Background

Renal disease is a known risk factor for adverse bariatric surgical outcomes. Chronic kidney disease (CKD) stage may be estimated based on several formulas for glomerular filtration rate (GFR), but it is not clear how formula accuracies differ in the morbidly obese population.

Methods

From the 2017 MBSAQIP database, we identified patients who had metabolic and bariatric surgery. GFR was calculated using several common formulas: Cockcroft-Gault formulas using actual, adjusted, and ideal body weight, Modification of Diet in Renal Disease (MDRD) formula, and the CKD-Epidemiology formula. Calculated GFR was compared between formulas as a predictor of adverse renal outcomes.

Results

149,430 patients had the requisite data to calculate GFR by all methods. There were significant differences in the CKD stage calculated by each method, with between 46.6– 96.0% in stage 1, 2.4– 44.9% in stage 2, 0.3– 8.5% in stage 3a, 0.2– 2.5% in stage 3b, 0.1– 0.5% in stage 4 CKD, and 1.0 – 1.1% in stage 5 CKD. 217 (0.1%) had adverse renal outcomes. Area under the ROC curve for the formulas varied from 0.634 (denoting poor accuracy) to 0.727 (denoting fair accuracy), with the CKD-EPI most predictive of adverse renal sequelae (Figure 1).

Conclusions

Estimations of glomerular filtration rate vary dramatically by the choice of formula used, which can significantly under- or over-predict the likelihood of renal complications. In this large bariatric dataset, the most accurate predictor was the CKD-EPI formula, which is independent of weight. In morbid obesity, weight-based GFR formulas may be unsuitable.