Comparative Safety of Sleeve Gastrectomy and Gastric Bypass: An Instrumental Variables Approach

Abstract

Objective: To compare the safety of sleeve gastrectomy and gastric bypass in a large cohort of commercially insured bariatric surgery patients from the IBM MarketScan claims database, while accounting for measurable and unmeasurable sources of selection bias in who is chosen for each operation.

Summary of background data: Sleeve gastrectomy has rapidly become the most common bariatric operation performed in the United States, but its longer-term safety is poorly described, and the risk of worsening gastroesophageal reflux requiring revision may be higher than previously thought. Prior studies comparing sleeve gastrectomy to gastric bypass are limited by low sample size (in randomized trials) and selection bias (in observational studies).

Methods: Instrumental variables analysis of commercially insured patients in the IBM MarketScan claims database from 2011 to 2018. We studied patients undergoing bariatric surgery from 2012 to 2016. We identified re-interventions and complications at 30 days and 2 years from surgery using Comprehensive Procedural Terminology and International Classification of Disease (ICD)-9/10 codes. To overcome unmeasured confounding, we use the prior year's sleeve gastrectomy utilization within each state as an instrumental variable-exploiting variation in the timing of payers' decisions to cover sleeve gastrectomy as a natural experiment.

Results: Among 38,153 patients who underwent bariatric surgery between 2012 and 2016, the share of sleeve gastrectomy rose from 52.6% (2012) to 75% (2016). At 2 years from surgery, patients undergoing sleeve gastrectomy had fewer re-interventions (sleeve 9.9%, bypass 15.6%, P < 0.001) and complications (sleeve 6.6%, bypass 9.6%, P = 0.001), and lower overall healthcare spending ($47,891 vs $55,213, P = 0.003), than patients undergoing gastric bypass. However, at the 2-year mark, revisions were slightly more common in sleeve gastrectomy than in gastric bypass (sleeve 0.6%, bypass 0.4%, P = 0.009).

Conclusions and relevance: In a large cohort of commercially insured patients, sleeve gastrectomy had a superior safety profile to gastric bypass up to 2 years from surgery, even when accounting for selection bias. However, the higher risk of revisions in sleeve gastrectomy merits further exploration.

Figure 1 –. Variation in State-Level Utilization of Sleeve Gastrectomy

The Y-axis reflects the % of all bariatric operations performed in the MarketScan dataset that were sleeve gastrectomy during each year and within each state. This figure reflects the 9 states with the highest case volume during the study period. Due to data vendor privacy restrictions, the 2014 sleeve rate in CA is not displayed.

Figure 2 –. Instrumental Variable Analysis of 30-day Outcomes

Outcomes reflect IV-adjusted absolute risk differences of each endpoint between sleeve gastrectomy and gastric bypass. Error bars reflect 95% CIs.

Figure 3 –. Instrumental Variable Analysis of 2-year Outcomes

Outcomes reflect IV-adjusted absolute risk differences of each endpoint between sleeve gastrectomy and gastric bypass. Error bars reflect 95% CIs. Asterisks (*) indicates the outcome had a statistically significant test for endogeneity (a first-stage regression residual with p<0.05).