This was a single-center, randomized, open-label, 3-armed, parallel-group, pragmatic trial conducted at Cleveland Clinic’s Department of Endocrinology, Diabetes, and Metabolism. The study was conducted in accordance with the International Conference on Harmonisation Guideline for Good Clinical Practice, the principles of the Declaration of Helsinki, and all applicable local ethical and legal requirements. The Cleveland Clinic institutional review board approved the study protocol (approval number IRB 20-648). All participants provided written informed consent. Study data were collected and managed using REDCap electronic data capture tools hosted at Cleveland Clinic [10, 11].

Eligible persons were employees of Cleveland Clinic, and their partners, who had health coverage through Cleveland Clinic, Medical Mutual, or Bravo Health, and who had a BMI ≥ 30 kg/m2, a diagnosis of type 2 diabetes, and an HbA1c > 7.5% within 90 days preceding screening. Major exclusion criteria included a glomerular filtration rate < 30 mL/min/1.73 m2; current glucocorticoid therapy; use of AOM within the previous 3 months; medical history that would contraindicate use of AOM (e.g., personal/family history of medullary thyroid carcinoma or multiple endocrine neoplasia syndrome type 2); history of acute pancreatitis or severe disease of the liver or digestive tract; history of bariatric or metabolic surgery; visit with an endocrinologist within the past year for treatment of type 2 diabetes; and prior participation in the Cleveland Clinic Endocrinology and Metabolism Institutes Integrated Weight Management Program.

Participants were randomized 1:1:1 to one of three management arms for a planned duration of 24 months: “Usual-Care” group (glucose-centric approach), or one of two obesity-centric management strategies: participation in a WMP plus adjunctive AOM (“WMP + AOM” group), or participation in a WMP without AOM (“WMP-Only” group). All patients remained under the care of their usual primary care provider and any other regular physicians.

Participants in the Usual-Care arm were managed for type 2 diabetes, hypertension, and hyperlipidemia according to a traditional approach which entailed an initial consultation with an endocrinologist and follow-up visits every 3 months.

Cleveland Clinic’s Endocrinology and Metabolism Institute’s Integrated Weight Management Program consisted of an initial 1:1 personal evaluation by an obesity medicine specialist to establish a plan of care, then participation in shared medical appointments (SMAs) (4–5 participants per session) once per month during year 1 and once every 3 months during year 2. The SMAs (see electronic Supplementary Materials) were run by an obesity specialist and a nutritionist and were approximately 75–90 min in length. The five main areas reviewed at every session included nutrition, physical activity, appetite control, sleep issues, and anxiety/depression/stress.

Participants in the WMP + AOM arm were enrolled in the same weight management program, but were also eligible to initiate treatment with one of five US Food and Drug Administration (FDA)-approved medications for chronic obesity treatment (orlistat, phentermine/topiramate, naltrexone/bupropion, liraglutide 3.0 mg, or semaglutide 2.4 mg). Choices of medication and dosage were at the discretion of the investigator, according to routine clinical practice. Participants could discontinue the medication at any time but were encouraged to initiate treatment with a different AOM. AOMs were dispensed by one of Cleveland Clinic’s ambulatory pharmacies; to simulate real-world conditions, participants were charged a fee commensurate with a typical retail pharmacy co-pay. Use of weight-loss medications other than the five FDA-approved medications was not allowed.

Participants assigned to the Usual-Care and WMP-Only arms were not allowed to use any medication for the primary intent of weight loss. In all study arms, intensification of anti-diabetes, hypertension, and/or hyperlipidemia therapy was allowed as indicated by HbA1c, blood pressure, and lipid profile values, at the discretion of the investigator according to current practice standards.

Body weight, HbA1c, blood pressure, serum low-density and high-density lipoprotein (LDL and HDL) and triglycerides, and blood pressure were assessed at screening/baseline, and at study months 6, 12, and 24.

Primary outcomes were percentage weight loss and change in HbA1c from baseline to month 12. The primary aims were to demonstrate noninferiority of WMP + AOM versus WMP-Only for both outcomes, and to demonstrate superiority in ≥ 1 of the outcomes for WMP + AOM versus WMP-Only. In addition, we aimed to demonstrate similar noninferiority in both outcomes and superiority in ≥ 1 of the outcomes for comparisons between both obesity-centric (weight management program) versus glycemic-centric (usual care) approaches.

Secondary outcomes included percentage of participants achieving > 5% weight loss (considered clinically relevant [12]) or HbA1c < 7%; changes in serum LDL, HDL, and triglycerides; and percentage of participants with blood pressure < 140/90 mmHg at month 12.

Prior research by our group [13] suggested that a change in HbA1c of 0.4–0.5% was expected to accompany changes in body weight of around 4–5%. Power calculations were performed using a simulation approach [14]. On the basis of the noted assumptions and our prior research findings, a sample size of 300 participants (100 per arm) was planned, assuming 20% of patients would not complete the evaluation at 1 year, leaving 80 completers per group, and assuming a weight loss standard deviation of 1.75%. This sample size would provide 82% power to detect noninferiority in both HbA1c and weight loss, and to detect superiority in weight loss.

The primary analysis was performed using the intent-to-treat group, which included all randomized participants. Primary end points were analyzed using linear mixed-effect models, with baseline body weight and HbA1c as covariates. Time points common to all study arms (months 3, 6, 9, and 12) were used for analysis; time–group interactions were included in the models. The method of joint hypothesis test was applied for this analysis. Noninferiority was tested at the 0.05 level at 1 year and performed pairwise comparison with Bonferroni-adjusted significance levels and P values. The noninferiority regions were set to be 1% for weight loss change and 0.5% for HbA1c. When both primary end points were noninferior, superiority testing at the 0.025 overall error level with Bonferroni adjustment for each end point at 1 year was then performed.

Between-group percentages of participants achieving weight loss > 5%, HbA1c treatment target < 7%, and blood pressure < 140/90 mmHg were compared through separate logistic regression models by odds ratio (OR), with relevant baseline values used as covariates. Linear mixed-effect models were used to analyze changes in serum LDL, HDL, and triglyceride levels, with baseline levels as covariates. Superiority testing at 0.05 overall error level with Bonferroni adjustment at 1 year was performed.

For primary endpoints (changes in HbA1C levels and weight percentage) and secondary endpoints (changes in serum LDL, HDL, triglycerides) where one-sided non-inferiority tests or superiority tests within a significance threshold of 0.05 were performed, 90% Bonferroni-adjusted confidence intervals were presented to ensure the upper confidence limits reflected a one-sided 95% confidence. Data were managed and analyzed using R software (version 4.3.1; Vienna, Austria).