Predictors of long-term therapeutic failure after bariatric surgery: a logistic regression model based on nutritional and metabolic parameters

Objective: this study analyzed long-term therapeutic failure after bariatric surgery and validated a 10-year predictive risk model. We aimed to integrate clinical and postoperative variables to identify failure-associated factors and establish a clinically useful threshold for follow-up. Methods: A retrospective observational cohort study was conducted including 96 patients (78.1% women, mean age 41.3 years, BMI 44.6 kg/m²) who underwent gastric bypass (n = 75) or sleeve gastrectomy (n = 19). Therapeutic failure was defined as <50% excess weight loss or recurrence of comorbidities. A logistic regression model was developed including baseline BMI, %EWL at 1 year, BMI reduction, type 2 diabetes (T2DM), and hypertension (HTN). Model performance was assessed using ROC curves and the Hosmer-Lemeshow test. Results: failure rates increased from 18 % to 32 % over 10 years. Risk factors: higher baseline BMI (OR = 1.24), T2DM (OR = 3.77), and HTN (OR = 3.83). One-year BMI loss was protective (OR = 0.72). The model showed strong predictive capacity (AU C = 0.858), with an optimal threshold of 28.6 % (sensitivity 84.6 %, specificity 80.6 %, NPV 93.1 %). Gastric bypass had lower failure rates (28 % vs. 38 %) and better predictive performance (AUC = 0.89 vs. 0.78). Conclusions: the model effectively predicts long-term failure, with gastric bypass demonstrating superior outcomes. We recommend prioritizing this procedure for patients with higher BMI or comorbidities.