Background

We have previously shown that the MBSAQIP risk calculator has a good predictive ability for postoperative complications and weight loss and is important for patient decision making. However, Its accuracy for revisional procedures is unclear. We aimed to assess the predictive accuracy of the calculator for revisional bariatric procedures.

Methods

We queried our institutional MBSAQIP registry for adult patients who underwent revisional bariatric surgery from 2016-2021. We recorded patient demographics, procedure type, preoperative comorbidities, 30-day outcomes, BMI at 6 and 12 months, and comorbidity remission at 12 months. We then used the online calculator to record predictions for the same patients. Receiver operating characteristic curves were constructed to assess predictive utility for dichotomous outcomes. For BMI, predictive error (predicted - actual BMI) was also calculated.

Results

A total of 169 revisional procedures were included, of which, 83.5% were gastric bypass. The area under the curve for any complication, serious complications, readmissions, and interventions was 0.61, 0.73, 0.58, and 0.61, respectively (Figure 1). The correlation coefficient for predicted and observed 1-year BMI was 0.56, p<0.001; the mean predictive error for 1-year BMI was -3.8 +- 5, and 75% of the cases had a negative predictive error (Figure 2).

Conclusions

For revisional bariatric operations, the MBSAQIP risk calculator fails to reliably predict postoperative complications and readmissions and overpredicts weight loss. Given the rising frequency of revisions and the demonstrated importance of the calculator for patient decision making it is important to develop a revision-specific calculator with more accurate estimates.