Comparative Safety and Effectiveness of Roux-en-Y Gastric Bypass and Sleeve Gastrectomy for Weight Loss and Type 2 Diabetes Across Race and Ethnicity in the PCORnet Bariatric Study Cohort

Key Points

Question  Are there differences in weight, diabetes, and safety and utilization outcomes for Roux-en-Y gastric bypass (RYGB) and sleeve gastrectomy (SG) across racial and ethnic groups in the National Patient-Centered Clinical Research Network (PCORnet) Bariatric Study?

Findings  In this comparative effectiveness research study including 36 871 adult and adolescent bariatric surgery patients, 34% of whom were Black or Hispanic, the variability in weight loss and diabetes outcomes for different racial and ethnic groups having RYGB or SG was clinically small. However, differences for safety and utilization outcomes were both clinically and statistically significant for Black and Hispanic patients who had RYGB compared with SG.

Meaning  These differences may be associated with the preoperative preparation and postoperative care processes for these patients rather than the bariatric operations themselves; work should focus on how to improve care processes and decision-making between bariatric operations for diverse patient populations.

Importance  Bariatric surgery is the most effective treatment for severe obesity; yet it is unclear whether the long-term safety and comparative effectiveness of these operations differ across racial and ethnic groups.

Objective  To compare outcomes of Roux-en-Y gastric bypass (RYGB) and sleeve gastrectomy (SG) across racial and ethnic groups in the National Patient-Centered Clinical Research Network (PCORnet) Bariatric Study.

Design, Setting, and Participants  This was a retrospective, observational, comparative effectiveness cohort study that comprised 25 health care systems in the PCORnet Bariatric Study. Patients were adults and adolescents aged 12 to 79 years who underwent a primary (first nonrevisional) RYGB or SG operation between January 1, 2005, and September 30, 2015, at participating health systems. Patient race and ethnicity included Black, Hispanic, White, other, and unrecorded. Data were analyzed from July 1, 2021, to January 17, 2022.

Exposure  RYGB or SG.

Outcomes  Percentage total weight loss (%TWL); type 2 diabetes remission, relapse, and change in hemoglobin A_1c (HbA_1c) level; and postsurgical safety and utilization outcomes (operations, interventions, revisions/conversions, endoscopy, hospitalizations, mortality, 30-day major adverse events) at 1, 3, and 5 years after surgery.

Results  A total of 36 871 patients (mean [SE] age, 45.0 [11.7] years; 29 746 female patients [81%]) were included in the weight analysis. Patients identified with the following race and ethnic categories: 6891 Black (19%), 8756 Hispanic (24%), 19 645 White (53%), 826 other (2%), and 783 unrecorded (2%). Weight loss and mean reductions in HbA_1c level were larger for RYGB than SG in all years for Black, Hispanic, and White patients (difference in 5-year weight loss: Black, −7.6%; 95% CI, −8.0 to −7.1; P < .001; Hispanic, −6.2%; 95% CI, −6.6 to −5.9; P < .001; White, −5.9%; 95% CI, −6.3 to −5.7; P < .001; difference in change in year 5 HbA_1c level: Black, −0.29; 95% CI, −0.51 to −0.08; P = .009; Hispanic, −0.45; 95% CI, −0.61 to −0.29; P < .001; and White, −0.25; 95% CI, −0.40 to −0.11; P = .001.) The magnitude of these differences was small among racial and ethnic groups (1%-3% of %TWL). Black and Hispanic patients had higher risk of hospitalization when they had RYGB compared with SG (hazard ratio [HR], 1.45; 95% CI, 1.17-1.79; P = .001 and 1.48; 95% CI, 1.22-1.79; P < .001, respectively). Hispanic patients had greater risk of all-cause mortality (HR, 2.41; 95% CI, 1.24-4.70; P = .01) and higher odds of a 30-day major adverse event (odds ratio, 1.92; 95% CI, 1.38-2.68; P < .001) for RYGB compared with SG. There was no interaction between race and ethnicity and operation type for diabetes remission and relapse.

Conclusions and Relevance  Variability of the comparative effectiveness of operations for %TWL and HbA_1c level across race and ethnicity was clinically small; however, differences in safety and utilization outcomes were clinically and statistically significant for Black and Hispanic patients who had RYGB compared with SG. These findings can inform shared decision-making regarding bariatric operation choice for different racial and ethnic groups of patients.

Severe obesity has increased in prevalence disproportionately for racial and ethnic groups over the last decade.¹ Unfortunately, many intensive, multicomponent lifestyle interventions have not shown substantial long-term weight loss.² This has led to the development of surgical treatments for severe obesity, referred to as metabolic and bariatric surgery. Both observational studies and clinical trials over the last 10 years have shown that there is no other intervention that is as effective as metabolic and bariatric surgery for durable weight loss and comorbidity remission (eg, type 2 diabetes [T2D]).^3-6

There is limited evidence, however, that racial and ethnic groups of patients may not benefit in the same way from metabolic and bariatric surgery.⁷ What evidence does exist suggests that Black/African American/non-Hispanic Black (Black) patients consistently lose less weight than White/non-Hispanic White (White) patients.^7-11 The racial and ethnic disparities in surgical weight outcomes could be operation dependent, with many studies suggesting that disparities appear with Roux-en-Y gastric bypass (RYGB) and not sleeve gastrectomy (SG).^8,11-13

There is almost nothing known about the comparative differences in diabetes and safety outcomes by bariatric operation for different racial and ethnic groups. A recent systematic review found that there were no differences in racial and ethnic groups for T2D and other comorbidity outcomes after primarily RYGB operations.⁷ The few articles that have reported safety outcomes by race and ethnicity have shown that Black patients have a higher rate of adverse events than other racial and ethnic groups,^14,15 although the rates of adverse events are still within the range reported for all patients.¹⁶

There are several methodological concerns with the studies in this area. Most studies only include 1 bariatric operation, and if more than 1 is included, the comparisons do not use statistical designs that control for nonrandom assignment of operation. In addition, most have small samples of racial and ethnic groups, no long-term follow-up (≥3 years), and variable definitions of weight loss and comorbidity resolution.

In 2016, the Patient-Centered Outcomes Research Institute funded the PCORnet Bariatric Study (PBS) to demonstrate the utility of the National Patient-Centered Clinical Research Network (PCORnet) for promoting evidence-based and patient-centered health care and to compare the safety and effectiveness of RYGB and SG used in 11 geographically diverse, partnering networks in PCORnet.¹⁷ Several publications have detailed the findings from this study^18-20; however, questions about heterogeneity of treatment effects were not their main focus. The goal of the current study was to use existing data from the PBS cohort to test if there were comparative differences in weight loss; T2D remission, relapse, and change in hemoglobin A_1c (HbA_1c) level; and safety and utilization between SG and RYGB for different racial and ethnic groups up to 5 years after surgery. Based on prior research,^7-15 we hypothesized that RYGB would result in greater weight loss, higher rates of T2D remission, and higher complication rates than SG for all racial and ethnic groups.

The PBS was a retrospective observational comparative effectiveness cohort study. The PBS cohorts and protocol were previously described.^17-20 The PBS population consisted of adults and adolescents (aged 12-79 years) who underwent a primary (first nonrevisional) RYGB or SG operation between January 1, 2005, and September 30, 2015, in the participating health systems (n = 41). A waiver of informed consent was approved by the Institutional Review Board (IRB) of Kaiser Permanente Washington Health Research Institute (lead site), and this study was determined to be exempt from review by participating sites through individual IRB review or reliance agreements. These IRBs determined that written informed consent was not required because the study posed minimal risks to the participants and obtaining written consent for retrospective deidentified observational data would have been prohibitive for the study. Operations were identified using procedural codes that are provided in eTable 1 in Supplement 1. To be eligible for cohort inclusion, patients had at least 1 body mass index (BMI) measurement of 35 or greater (calculated as weight in kilograms divided by height in meters squared) recorded in their electronic health record in the year before the surgery and no prior revisional bariatric procedure codes (eTable 2 in Supplement 1), gastrointestinal cancer diagnosis, or fundoplasty operation during this period. In addition, patients could not have multiple conflicting bariatric procedure codes on the same day or any emergency department encounter on the day of the index operation.

From this general population, 3 cohorts were created for the analyses of each outcome: percentage total weight loss (%TWL), T2D remission and relapse, and safety and utilization. The details for the formation of each of these cohorts are provided in the eAppendix in Supplement 1. The analyses described in this article were added during an extension of the project and began by inviting all 41 clinical data network sites from the original study to participate in the analyses of race and ethnicity. Of these 41 sites, 25 agreed to participate (eTable 3 in Supplement 1 contains a list of sites for each analytic cohort). This study followed the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) reporting guidelines.²¹

All data were obtained from health systems participating in PCORnet by querying its Common Data Model.²² The details of the data used from the PCORnet Common Data Model have been published elsewhere.¹⁷ Data were available for patient demographics, procedures, diagnoses, weight, and laboratory values. Nine sites also had insurance claims and mortality data (eTable 3 in Supplement 1).

Race and ethnicity data were obtained from the electronic health records of participating health systems. Health systems were implementing meaningful use requirements²³ to collect self-reported race and ethnicity from their patients. To consolidate race and ethnicity categories for analyses, we followed national recommendations for mutually exclusive categories.^24,25 Regardless of the race category they endorsed, patients self-reporting Hispanic ethnicity were considered Hispanic. This convention has been used for data analyses across multiple diverse health systems²⁶ and aligns with recommendations from a US survey that found that Hispanic people considered themselves a race of people and not an ethnicity.²⁵ Final categories included: Black/African American/non-Hispanic Black (Black); Hispanic/Latino (Hispanic); and White/non-Hispanic White (White). Owing to small sample sizes, the other race and ethnicity category included non-Hispanic Asian/Alaskan Native or Native American/Native Hawaiian, Asian, Pacific Islander, and other and multiple races. We also included an unrecorded category for those patients who were missing race and ethnicity data.

Weight loss outcome was percentage total weight loss (%TWL) calculated as [(weight – weight at surgery) / (weight at surgery)] × 100,¹⁸ at each year of follow-up. Using this formula resulted in negative values for weight loss. The %TWL was calculated if a patient had at least 1 weight measurement at some time during the following periods: 1 year (6-18 months post-surgery), 3 year (30-42 months), and 5 year (54-66 months). This was done to represent clinical practice (weights are not collected exactly at annual periods) and to maximize data capture during follow-up.

T2D remission was defined as the first postsurgical occurrence of HbA_1c level less than 6.5% after at least 6 months without diabetes medication prescription orders.¹⁹ The occurrence of a HbA_1c level of 6.5% or greater and/or a prescription for diabetes medication after remission defined relapse.¹⁹ Change in HbA_1c level at 1, 3, and 5 years after surgery was calculated as HbA_1c level at follow-up – HbA_1c level at surgery. Using this formula resulted in negative values for HbA_1c improvement.

The primary outcome of interest for the safety and utilization analysis was time to the first occurrence of a gastrointestinal operation or intervention (O/I) across 5 years of follow-up identified using the International Classification of Diseases, Ninth Revision, and Current Procedural Terminology 4 procedure codes (available from authors on request).²⁰ Secondary outcomes across 5 years of follow-up included the time to first occurrence of revision or conversion operations (also included in the O/I primary outcome), endoscopy, and all-cause mortality. In addition, a composite indicator of any postoperative major adverse event was calculated using the first occurrence of the following events within 30 days of discharge from the index bariatric operation: death, venous thromboembolism, operation or intervention, and/or failure to be discharged from the hospital.²⁰ More detail about the creation of our safety and utilization outcomes can be found in the eAppendix in Supplement 1.

Confounders were chosen based upon our previous studies with the PBS cohort.^17-20 They included baseline weight, demographic variables (age, sex), Charlson/Elixhauser comorbidity score (range, −2 to 20; higher score was higher comorbidity burden),²⁷ number of days between baseline weight measurement and metabolic and bariatric operation, number of hospitalized days in the year before surgery, diastolic and systolic blood pressure, year of surgery, health system, and all comorbidities listed in eTable 4 in Supplement 1.

The focus for the weight loss and change in HbA_1c level analyses was a 3-way interaction including race and ethnicity (Black, White, Hispanic, other, unrecorded), operation type (RYGB, SG), and time since metabolic and bariatric surgery. For the time-to-event analyses, the focus was a 2-way interaction of race and ethnicity and operation type. For all analyses, there were also potential main effects for each variable such that differences within the categories in each variable might be significant regardless of the other variables of interest (eg, RYGB was different than SG across all races and ethnicities and time periods). All confounding variables listed above were used as adjustment variables in all analyses to control for confounding in the comparison of bariatric operations. All analyses were conducted using R, version 3.6.1 (R Foundation for Statistical Computing). P values were 2 sided, and P < .05 was considered significant. Data were analyzed from July 1, 2021, to January 17, 2022.

Each operation’s adjusted mean weight and change in HbA_1c level at 1, 3, and 5 years was estimated using a linear mixed-effects (random-effects) model with a cubic polynomial b-spline basis with 5 knots for flexible curves over time.²⁸ Separate models estimated a population-level curve for the mean weight and mean difference in HbA_1c level from the time of surgery to the end of the 5-year follow-up. These models included random effect terms for individuals (intercepts).

For clinical relevance in presentation, model-based mean weight and SE estimates were used to compute mean %TWL and corresponding 95% CIs and P values. Differences between operations in %TWL and HbA_1c level were calculated as RYGB − SG. For both %TWL and HbA_1c, change was negative (reduction after surgery); thus, if RYGB had greater change than SG, the difference would also be negative and favor RYGB.

Potential confounders included in our T2D outcome models were identical to the weight analysis, except that baseline weight was removed and we included baseline BMI and HbA_1c categories, indicators of diabetes medication classes, an indicator for insulin prescription, and the number of diabetes medications other than insulin. Cox proportional hazards models with an interaction term between operation type and race and ethnicity were used to estimate the adjusted hazard ratio (HR) for T2D remission and the adjusted cumulative proportion of individuals remitting at 1, 3, and 5 years for RYGB and SG. Similar analyses were conducted for T2D relapse. Additional details on the diabetes analyses are provided in the eAppendix in Supplement 1.

A Cox proportional hazards model was used to estimate the adjusted HR for time to our primary safety and utilization outcome, O/I, comparing RYGB and SG. Analyses of time-to-event for the first occurrence of revision or conversion, endoscopy, hospitalization, and all-cause mortality followed the same approach. Logistic regression was used for odds of the 30-day composite of major adverse events.

A total of 36 871 patients (mean [SD] age, 45.0 [11.7] years; 7123 male patients [19%]; 29 746 female patients [81%]) were included in the weight analysis. Patients identified with the following race and ethnic categories: 6891 Black (19%), 8756 Hispanic (24%), 19 645 White (53%), 826 other (2%), and 783 unrecorded (2%). Baseline characteristics of the patients included in each of the analytic cohorts (n = 36 871 for weight; n = 8407 for T2D; and n = 32 853 for safety and utilization) are presented in eTable 4 in Supplement 1 by operation. Compared with patients who had RYGB, patients who had SG were younger (weight cohort, 44.1 years vs 45.7 years), with a lower BMI (weight cohort, 49.0 vs 49.8), and were more likely to be Black (weight cohort, 3963 vs 2898) and Hispanic (weight cohort, 4688 vs 4068). Compared with patients who had RYGB, patients who had SG also had a lower rate of many comorbid conditions at the time of surgery, including depression (weight cohort, 27% [4385 of 16 158] vs 33% [6828 of 20 713]), diabetes (weight cohort, 28% [4512 of 16  158] vs 44% [9188 of 20 713]), dyslipidemia (weight cohort, 45% [7207 of 16 158] vs 52% [10 825 of 20 713]), eating disorders (weight cohort, 4% [708 of 16 158] vs 18% [3634 of 20 713]), gastroesophageal reflux disease (weight cohort, 32% [5118 of 16 158] vs 45% [9253 of 20 713]), hypertension (weight cohort, 54% [8695 of 16 158] vs 64% [13 224 of 20 713]), nonalcoholic fatty liver disease (weight cohort, 14% [2207 of 16 158] vs 26% [5344 of 20 713]), and sleep apnea (weight cohort, 40% [6509 of 16 158] vs 55% [11 405 of 20 713]). Other baseline characteristics were similar between operations.

Follow-up rates for weight at years 1, 3, and 5 are provided in eTable 5 in the Supplement 1. Follow-up rates varied as follows: year 1, 95% to 98%; year 3, 68% to 74%; and year 5, 46% to 64.5%. Differences in follow-up between SG and RYGB and between the main race and ethnicity categories (Black, Hispanic, White) were small.

Weight loss results (Table 1, Figure 1, and eFigures 1 and 2 and eTable 6 in Supplement 1) were larger for RYGB than SG in all years for Black, Hispanic, and White patients (eg, at year 5, the difference in mean %TWL between RYGB and SG was Black, −7.6%; 95% CI, −8.0 to −7.1; P < .001; Hispanic, −6.2%; 95% CI, −6.6 to −5.9; P < .001; White, −5.9%; 95% CI, −6.3 to −5.7; P < .001; and other, −6.0%; 95% CI, −7.3 to −4.7; P < .001). The differences between racial and ethnic groups in %TWL between SG and RYGB were small (1%-3% TWL).

Only Hispanic patients had higher T2D remission rates with RYGB when compared with SG (HR, 1.19; 95% CI, 1.08-1.32). None of the other racial and ethnic groups had differential remission with RYGB. Unlike remission, all race and ethnic groups experienced lower T2D relapse rates with RYGB compared with SG (Black [HR, 0.73; 95% CI, 0.56-0.95], Hispanic [HR, 0.75; 95% CI, 0.60-0.93], White [HR, 0.76; 95% CI, 0.63-0.92], and other [HR, 0.56; 95% CI, 0.32-0.97]) (Table 2, and eFigures 5 and 6 in Supplement 1).

Differences in mean HbA_1c level are presented in Table 1 and eFigures 3 and 4 and eTable 6 in Supplement 1. As with weight loss, differences were larger for RYGB than SG in all years for Black, Hispanic, and White patients (eg, at year 5, the difference between RYGB and SG favoring RYGB for lowering mean change in HbA_1c level was Black, −0.29; 95% CI, −0.51 to −0.08; P = .009; Hispanic, −0.45; 95% CI, −0.61 to −0.29; P < .001; White, −0.25; 95% CI, −0.40 to −0.11; P = .001). Mean differences in HbA_1c levels are also shown in eFigures 3 and 4 in Supplement 1).

Cumulative incidence (95% CIs) of adverse events at year 5 after each bariatric operation by racial and ethnic group are shown in eTable 7 in Supplement 1. Black (HR, 1.45; 95% CI, 1.17-1.79), Hispanic (HR, 1.48; 95% CI, 1.22-1.79), and White (HR, 1.34; 95% CI, 1.16-1.54) patients had higher risk of intervention or operation with RYGB than SG (Table 3 and Figure 2; other and undefined race and ethnicity are shown in eFigure 7 in Supplement 1). Black patients had significantly lower rates of revision or conversion with RYGB compared with SG (HR, 0.60; 95% CI, 0.41-0.88) (Table 3; eFigure 8 in Supplement 1).

Hispanic patients had the highest risk of endoscopy with RYGB compared with SG (HR, 2.88; 95% CI, 2.34-3.54), followed by White (HR, 2.06; 95% CI, 1.80-2.35) and Black (HR, 1.88; 95% CI, 1.54-2.30) patients (Table 3; eFigure 9 in Supplement 1). Hispanic patients had the highest risk of hospitalization with RYGB compared with SG (HR, 1.37; 95% CI, 1.24-1.51; P < .001) followed by Black patients (HR, 1.30; 95% CI, 1.16-1.45; P < .001) (Table 3; eFigure 10 in Supplement 1). Hispanic patients also had higher risk of all-cause mortality following RYGB compared with SG (HR, 2.41; 95% CI, 1.24-4.70; P = .01) (Table 3; eFigure 11 in Supplement 1) and higher odds of experiencing a 30-day major adverse event with RYGB than SG (adjusted odds ratio, 1.92; 95% CI, 1.38-2.68; P < .001) (Table 3).

In this large, multicenter, comparative effectiveness research study, there was greater improvement in %TWL and HbA_1c level for RYGB than SG in all racial and ethnic groups across all years of follow-up, and the variability of comparative effectiveness of operations across racial and ethnic groups for %TWL and HbA_1c level was clinically small. The 3 largest racial and ethnic groups all had higher rates of postsurgical intervention or operation with RYGB compared with SG (Black, HR, 1.45; Hispanic, HR, 1.48; White, HR, 1.34) and higher rates of endoscopy for RYGB compared with SG (Black, HR, 1.88; Hispanic, HR, 2.88; White, HR, 2.06). However, the differences in safety and utilization outcomes for Black and Hispanic patients having RYGB as compared with SG were much larger than those seen in weight and diabetes outcomes. Black and Hispanic patients had higher rates of hospitalization after they had RYGB compared with SG (HR, 1.30 and 1.37, respectively), and Hispanic patients had significantly greater risk of all-cause mortality (HR, 2.41) and higher odds of a 30-day major adverse event (odds ratio, 1.92) for RYGB compared with SG.

To our knowledge, few studies have compared metabolic and bariatric operations directly for health and safety outcomes in different racial and ethnic groups; therefore, it is difficult to compare our findings with this literature. A recent systematic review⁷ found that Black patients lost less weight than other racial and ethnic groups; however, this was primarily with RYGB, and most studies neither compared RYGB with other operations nor had long-term follow-up. This same review⁷ did not find clear support for racial and ethnic differences in comorbidity resolution, and no studies reported comparative data for bariatric and metabolic operations within racial and ethnic groups for these outcomes. Only 1 out of 13 studies on T2D remission found differences between Black and White patients.⁷

Our study findings suggest that patients from the 3 largest racial and ethnic groups all had a higher risk of intervention or operation after RYGB compared with SG with similar findings for risk of endoscopy. Many publications, including those from the PCORnet Bariatric Study,²⁰ have shown higher rates of complications with RYGB when compared with SG.^16,29 Recent publications have documented Black patients, in particular, had higher rates of adverse events immediately after primarily RYGB operations.^12-15,30,31 We have extended these findings by adding SG operations and Hispanic patients, and our findings also suggest that the risk of hospitalization for Black and Hispanic patients who had RYGB was higher than for those who had SG. Of particular concern was that Hispanic patients who had RYGB also had higher risk of all-cause mortality and higher odds of a 30-day adverse event than those who had SG.

There could be several reasons why Black and especially Hispanic patients had higher rates of complications with RYGB than SG. Given that there are higher risks with RYGB generally,^32-34 the findings for Black and Hispanic patients could be exacerbated by poor access to or coverage for postoperative health care.³⁵ In addition, these groups of patients often have limited access to healthy foods and physical activity, which are critical to weight loss maintenance and the resulting disease remission regardless of the way people lose weight.^36,37 In addition, there are many other health system and societal factors shown to determine racial and ethnic disparities in health outcomes generally, such as the number of racial and language concordant physicians available to patients,^38,39 structural racism⁴⁰ and discrimination⁴¹ during clinical care encounters, and immigration status.⁴² Future research should be done to understand the specific barriers (eg, structural racism and inequities in social determinants of health) that racial and ethnic groups may experience in accessing effective short- and long-term postoperative care necessary to prevent potential complications after metabolic and bariatric surgery.

Owing to the limited data available in the PCORnet data marts, our study was not designed to address the mechanisms that might explain the differential effect of operation in certain racial and ethnic groups for weight loss, T2D outcomes, and safety and utilization. Other published work has suggested that the differences between race and ethnicity and operation type could be the result of differences in age and comorbidity burden between patients who have SG and RYGB.¹⁰ In our descriptive findings (eTable 3 in Supplement 1), we found that patients who had SG were more likely to be Black or Hispanic and were younger. However, it is unlikely these factors explained the differences we reported because we adjusted for age and comorbidity burden in our comparative analyses. This highlights the importance of adjustment for the nonrandom assignment of metabolic and bariatric operation when studying comparative outcomes. This was 1 of the main limitations noted by the recent systematic review on racial and ethnic differences in metabolic and bariatric surgery outcomes.⁷

In this comparative effectiveness research study, findings suggest that the variability of comparative effectiveness of operations across racial and ethnic groups for %TWL and HbA_1c level was clinically small; however, differences for safety and utilization outcomes were both clinically and statistically significant for Black and Hispanic patients who had RYGB as compared with SG. Reasons for these findings are unclear and could be related to the availability of health care for these patients rather than the bariatric operations themselves. Our findings can inform shared decision-making conversations about the choice of bariatric operation that could be tailored for different racial and ethnic groups of patients.

Accepted for Publication: June 2, 2022.

Published Online: August 31, 2022. doi:10.1001/jamasurg.2022.3714

Corresponding Author: Karen J. Coleman, PhD, Department of Research and Evaluation, Kaiser Permanente Southern California, 100 S Los Robles, 6th Floor, Pasadena, CA 91101 (karen.j.coleman@kp.org).

Author Contributions: Dr Arterburn and Mr Wellman had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Coleman, Wellman, Fitzpatrick, Conroy, Lewis, Coley, McTigue, Tobin, Desai, Clark, Toh, Horgan, Duke, Anau, Michalsky, Cook, Arterburn, Apovian.

Acquisition, analysis, or interpretation of data: Coleman, Wellman, Fitzpatrick, Conroy, Hlavin, Lewis, Coley, McTigue, Tobin, McBride, Clark, Toh, Sturtevant, Horgan, Williams, Anau, Horberg, Michalsky, Arterburn, Apovian.

Drafting of the manuscript: Coleman, Wellman, Desai, Sturtevant, Arterburn.

Critical revision of the manuscript for important intellectual content: Fitzpatrick, Conroy, Hlavin, Lewis, Coley, McTigue, Tobin, McBride, Desai, Clark, Toh, Horgan, Duke, Williams, Anau, Horberg, Michalsky, Cook, Apovian.

Statistical analysis: Wellman, Coley, Toh, Cook.

Obtained funding: McTigue, Arterburn.

Administrative, technical, or material support: Lewis, Coley, Tobin, Sturtevant, Anau, Michalsky, Apovian.

Supervision: Coley, McBride, Duke, Anau.

Conflict of Interest Disclosures: Dr Coleman reported receiving grants from Patient-Centered Outcomes Research Institute (PCORI), the National Institutes of Health (NIH), Janssen, and the US Food and Drug Administration outside the submitted work. Mr Wellman reported receiving grants from PCORI during the conduct of the study. Dr Fitzpatrick reported receiving grants from PCORI and the NIH and financial support (salary) from WW (formerly Weight Watchers). Dr Lewis reported receiving honoraria from National Committee for Quality Assurance (NCQA) for serving as a faculty member on a continuing medical education video about the treatment of obesity outside the submitted work. Dr McTigue reported receiving grants from the University of Pittsburgh research contract from PCORI during the conduct of the study. Dr Tobin reported receiving grants from the NIH National Heart, Lung, and Blood Institute, US Department of Health and Human Services Administration for Community Living, and PCORI during the conduct of the study. Dr Clark reported receiving grants from Johns Hopkins School of Medicine during the conduct of the study. Dr Toh reported receiving grants from PCORI during the conduct of the study. Dr Williams reported receiving grants from PCORI during the conduct of the study. Dr Anau reported receiving grants from PCORI during the conduct of the study. Dr Horberg reported receiving grants from PCORI during the conduct of the study. Dr Michalsky reported receiving honorarium from and being a shareholder of Intuitive Surgical outside the submitted work. Dr Cook reported receiving grants from PCORI, the NIH, and the Centers for Disease Control and Prevention outside the submitted work. Dr Arterburn reported receiving grants from PCORI, the NIH, and Sharecare and receiving travel support from the World Congress for Interventional Therapy for Diabetes and the International Federation for the Surgery of Obesity and Metabolic Disorders Latin America Chapter outside the submitted work. Dr Apovian reported receiving grants from Orexigen, Aspire Bariatrics, GI Dynamics, Myos, Takeda, the Vela Foundation, the Dr. Robert C. and Veronica Atkins Foundation, Coherence Lab, Energesis, the NIH, and PCORI; advisory board fees from Altimmune, Cowen and Company, Gelesis, L-Nutra, NeuroBo Pharmaceuticals, Nutrisystem, Zafgen, Sanofi-Aventis, Orexigen, Novo Nordisk, GI Dynamics, Takeda, Scientific Intake, Pain Script Corporation, Riverview School, Rhythm Pharmaceuticals, Xeno Biosciences, Eisai, EnteroMedics, and Bariatrix Nutrition; and having a previous ownership of stock in Science-Smart outside the submitted work. No other disclosures were reported.

Funding/Support: The PCORnet Study reported in this publication was conducted using PCORnet, the National Patient-Centered Clinical Research Network. PCORnet has been developed with contract OBS-1505-30683 funding from the Patient-Centered Outcomes Research Institute.

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Group Information: See Supplement 2.

Disclaimer: The views expressed in this manuscript are solely the responsibility of the authors and do not necessarily represent the views of other organizations participating in, collaborating with, or funding PCORnet or of PCORI.

Additional Contributions: We thank the patients who contributed data to this study through their electronic medical records without whom none of this would have been possible.