated: Dec 1, 2025
Author Spotlight: Assessing Surgical Frailty with Point-of-Care Ultrasound of Quadriceps Muscles
Published on: July 26, 2024
Predicting morbidity in liver resection surgery: external validation of the revised frailty index and development of
Don Milliken1, Sam Curtis1, Clare Melikian1
1Royal Free Perioperative Research Group, Royal Free London NHS Foundation Trust, Pond Street, London, NW3 2QG, UK.
Background:
Perioperative complications of liver resection surgery are common but individual patient-level prediction is difficult. Most risk models are unvalidated and may not be clinically useful. We aimed to validate a risk prediction model for complications of liver resection, the Revised Frailty Index (rFI), at a high volume centre. We also aimed to derive a predictive model for complications in our cohort.
Methods:
Records were reviewed for 300 patients undergoing liver resection. The rFI's discrimination of 90-day major complications was assessed by receiver operating curve analysis. Logistic regression analysis was then used to fit rFI covariates to our dataset. A further analysis produced a model with optimal discrimination of 90-day major complications.
Results:
The rFI was a poor discriminator of 90-day major complications (AUROC 0.562) among patients at our centre. The rFI optimised fit model demonstrated improved discrimination of 90-day major complications (AUROC 0.685). We developed a novel model with improved fit and similar discrimination (AUROC 0.710).
Conclusion:
We were unable to validate the rFI as a predictor of complications. We developed a novel model with discrimination at least equal to other published risk models. However, there is an unmet need for well-validated, clinically useful risk tools in this area.
