Parental Obesity Predisposition and Age of Onset Associate with Poor Response to Bariatric and Metabolic Surgery

Figures

• 

  Fig. 1. Flow chart of cohort selection. Of ~ 75,000 observations in the StuDoQ | MBE register at the time of application, a search was conducted with the primary inclusion criteria “familial predisposition to obesity,” “age of onset,” “age & weight at surgery,” “ > follow-up ≥ 1 year.” There were 15,073 cases identified. We excluded non-primary procedures, non-standard and rare procedures as well as observations that failed a plausibility check during quality control of the data. The final study cohort entailed 14,404 observations

• 

  Fig. 2. 1-year TWL distribution of the StuDoQ | MBE study cohort. Boxes represent deciles and numbers above boxes indicate patient number. The vertical red line indicates the cut-off between responders and non-responders

• 

  Fig. 3. Adjusted odds ratio coefficients for risk factors associated with poor MBS response (1-year TWL < 20%). The estimated coefficients in this plot are derive from the multivariate logistic regression model described above. Bar thickness provides estimates of the 95, 99, and 99.9 confidence intervals. For age, pre-operative weight gain, pre-operative BMI, the odds ratio depicted is for a 1 unit increase of the respective covariate. “Adult-onset” indicates > = 18 years. Roux-en-y gastric bypass is the reference category (OR = 1, not depicted) for sleeve gastrectomy and one-anastomosis gastric bypass

• 

  Fig. 4. Comparison of two predictive models of 1-year TWL failure. We used the DeLong method to compare ROC-curves. The ROC curves show the predictive power of two models of 1-year TWL failure, one using all covariates (Fig. 3) without surgery type (blue curve) vs all covariates with surgery type (red curve). Adding surgery type independently improves the AUC from 0.732 to 0.749 (p < 0.001), which means that there is higher sensitivity when predicting surgical failure)