Development and Internal Validation of a Metabolic and Bariatric Surgery Frailty Index (MBS-FI)
Related Experiment Video
Updated: Oct 31, 2026
Frailty Assessment in an Aging Mouse Model
Published on: September 23, 2025
Kaique Filardi1, Carolina Michelle Mejia-Alaniz2, Sara Saeidishahri3
1Center for Obesity Research, Innovation and Education, Digestive Health Institute, Hartford Hospital, Hartford, CT, USA. kaique.filardi@hhchealth.org.
Introduction:
Frailty is associated with adverse outcomes after metabolic and bariatric surgery (MBS). Existing bariatric frailty scores include MBSAQIP-derived tools, but reported discrimination is modest and prior accumulating-deficit indices used earlier or restricted cohorts. We developed and internally validated a contemporary adult MBSAQIP deficit index and compared it with the modified Frailty Index-5 (mFI-5) and Bariatric Frailty Score (BFS).
Methods:
We conducted a retrospective analysis using the MBSAQIP 2023-2024 cohort, including adults undergoing sleeve gastrectomy or Roux-en-Y gastric bypass as a primary or conversion MBS. Preoperative health-deficit variables were coded and summed to create the MBS-FI. Candidate deficits were excluded if missing > 10%, too rare (< 1%), too common (> 80%), or unrelated to age. The data were randomly split into FI-derivation and validation cohorts (70%/30%, respectively). Outcomes included in-hospital mortality, 30-day mortality, serious complications, cardiovascular complications, reoperation, and length of stay > 2 days.
Results:
The study identified 359,767 patients, of whom 252,043 were in the derivation cohort. Of 28 candidate variables, 18 remained in the final model. The MBS-FI displayed a right-skewed distribution and a positive correlation with age. Higher MBS-FI scores were progressively associated with adverse outcomes. Although the MBS-FI demonstrated numerically higher discrimination for 30-day and in-hospital mortality, these differences were not statistically significant in the overall cohort (p = 0.33 and p = 0.49, respectively. In the validation cohort, the adjusted model yielded an AUC of 0.74 (95% CI, 0.68-0.79) for 30-day mortality based on 75 events and 0.80 (95% CI, 0.69-0.91) for in-hospital mortality based on 14 events.
Conclusion:
The MBS-FI was internally validated and provided incremental discrimination for several non-mortality outcomes. Improvements for mortality were limited, and external validation and assessment of clinical utility are required before routine use.
