Role-Playing–Based Physician Training and Metabolic Outcomes Among Patients With Type 2 Diabetes: A Cluster Randomized Clinical Trial

Yifei Zhang; Ying Peng; Yufei Chen; Tingyu Ke; Fengmei Xu; Shengli Wu; Yuancheng Dai; Lin Sun; Qidong Zheng; Zhuomeng Hu; Qijuan Dong; Juan Shi; Xueyi Wu; Yu Shi; Rong Tang; Yubo Sha; Rongyue Chen; Bin Xu; Shu Li; Lianyong Liu; Mingdian Gao; Dong Zhao; Qinghua Yi; Zhiqiang Kang; Weiqing Wang

doi:10.1001/jamanetworkopen.2026.27376

. 2026 Aug 6;9(8):e2627376. doi: 10.1001/jamanetworkopen.2026.27376

Role-Playing–Based Physician Training and Metabolic Outcomes Among Patients With Type 2 Diabetes

A Cluster Randomized Clinical Trial

Yifei Zhang ^1,², Ying Peng ^1,², Yufei Chen ^1,², Tingyu Ke ³, Fengmei Xu ⁴, Shengli Wu ⁵, Yuancheng Dai ⁶, Lin Sun ^1,², Qidong Zheng ⁷, Zhuomeng Hu ^1,², Qijuan Dong ⁸, Juan Shi ^1,², Xueyi Wu ⁹, Yu Shi ¹⁰, Rong Tang ¹¹, Yubo Sha ¹², Rongyue Chen ¹³, Bin Xu ¹⁴, Shu Li ¹⁵, Lianyong Liu ¹⁶, Mingdian Gao ¹⁷, Dong Zhao ¹⁸, Qinghua Yi ¹⁹, Zhiqiang Kang ²⁰, Weiqing Wang ^1,^2,^✉, for the Chinese Endocrinologists Health Education Study (CREATION) group

¹Department of Endocrine and Metabolic Diseases, Shanghai Institute of Endocrine and Metabolic Diseases, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China

²Shanghai National Clinical Research Center for Metabolic Diseases, Key Laboratory for Endocrine and Metabolic Diseases of the National Health Commission of the PR China, Shanghai National Center for Translational Medicine, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China

³Department of Endocrinology, The Second Affiliated Hospital of Kunming Medical University, Kunming, China

⁴Department of Endocrine and Metabolic Diseases, Hebi Coal (Group) Ltd, General Hospital, Hebi, China

⁵Endocrine and Metabolic Center, Karamay Hospital of Integrated Traditional Chinese and Western Medicine (Karamay People’s Hospital), Karamay, China

⁶Department of Endocrinology, Sheyang Diabetes Hospital, Yancheng, China

⁷Department of Endocrinology, Yuhuan Second People’s Hospital, Taizhou, China

⁸Department of Endocrine and Metabolic Diseases, People’s Hospital of Zhengzhou, Zhengzhou, China

⁹Department of Endocrinology, The Second People’s Hospital of Guiyang (Jinyang Hospital)/The Affiliated Jinyang Hospital of Guizhou Medical University, Guiyang, China

¹⁰Department of Endocrinology, Qidong People’s Hospital/Qidong Liver Cancer Institute/Affiliated Qidong Hospital of Nantong University, Nantong, China

¹¹Department of Endocrinology, Panzhou No. 2 People’s Hospital, Liupanshui, China

¹²Department of Endocrine and Metabolic Diseases, People’s Hospital of Dali City, Dali, China

¹³Department of Endocrinology, Xuchang Hongyue Diabetes Hospital, Xuchang, China

¹⁴Department of Endocrinology, Biyang County People’s Hospital, Zhumadian, China

¹⁵Department of Endocrinology, Huizhou Central People’s Hospital, Huizhou, China

¹⁶Department of Endocrinology, Shanghai Punan Hospital of Pudong New District, Shanghai, China

¹⁷Department of Endocrinology, Changyi People’s Hospital, Weifang, China

¹⁸Center for Endocrine Metabolism and Immune Diseases, Beijing Luhe Hospital, Capital Medical University, Beijing, China

¹⁹Department of Endocrinology and Nephrology, The People’s Hospital of Guandu District, Kunming, China

²⁰Department of Endocrinology, Zhengzhou Central Hospital, Zhengzhou University, Zhengzhou, China

Accepted for Publication: June 15, 2026.

Published: August 6, 2026. doi:10.1001/jamanetworkopen.2026.27376

Open Access: This is an open access article distributed under the terms of the CC-BY-NC-ND License, which does not permit alteration or commercial use, including those for text and data mining, AI training, and similar technologies. © 2026 Zhang Y et al. JAMA Network Open.

^✉

Corresponding Author: Weiqing Wang, MD, Department of Endocrine and Metabolic Diseases, Shanghai Institute of Endocrine and Metabolic Diseases, Ruijin Hospital, Shanghai Jiaotong University School of Medicine, 197 Ruijin 2nd Rd, Shanghai 200025, China (wqingw@shsmu.edu.cn).

Author Contributions: Dr Wang had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs Zhang, Peng, Y. Chen, Ke, Ms F. Xu, Mr S. Wu, Mr Dai, and Dr Sun contributed equally to this work.

Concept and design: Zhang, Wang.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Zhang, Peng, Y. Chen, Sun, Hu.

Critical review of the manuscript for important intellectual content: Zhang, Y. Chen, Ke, F. Xu, S. Wu, Dai, Zheng, Dong, J. Shi, X. Wu, Y. Shi, Tang, Sha, R. Chen, B. Xu, Li, Liu, Gao, Zhao, Yi, Kang, Wang.

Statistical analysis: Peng, Y. Chen, Ke, F. Xu, S. Wu, Dai, Sun, Hu.

Obtained funding: Zhang, Wang.

Administrative, technical, or material support: Y. Chen, Zhao, Wang.

Supervision: Wang.

Conflict of Interest Disclosures: None reported.

Funding/Support: This research was supported by the Noncommunicable Chronic Diseases–National Science and Technology Major Project (grant 2023ZD0508100), the National Key Research and Development Program of China (grant 2018YFC1311800), Capacity Building for Multidisciplinary Cooperation in Diagnosis and Treatment Of Major Metabolic Diseases (grant Z155080000004), and the Leader Project of the Oriental Talent Program in 2022 (grant 153).

Role of the Funder/Sponsor: The funders 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: The members of the Chinese Endocrinologists Health Education Study (CREATION) group are listed in Supplement 3.

Data Sharing Statement: See Supplement 4.

Additional Contributions: We thank all patients with diabetes who participated in this study and the physicians and staff at the participating Metabolic Management Centers for their contributions to data collection and study implementation.

^✉

Corresponding author.

PMCID: PMC13448795 PMID: 42560674

Key Points

Question

Was role-playing–based physician training associated with superior metabolic outcomes among patients with type 2 diabetes compared with regular physician training?

Findings

In this cluster randomized clinical trial including 205 physicians and 2017 patients with type 2 diabetes, patients whose physicians received role-playing–based intensive training had a significantly better hemoglobin A_1c control rate at 6 months compared with those whose physicians received regular training (58.0% vs 42.9%).

Meaning

Role-playing–based intensive training for physicians could be a feasible strategy to narrow the guideline-implementation gap in diabetes care.

Abstract

Importance

Global glycemic control in diabetes is suboptimal, with persistent gaps between guideline recommendations and clinical practice.

Objective

To evaluate the effectiveness of a role-playing–based intensive training program for physicians in improving glycemic control among patients with type 2 diabetes.

Design, Setting, and Participants

This was an open-label, 2-arm parallel, cluster randomized clinical trial conducted at 205 centers in China, with enrollment from February 13 to September 19, 2023, and 12-month follow-up completed on October 25, 2024. In phase 1, 205 physicians from qualified diabetes centers across China were enrolled from February 13 to April 29, 2023, and randomly assigned to the intensive training group (n = 103) and the regular training group (n = 102). In phase 2, 2017 patients with type 2 diabetes were enrolled from February 28 to September 19, 2023. Data were analyzed from January 1 to June 1, 2025.

Interventions

One-week on-site role-playing–based intensive training (involving simulating patient roles in experiencing diabetes-related health examinations, complication scenarios, lifestyle intervention, and face-to-face conversations) vs regular training.

Main Outcomes and Measures

The proportion of patients achieving a hemoglobin A_1c (HbA_1c) level of less than 7.0% at 6 months. The primary analysis was based on the intention-to-treat principle.

Results

A total of 205 physicians (mean [SD] age, 36.2 [5.1] years; 155 women [75.6%]) and 2017 patients (mean [SD] age, 53.0 [7.0] years; 713 women [35.3%] and 1304 men [64.7%]) were enrolled, with 1009 patients in the intensive training group and 1008 patients in the regular training group. At 6 months, a significantly greater proportion of patients in the intensive training group achieved an HbA_1c level less than 7.0% compared with the regular training group (58.0% [476 of 820] vs 42.9% [351 of 818]), with an adjusted between-group difference of 16.6% (95% CI, 7.2%-25.7%; P < .001). At 12 months, the difference between the intensive training group and the regular training group remained significant (60.9% [502 of 824] vs 44.6% [371 of 832]; adjusted between-group difference, 17.0% [95% CI, 7.0%-26.8%]; P < .001). The intensive group had greater reductions in body mass index (adjusted between-group difference, −0.3 [95% CI, −0.5 to −0.1]), waist circumference (adjusted between-group difference, −1.4 cm [95% CI, −2.1 to −0.8 cm]), fasting glucose (adjusted between-group difference, −6.3 mg/dL [95% CI, −10.7 to −1.9 mg/dL]), and systolic blood pressure (adjusted between-group difference, −1.5 mm Hg [95% CI, −2.9 to −0.1 mm Hg]). Serious adverse events were comparable between groups (intensive group, 2.6% [26 of 1009]; regular group, 2.4% [24 of 1008]).

Conclusions and Relevance

In this cluster randomized clinical trial of role-playing–based intensive training for physicians managing type 2 diabetes, the intervention significantly improved patient glycemic control. These findings suggest a feasible strategy to narrow the guideline-implementation gap in diabetes care.

Trial Registration

ClinicalTrials.gov Identifier: NCT05715307

This randomized clinical trial evaluates the effectiveness of a role-playing–based intensive training program for physicians compared with regular physician training in improving glycemic control among patients with type 2 diabetes.

Introduction

It is estimated that diabetes will affect approximately 1.3 billion people by 2050.^1,2 Despite evidence-based guidelines advocating comprehensive management strategies, a persistent gap between clinical recommendations and actual implementation remains a critical barrier to improving outcomes.^3,4 For instance, China has the largest number of adults with diabetes worldwide.^1,4 However, only 32.9% of patients with diabetes in China have received treatment, among whom only 50.1% have achieved effective glycemic control.⁵ Similarly, a global pooled analysis encompassing data from 200 countries and territories estimated that 59% of adults aged 30 years or older with diabetes did not receive treatment in 2022.⁶ Furthermore, a cross-sectional analysis including pooled data from 55 national surveys showed that less than 10% of adults with diabetes were comprehensively treated to guideline standards based on 6 core clinical treatment indicators.⁷ These findings underscore systemic challenges in translating knowledge into practice.^7,8

Addressing this challenge requires multifaceted strategies, and continuous education for health care professionals is fundamental.^9,10,11 Conventional training programs predominantly relying on passive knowledge dissemination are not sufficient to foster long-term adherence to guidelines in chronic disease management. Although innovative approaches—such as the use of online platforms—have enhanced the flexibility and efficiency of medical education, their inability to simulate actual clinical interactions and cultivate empathy may hinder the development of patient-centered care skills.^12,13 Therefore, effective implementation of training programs that integrate experiential learning with adaptable online methods is needed.^14,15,16,17

To address these limitations, we designed the Chinese Endocrinologists Health Education Study (CREATION), a multicenter, cluster randomized clinical trial to evaluate a novel role-playing–based intensive training program for physicians managing patients with type 2 diabetes. This intervention uniquely combines immersive offline physician-as-patient simulation with sustained digital reinforcement through a nationwide standardized care platform, with the dual objectives of enhancing physicians’ adherence to clinical practice guidelines and improving metabolic control outcomes among their patients. Accordingly, CREATION was designed to evaluate whether a role-playing–based intensive training program for physicians could improve metabolic outcomes among patients with type 2 diabetes compared with regular training.

Methods

Study Design and Participants

CREATION was a multicenter, cluster randomized, 2-arm parallel clinical trial conducted in China, with enrollment from February 13 to September 19, 2023, and 12-month follow-up completed on October 25, 2024. All study centers were qualified centers from the Metabolic Management Center (MMC) nationwide project, initiated in 2016 to standardize and enhance the management of diabetes and metabolic diseases across China.^10,18 The detailed study protocol and statistical analysis plan are presented in Supplement 1. This study is reported following the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline. This study was conducted without the involvement of patients or the public in its design, conduct, or reporting, and it was conducted in accordance with the Declaration of Helsinki.¹⁹ The protocol was approved by the Ethics Committee of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine, and all participating physicians and patients provided written informed consent.

The trial consisted of 2 phases (trial protocol in Supplement 1 and the eFigure in Supplement 2). In phase 1, 205 eligible physicians (1 per study center; Figure 1; eFigure in Supplement 2) were randomly assigned at 1:1 ratio to either the intensive training group or the regular training group. Prior to initiating the assigned intervention, both groups received online training covering diabetes management workflow, the study protocol, and the use of the MMC platform. The intensive training group then underwent an additional 1-week on-site role-playing–based training program in comprehensive diabetes management. Subsequently, each physician was required to enroll and follow up with a mean of 10 patients with type 2 diabetes according to prespecified inclusion and exclusion criteria in phase 2, with a target total enrollment of approximately 2000 patients (1000 per group). Patients were managed and followed up for 12 months, with all participating physicians providing care according to the MMC management standards and current Chinese diabetes management guidelines.^10,20,21

Flow diagram of center, physician, and participant enrollment and follow-up. Top center rectangle: 248 Centers from M M C network invited. A rightward arrow points to a light gray rectangle: 36 Centers declined to participate. A vertical connector continues to a large light gray oval centered: 212 Centers from M M C network participated and randomized with 1 physician from each center. Two diagonal connectors branch to two parallel arms. Left arm: light gray rectangle: 108 Physicians randomized to intensive training group. A rightward arrow points to a light gray box listing attrition: 5 Discontinued; 4 Transferred to other departments; 1 Enrolled 0 participants with type 2 diabetes. A downward arrow continues to a light gray rectangle: 103 Physicians enrolled. Downward arrow to light gray rectangle: 1497 Participants with type 2 diabetes screened. A rightward arrow points to a light gray exclusion box: 488 Excluded; 465 Did not meet inclusion criteria; 23 Left the original center. Downward arrow to light gray rectangle: 1009 Participants enrolled. Downward arrow to light gray rectangle: 189 Missing 6-mo visit. Downward arrow to light gray rectangle: 820 Participants completed follow-up at 6 mo a. Downward arrow to light gray rectangle: 185 Missing 12-mo visit. Downward arrow to light gray rectangle: 824 Participants completed follow-up at 12 mo a. Right arm: light gray rectangle: 104 Physicians randomized to regular training group. A rightward arrow points to a light gray box listing attrition: 2 Discontinued; 2 Enrolled 0 participants with type 2 diabetes. Downward arrow to light gray rectangle: 102 Physicians enrolled. Downward arrow to light gray rectangle: 1512 Participants with type 2 diabetes screened. A rightward arrow points to a light gray exclusion box: 504 Excluded; 482 Did not meet inclusion criteria; 22 Left the original center. Downward arrow to light gray rectangle: 1008 Participants enrolled. Downward arrow to light gray rectangle: 190 Missing 6-mo visit. Downward arrow to light gray rectangle: 818 Participants completed follow-up at 6 mo a. Downward arrow to light gray rectangle: 176 Missing 12-mo visit. Downward arrow to light gray rectangle: 832 Participants completed follow-up at 12 mo a. — MMC indicates Metabolic Management Center.

^aPatients with hemoglobin A_1c data at the 6- or 12-month follow-up visit. Participants missing the 6-month visit were not withdrawn and could still complete the 12-month visit; each number is relative to enrolled participants.

In phase 1, eligible physicians were required to be 25 to 50 years of age, able to complete the assigned training program, and engage in a 1-year patient follow-up period. Those from private, community, or rural hospitals or centers with fewer than 30 new patients with diabetes monthly were excluded.

In phase 2, trained physicians were required to enroll participants with type 2 diabetes who met the following criteria: aged 40 to 65 years; a body mass index (BMI; calculated as weight in kilograms divided by height in meters squared) of 24.0 to 35.0; hemoglobin A_1c (HbA_1c) level of 7.5% to 10.0% (to convert to proportion of total hemoglobin, multiply by 0.01), and fasting blood glucose level of 144.1 to 239.6 mg/dL (to convert to millimoles per liter, multiply by 0.0555); diabetes duration less than 10 years; inadequate glycemic control while using 1 to 2 types of noninsulin hypoglycemic medications without regimen changes for at least 2 months; and visiting the MMC center for the first time. Patients with type 1, monogenic, or secondary diabetes; recent acute diabetic complications; severe cardiovascular, hepatic, or kidney dysfunction; mental disorders, recent insulin use, drug abuse, sexually transmitted diseases, active infections, pregnancy, lactation, or concurrent trial enrollment were excluded. Other conditions judged ineligible by investigators were also ruled out. Full eligibility criteria are provided in Supplement 1.

Randomization and Masking

Participating physicians were randomized using a computer-generated randomization list prepared by an independent statistician with no involvement in participant recruitment, intervention delivery, or outcome assessment to ensure allocation concealment. The allocation list was stored in a secured system accessible only to the independent statistician until final database lock. Physicians were randomized 1:1 to either the intensive or regular training group, stratified by hospital grade (tertiary or secondary) and management ranking of MMCs (high [top 35%], middle [middle 35%], low [bottom 30%]). Due to the nature of the cluster randomized design and intervention program in this study, participating physicians and clinical data collectors were unblinded to the group allocation. The statisticians remained blinded prior to the database lock by conducting dry runs using dummy codes.

Intervention

In phase 1, physicians in the intensive training group further received an additional 1-week on-site experiential training at the leading MMC (Ruijin Hospital, Shanghai, China). The training was delivered in small-group sessions of 10 to 12 physicians each and led by a multidisciplinary team including certified diabetologists, diabetes educators, dietitians, and exercise specialists. It focused on role-playing simulations across 4 core modules—examinations, experiencing, interviewing, and feedback (eTable 1 in Supplement 2). During sessions, physicians experienced diabetic examinations and complications, attended nutrition and exercise workshops, engaged in face-to-face conversations with real patients, discussed clinical dilemmas, and assessed training efficacy via questionnaires, enhancing patient-centered empathy and improving clinical guideline implementation among physicians.

In phase 2, each physician consecutively enrolled and followed up with approximately 10 patients with type 2 diabetes, targeting a total enrollment of about 1000 patients in each group. Patients underwent a screening visit (visit 0) to assess eligibility criteria, then enrolled patients were expected to visit their physician at baseline (visit 1), month 1 (±1 week; visit 2), and months 3, 6, 9, and 12 (±2 weeks for each; visits 3-6). At each visit, protocol-specified assessments were conducted to evaluate changes in patients’ clinical parameters, lifestyle behaviors, and medication regimens (see Supplement 1 for detailed schedule of assessments).

Before training initiation and after the 12-month patient follow-up period, standardized questionnaires were administered to both groups of physicians to assess their general profiles, lifestyle behaviors, and guideline-related professional knowledge. The intensive training group completed an additional reassessment for professional knowledge immediately after the 1-week intensive training program.

Outcomes

The primary outcome was the HbA_1c control rate (percentage of patients achieving HbA_1c <7.0%) at 6-month follow-up. The secondary outcomes included the following: for patients, (1) HbA_1c control rate at 12-month follow-up; (2) changes in clinical parameters assessed at both 6- and 12-month follow-ups, including glycemic parameters (HbA_1c, fasting and 2-hour postprandial blood glucose), systolic and diastolic blood pressure, anthropometric measurements, lipid profiles, and lifestyle changes; (3) between-group differences in medication prescription choices for glycemic, lipid, and blood pressure control at 6- and 12-month follow-ups; (4) a safety outcome including the adverse events reported by the participating physicians; and (5) for physicians, between-group differences in standard questionnaire results at 12-month follow-up (Supplement 1).

All laboratory test data and questionnaire information about enrolled patients were collected through the well-established data management platform of the MMC as described in the previous study.¹⁰ For participating physicians, identical self-administered questionnaires were completed at baseline and at the 12-month end point through Doctor’s Studio App mobile portal.

Statistical Analysis

Statistical analyses were performed from January 1 to June 1, 2025. For this cluster randomized clinical trial, we aimed to include 200 centers, with 1 physician from each center invited to participate in this study. Based on management experience and unpublished follow-up data from the MMC, approximately 30% of patients with type 2 diabetes who had uncontrolled HbA_1c (≥7.0%) at baseline could achieve glycemic control (HbA_1c <7.0%) after 6 months of management. Assuming HbA_1c control rates of 40% in the intensive training group and 30% in the regular training group, an intraclass correlation coefficient of 0.05, 95% power, and a 2-sided α of .05, 800 patients per group were required. Allowing for approximately 20% attrition, the planned sample size was 2000 patients. No interim analyses or stopping guidelines were planned.

The primary analysis was based on the intention-to-treat principle, including patients with available HbA_1c data at the 6- and 12-month follow-up visits. Sensitivity analyses were performed using last observation carried forward (LOCF), multiple imputation, nonresponder imputation, and an analysis excluding centers with physician changes (eTables 2-4 in Supplement 2). Detailed procedures are specified in the statistical analysis plan (Supplement 1).

Differences in patient outcomes between the 2 groups at 6 and 12 months were estimated using mixed-effects models with physician as a random intercept. Linear mixed-effects models were used for continuous outcomes, and logistic mixed-effects models were used for binary outcomes, with absolute risk differences derived by marginal standardization for binary outcomes. Adjusted models for the primary and key secondary outcomes included patient age, sex, diabetes duration, and the corresponding baseline value; age and diabetes duration were modeled as continuous linear variables in years. Medication use was compared using χ² tests. Within-physician changes in knowledge accuracy and lifestyle behaviors were assessed using paired t tests or McNemar tests, as appropriate; between-group differences in these physician-level changes were estimated using generalized linear models with gaussian identity links for continuous changes and binomial logit links for binary outcomes. Prespecified subgroup analyses assessed heterogeneity across physician education, professional title, diabetes care experience, management quality, and hospital grade. Two-sided P < .05 indicated statistical significance. Statistical analyses were performed using R, version 4.2.3 (R Project for Statistical Computing). Detailed statistical methods are provided in Supplement 1.

Results

Baseline Characteristics of Participants

A total of 205 physicians (mean [SD] age, 36.2 [5.1] years; 50 men [24.4%] and 155 women) and 2017 patients (mean [SD] age, 53.0 [7.0] years; 1304 men [64.7%] and 713 women [35.3%]) were enrolled (Table 1). In phase 1 (February 13 to April 29, 2023), a total of 248 centers from MMC network were initially invited to participate. Of these, 212 centers (1 physician per center) from the MMC network were recruited. Ultimately, 205 physicians (103 randomized to the intensive training group and 102 to the regular training group) completed the full research procedure in the study (Figure 1). In phase 2 (February 28 to September 19, 2023), 3009 patients who first visited the centers were screened consecutively for eligibility, of whom 2017 were enrolled (1009 in the intensive training group and 1008 in the regular training group). Baseline characteristics of physicians and patients were well balanced between the 2 groups (Table 1). Among enrolled patients, the median duration of diabetes was 4.0 years (IQR, 1.3-7.1 years), the mean (SD) BMI was 26.8 (2.4), and the mean (SD) HbA_1c level was 8.7% (0.8%).

Table 1. Baseline Characteristics of Participating Centers, Physicians, and Participants With Type 2 Diabetes^a.

Characteristic	Overall, No. (%)	Intensive training, No. (%)	Regular training, No. (%)
Characteristics of participating centers
No. of centers	205	103	102
Hospital grades of each center
Secondary hospital	51 (24.9)	26 (25.2)	25 (24.5)
Tertiary hospital	154 (75.1)	77 (74.8)	77 (75.5)
Management quality ranking^b
Top 35%	72 (35.1)	36 (35.0)	36 (35.3)
Middle 35%	73 (35.6)	36 (35.0)	37 (36.3)
Bottom 30%	60 (29.3)	31 (30.1)	29 (28.4)
Characteristics of participating physicians
No. of physicians	205	103	102
Age, mean (SD), y	36.2 (5.1)	36.6 (5.0)	35.8 (5.2)
Sex
Female	155 (75.6)	74 (71.8)	81 (79.4)
Male	50 (24.4)	29 (28.2)	21 (20.6)
BMI, mean (SD)	23.0 (3.3)	23.3 (3.1)	22.7 (3.4)
Years of experience in treating patients with diabetes, mean (SD)	8.9 (5.1)	9.2 (5.1)	8.6 (5.2)
Characteristics of participants with type 2 diabetes
No. of participants	2017	1009	1008
Age, mean (SD), y	53.0 (7.0)	53.1 (6.7)	53.0 (6.7)
Sex
Female	713 (35.3)	348 (34.5)	365 (36.2)
Male	1304 (64.7)	661 (65.5)	643 (63.8)
Diabetes duration, median (IQR), y	4.0 (1.3-7.1)	4.1 (1.3-7.1)	3.7 (1.3-7.2)
History of hypertension	771 (39.3)	394 (40.2)	377 (38.4)
History of hyperlipidemia	563 (28.7)	263 (26.8)	300 (30.6)
History of coronary heart disease	90 (4.6)	40 (4.1)	50 (5.1)
Systolic blood pressure, mean (SD), mm Hg	130.1 (15.5)	129.7 (15.4)	130.4 (15.7)
Diastolic blood pressure, mean (SD), mm Hg	80.7 (10.1)	80.5 (9.9)	80.9 (10.4)
Weight, mean (SD), kg	73.6 (10.0)	73.8 (10.1)	73.4 (9.9)
BMI, mean (SD)	26.8 (2.4)	26.8 (2.4)	26.8 (2.4)
Waist circumference, mean (SD), cm	92.8 (8.4)	92.8 (8.2)	92.8 (8.6)
Waist to hip ratio, mean (SD), %	95.4 (5.8)	95.5 (5.7)	95.4 (5.9)
HbA_1c, mean (SD), %	8.7 (0.8)	8.7 (0.8)	8.8 (0.8)
Fasting blood glucose, mean (SD), mg/dL	175.8 (25.6)	175.9 (26.4)	175.6 (24.7)
2-h Postprandial blood glucose, mean (SD), mg/dL	281.2 (75.8)	279.5 (75.8)	283.0 (75.8)
Fasting insulin, median (IQR), μIU/mL	9.3 (5.8-14.1)	9.6 (5.6-14.1)	9.2 (5.9-14.1)
2-h Postprandial insulin, median (IQR), μIU/mL	27.5 (17.2-47.6)	27.6 (17.5-49.7)	27.0 (16.9-45.6)
Triglycerides, median (IQR), mg/dL	178.5 (119.3-276.6)	178.0 (116.5-270.6)	178.9 (121.3-280.8)
Total cholesterol, mean (SD), mg/dL	193.0 (49.5)	193.6 (49.3)	192.3 (49.8)
Low-density lipoprotein cholesterol, mean (SD), mg/dL	114.1 (36.0)	115.2 (36.3)	113.0 (35.6)

Open in a new tab

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); HbA_1c, hemoglobin A_1c.

SI conversion factors: To convert HbA_1c to proportion of total hemoglobin, multiply by 0.01; glucose to millimoles per liter, multiply by 0.0555; insulin to picomoles per liter, multiply by 6.945; triglycerides to millimoles per liter, multiply by 0.0113; and total cholesterol and low-density lipoprotein cholesterol to millimoles per liter, multiply by 0.0259.

^{^a}

Percentages may not total 100% due to rounding.

^{^b}

The ranking of management quality across all metabolic management centers is comprehensively outlined in the study protocol (Supplement 1).

Overall, 820 participants (81.3%) in the intensive training group and 818 participants (81.2%) in the regular training group finished the 6-month follow-up visit, while 824 participants (81.7%) in the intensive training group and 832 participants (82.5%) in the regular training group finished the 12-month follow-up visit and were included in the main analysis (Figure 1). In addition, those who had at least 1 HbA_1c follow-up measurement (937 participants [92.9%] in the intensive training group and 933 participants [92.6%] in the regular training group) were included in the LOCF analysis (eTable 2 in Supplement 2).

Primary Outcome and HbA_1c Control

At 6 months, 58.0% of participants (476 of 820) in the intensive training group and 42.9% of participants (351 of 818) in the regular training group achieved an HbA_1c level of less than 7.0%, with an adjusted between-group difference of 16.6% (95% CI, 7.2%-25.7%; P < .001) (relative increase by 35.2%) (Table 2 and Figure 2). This difference persisted at 12 months, with 60.9% of participants (502 of 824) in the intensive training group achieving an HbA_1c level of less than 7.0%, compared with 44.6% of participants (371 of 832) in the regular training group, corresponding to an adjusted between-group difference of 17.0% (95% CI, 7.0%-26.8%; P < .001) (relative increase by 36.5%).

Table 2. Primary and Secondary Outcomes Among Participants With Type 2 Diabetes^a.

Outcome	Intensive training		Regular training		Mean difference (95% CI)	P value^b	Adjusted mean difference (95% CI)	Adjusted P value^c
Outcome	Value	Change from baseline, mean (95% CI)	Value	Change from baseline, mean (95% CI)	Mean difference (95% CI)	P value^b	Adjusted mean difference (95% CI)	Adjusted P value^c
Primary outcome
HbA_1c <7.0% at 6 mo, No./total No. (%)	476/820 (58.0)	58.0 (54.7 to 61.4)	351/818 (42.9)	42.9 (39.5 to 46.3)	15.7 (6.7 to 24.4)	<.001	16.6 (7.2 to 25.7)	<.001
Secondary outcomes
HbA_1c <7.0% at 12 mo, No./total No. (%)	502/824 (60.9)	60.9 (57.6 to 64.3)	371/832 (44.6)	44.6 (41.2 to 48.0)	16.3 (6.7 to 25.6)	<.001	17.0 (7.0 to 26.8)	<.001
HbA_1c, mean (SD), %
At 6 mo	7.0 (1.0)	−1.7 (−1.8 to −1.7)	7.3 (1.1)	−1.5 (−1.6 to −1.4)	−0.3 (−0.4 to −0.1)	.002	−0.3 (−0.4 to −0.2)	<.001
At 12 mo	6.9 (1.0)	−1.8 (−1.9 to −1.7)	7.3 (1.3)	−1.4 (−1.5 to −1.3)	−0.3 (−0.5 to −0.2)	<.001	−0.4 (−0.5 to −0.2)	<.001
Weight, mean (SD), kg
At 6 mo	71.9 (10.4)	−2.0 (−2.2 to −1.7)	71.7 (10.5)	−1.5 (−1.8 to −1.2)	−0.4 (−1.0 to 0.1)	.08	−0.4 (−0.9 to 0.1)	.09
At 12 mo	71.1 (10.3)	−2.9 (−3.2 to −2.6)	71.3 (10.3)	−2.0 (−2.3 to −1.7)	−0.9 (−1.5 to −0.3)	.003	−0.8 (−1.4 to −0.3)	<.001
BMI, mean (SD)
6 mo	26.1 (2.7)	−0.8 (−0.9 to −0.7)	26.2 (2.7)	−0.6 (−0.7 to −0.5)	−0.1 (−0.3 to 0.0)	.14	−0.1 (−0.3 to 0.0)	.14
12 mo	25.8 (2.7)	−1.1 (−1.2 to −0.9)	26.1 (2.7)	−0.8 (−0.9 to −0.7)	−0.3 (−0.5 to −0.1)	.005	−0.3 (−0.5 to −0.1)	<.001
Waist circumference, mean (SD), cm
At 6 mo	91.4 (8.3)	−1.8 (−2.2 to −1.5)	91.6 (8.3)	−1.0 (−1.5 to −0.6)	−0.7 (−1.5 to 0.1)	.08	−0.7 (−1.5 to 0.1)	.09
At 12 mo	90.0 (7.7)	−2.8 (−3.2 to −2.5)	91.7 (8.3)	−1.4 (−1.7 to −1.0)	−1.4 (−2.1 to −0.7)	<.001	−1.4 (−2.1 to −0.8)	<.001
Waist to hip ratio, mean (SD), %
At 6 mo	94.1 (5.7)	−1.5 (−1.9 to −1.1)	94.4 (5.9)	−0.8 (−1.2 to −0.4)	−0.6 (−1.2 to 0.1)	.11	−0.5 (−1.1 to 0.2)	.15
At 12 mo	93.8 (5.6)	−1.6 (−2.0 to −1.3)	95.0 (6.5)	−0.6 (−0.9 to −0.2)	−1.0 (−1.6 to −0.4)	.001	−1.0 (−1.6 to −0.5)	<.001
Systolic blood pressure, mean (SD), mm Hg
At 6 mo	128.3 (12.0)	−1.3 (−2.4 to −0.2)	129.5 (14.1)	−1.3 (−2.4 to −0.3)	0.3 (−1.5 to 2.1)	.78	−0.5 (−2.2 to 1.1)	.54
At 12 mo	125.1 (12.6)	−4.2 (−5.3 to −3.1)	126.9 (12.1)	−3.2 (−4.2 to −2.2)	−1.0 (−2.7 to 0.6)	.22	−1.5 (−2.9 to −0.1)	.04
Diastolic blood pressure, mean (SD), mm Hg
At 6 mo	79.1 (8.2)	−1.4 (−2.1 to −0.7)	78.3 (8.8)	−2.4 (−3.1 to −1.7)	1.0 (−0.3 to 2.2)	.13	0.9 (−0.3 to 2.0)	.15
At 12 mo	77.2 (8.6)	−3.1 (−3.9 to −2.4)	77.4 (8.4)	−3.0 (−3.7 to −2.3)	−0.2 (−1.5 to 1.1)	.75	−0.3 (−1.4 to 0.9)	.66
Fasting blood glucose, mean (SD), mg/dL
At 6 mo	133.7 (30.8)	–42.0 (–44.7 to –39.3)	138.4 (34.7)	−36.8 (−39.6 to −33.9)	−4.9 (−9.5 to –0.4)	.04	−4.3 (−8.4 to –0.2)	.04
At 12 mo	129.7 (31.2)	−46.1 (−48.7 to −43.5)	137.1 (37.8)	−38.7 (−41.7 to −35.7)	−6.3 (−11.3 to −1.4)	.01	−6.3 (−10.7 to −1.9)	.006
2-h Postprandial blood glucose, mean (SD), mg/dL
At 6 mo	206.9 (68.1)	−70.8 (−78.7 to −63.0)	205.4 (63.8)	−78.1 (−87.5 to −68.6)	6.1 (−12.3 to 24.5)	.51	3.7 (−8.4 to 15.8)	.55
At 12 mo	203.5 (67.4)	−74.1 (−81.5 to −66.6)	211.4 (67.8)	−67.6 (−75.7 to −59.4)	−5.7 (−23.4 to 11.9)	.52	−8.8 (−21.2 to 3.5)	.16
Triglycerides, median (IQR), mg/dL
At 6 mo	141.7 (98.3-206.1)	−55.9 (−74.5 to −37.3)	143.5 (100.1-202.8)	−66.1 (−82.9 to −49.4)	10.3 (−14.9 to 35.4)	.42	6.7 (−10.7 to 24.2)	.45
At 12 mo	141.7 (96.5-203.7)	−62.0 (−81.3 to −42.6)	154.1 (104.5-219.7)	−49.6 (−67.1 to −32.1)	−12.4 (−38.7 to 14.0)	.36	−18.9 (−40.0 to 2.1)	.08
Total cholesterol, mean (SD), mg/dL
At 6 mo	179.0 (46.2)	−14.6 (−18.5 to −10.8)	181.5 (44.9)	−11.9 (−16.0 to −7.8)	−2.2 (−8.7 to 4.4)	.52	–1.9 (−6.7 to 3.0)	.46
At 12 mo	178.4 (44.7)	−15.8 (−19.7 to −12.0)	181.3 (48.3)	−11.3 (−15.2 to −7.3)	−4.3 (−10.6 to 1.9)	.18	−3.7 (−9.0 to 1.5)	.17
Low-density lipoprotein cholesterol, mean (SD), mg/dL
At 6 mo	104.9 (35.4)	−10.2 (−13.1 to −7.3)	106.8 (34.9)	−6.9 (−9.9 to −3.9)	−2.6 (−7.5 to 2.2)	.28	–1.8 (−5.7 to 2.1)	.37
At 12 mo	104.3 (37.6)	−11.6 (−14.7 to −8.5)	104.9 (34.3)	−8.7 (−11.6 to −5.7)	–1.9 (−7.4 to 3.6)	.50	–1.0 (−5.6 to 3.5)	.65

Open in a new tab

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); HbA_1c, hemoglobin A_1c.

SI conversion factors: To convert HbA_1c to proportion of total hemoglobin, multiply by 0.01; glucose to millimoles per liter, multiply by 0.0555; triglycerides to millimoles per liter, multiply by 0.0113; and total cholesterol and low-density lipoprotein cholesterol to millimoles per liter, multiply by 0.0259.

^{^a}

This set includes participants with HbA_1c assessment at the specified time points (6 and 12 mo, respectively). For continuous outcomes, the values shown under “change from baseline” were calculated as the follow-up value minus the baseline value at the corresponding time point. The mean difference represents the unadjusted between-group difference in these changes, estimated using linear mixed-effects models with physician included as a random intercept. Adjusted mean differences were estimated using linear mixed-effects models further adjusted for the corresponding baseline value, age, sex, and diabetes duration, with physician included as a random intercept. For binary outcomes, the unadjusted and adjusted differences represent absolute risk differences between the 2 groups. These risk differences were derived from logistic mixed-effects models as differences in estimated marginal probabilities between the intensive training group and the regular training group. The adjusted logistic mixed-effects models were further adjusted for the corresponding baseline HbA_1c, age, sex, and diabetes duration, with physician included as a random intercept. Positive values indicate higher values, greater increases, or higher probabilities in the intensive training group compared with the regular training group, as applicable. P < .05 was considered statistically significant.

^{^b}

P values for unadjusted between-group differences.

^{^c}

P values for adjusted between-group differences.

Two-panel line graphs of H b A 1 c outcomes over 12 months in two training groups. Two side-by-side panels labeled A at upper left and B at upper left. Panel A title: Participants with H b A 1 c less than 7.0 percent. Vertical axis labeled Participants, percent, ranging from 0 to 100 with gridlines; horizontal axis labeled Month with tick marks at 0, 3, 6, 9, and 12. A legend near the upper left contains two entries: Intensive training group marked by a dark teal filled circle, and Regular training group marked by an orange filled circle. Both series use connected lines with circular markers and vertical error bars at months 3, 6, 9, and 12. At month 0, both groups start at approximately 0 percent. At month 3, intensive approximately 55 percent and regular approximately 46 percent. At month 6, intensive approximately 58 percent and regular approximately 42 percent. At month 9, intensive approximately 62 percent and regular approximately 50 percent. At month 12, intensive approximately 60 percent and regular approximately 44 percent. Error bars are short relative to the axis, extending a few percentage points above and below each marker. Panel B title: H b A 1 c levels among all participants. Vertical axis labeled H b A 1 c, percent, ranging from 6 to 9 with tick marks at 6, 7, 8, and 9; horizontal axis labeled Month with tick marks at 0, 3, 6, 9, and 12. Two lines match the legend colors from panel A, with circular markers and vertical error bars at months 3, 6, 9, and 12. At month 0, intensive approximately 8 point 7 percent and regular approximately 8 point 8 percent. At month 3, intensive approximately 7 point 0 percent and regular approximately 7 point 1 percent. At month 6, intensive approximately 7 point 0 percent and regular approximately 7 point 3 percent. At month 9, intensive approximately 6 point 9 percent and regular approximately 7 point 1 percent. At month 12, intensive approximately 6 point 9 percent and regular approximately 7 point 3 percent. Error bars span roughly one to two tenths of a percent around each mean. — A, Proportion of participants achieving HbA_1c levels less than 7.0%. B, Mean HbA_1c levels. Error bars indicate 95% CIs. To convert HbA_1c to proportion of total hemoglobin, multiply by 0.01.

Differences in the proportion of participants achieving an HbA_1c level of less than 7.0% at 6 months between the intensive and regular training groups were consistent across subgroups defined by highest academic degree, professional title, years of experience in treating patients with diabetes, center management quality ranking, and hospital grade. The intervention effect appeared greater among patients managed by physicians with a bachelor’s degree, junior or intermediate professional titles, or those from secondary hospitals (Figure 3).

Forest plot of subgroup differences in HbA 1c under 7 point 0 percent at 6 months. Left panel formatted as a table with four columns. Column one header: Characteristic of physician or center. Column two header: Difference in proportion of H b A 1 c less than 7 point 0 percent at 6 mo, No. slash total No. percent, with subcolumn header Intensive training. Column three header: Regular training. Column four header: Group difference, 95 percent C I, percent. Row Overall: Intensive training 476 slash 820, 58 point 0; Regular training 351 slash 818, 42 point 9; Group difference 15 point 1, 10 point 4 to 19 point 9. Section Highest academic degree with two rows: Bachelor’s 217 slash 358, 60 point 6; 134 slash 342, 39 point 2; 21 point 4, 14 point 2 to 28 point 7. Master’s or Doctor 259 slash 462, 56 point 1; 217 slash 476, 45 point 6; 10 point 5, 4 point 1 to 16 point 8. Section Professional title: Junior or intermediate 256 slash 438, 58 point 4; 197 slash 491, 40 point 1; 18 point 3, 12 point 0 to 24 point 7. Senior 220 slash 382, 57 point 6; 154 slash 327, 47 point 1; 10 point 5, 3 point 2 to 17 point 8. Section Years of experience in treating patients with diabetes: less than 9 y 176 slash 319, 55 point 2; 168 slash 406, 41 point 4; 13 point 8, 6 point 5 to 21 point 1. Greater than or equal to 9 y 300 slash 501, 59 point 9; 183 slash 412, 44 point 4; 15 point 5, 9 point 0 to 21 point 9. Section Management quality ranking: Top 35 percent 236 slash 389, 60 point 7; 145 slash 340, 42 point 6; 18 point 0, 10 point 9 to 25 point 2. Middle 35 percent 143 slash 245, 58 point 4; 130 slash 284, 45 point 8; 12 point 6, 4 point 1 to 21 point 1. Bottom 30 percent 97 slash 186, 52 point 2; 76 slash 194, 39 point 2; 13 point 0, 3 point 0 to 22 point 9. Section Hospital grade: Secondary 157 slash 255, 61 point 6; 99 slash 237, 41 point 8; 19 point 8, 11 point 1 to 28 point 5. Tertiary 319 slash 565, 56 point 5; 252 slash 581, 43 point 4; 13 point 1, 7 point 3 to 18 point 8. Right panel forest plot aligned to the same rows, with dark teal square markers and horizontal lines for 95 percent confidence intervals. A vertical dotted reference line near 15 percent. Top labels above the plot: Favors regular training on the left and Favors intensive training on the right. Bottom horizontal axis label: Difference in proportion of H b A 1 c less than 7 point 0 percent, with tick marks at 0, 10, 20, 30, and 40. — Group difference refers to the difference in the proportion of participants with type 2 diabetes achieving an HbA_1c level less than 7.0% at 6 months, with positive values indicating greater achievement rate in the intensive training group. Horizontal lines represent 95% CIs.

For HbA_1c levels, participants in the intensive training group showed greater reductions than those in the regular training group at both 6 months (−1.7% [95% CI, −1.8% to −1.7%] vs −1.5% [95% CI, −1.6% to −1.4%]; adjusted difference, −0.3% [95% CI, −0.4% to −0.2%]; P < .001) and 12 months (−1.8% [95% CI, −1.9% to −1.7%] vs −1.4% [95% CI, −1.5% to −1.3%]; adjusted difference, −0.4% [95% CI, −0.5% to −0.2%]; P < .001) (Table 2). Sensitivity analyses using LOCF, multiple imputation, nonresponder imputation, and exclusion of centers with physician changes yielded consistent results (eTables 2-4 in Supplement 2).

Other Clinical Outcomes

Participants in the intensive training group achieved better control of fasting blood glucose, BMI, waist circumference, and systolic blood pressure compared with those in the regular training group. These results were consistent at both 6- and 12-month follow-up assessments (Table 2; eTables 2-4 in Supplement 2). At 12 months, participants in the intensive training group exhibited significantly greater reductions in BMI (−1.1 [95% CI, −1.2 to −0.9] vs −0.8 [95% CI, −0.9 to −0.7]; adjusted between-group mean difference, −0.3 [95% CI, −0.5 to −0.1]; P < .001), waist circumference (−2.8 cm [95% CI, −3.2 to −2.5 cm] vs −1.4 cm [95% CI, −1.7 to −1.0 cm]; adjusted difference, −1.4 cm [95% CI, −2.1 to −0.8 cm]; P < .001), and waist to hip ratio (−1.6% [95% CI, −2.0% to −1.3%] vs −0.6% [95% CI, −0.9% to −0.2%]; adjusted difference, −1.0% [95% CI, −1.6% to −0.5%]; P < .001) compared with those in the regular training group (Table 2). Similarly, the difference at 12 months in fasting blood glucose reduction between the intensive training group and the regular training group (−46.1 mg/dL [95% CI, −48.7 to −43.5 mg/dL] vs −38.7 mg/dL [95% CI, −41.7 to −35.7 mg/dL]; adjusted mean difference, −6.3 mg/dL [95% CI, −10.7 to −1.9 mg/dL]; P = .006) and systolic blood pressure reduction (−4.2 mm Hg [95% CI, −5.3 to −3.1 mm Hg] vs −3.2 mm Hg [95% CI, −4.2 to −2.2 mm Hg]; adjusted difference, −1.5 mm Hg [95% CI, −2.9 to −0.1 mm Hg]; P = .04) were significant. Both groups showed reductions in 2-hour postprandial blood glucose, diastolic blood pressure, and lipid levels, without significant between-group differences.

Changes in Participants’ Lifestyle Behaviors

At 12 months, participants in the intensive training group showed healthier lifestyle behaviors than those in the regular training group (eg, fewer currently drinking, 4.7% [35 of 745] vs 7.2% [54 of 750]; more with vegetable and fruit consumption ≥500 g/d: 70.4% [536 of 761] vs 58.0% [431 of 743]) (eTable 5 in Supplement 2). Moreover, the proportion of participants with ideal nighttime sleep duration (7 to <9 hours) was increased by 5.1% (95% CI, 0.0%-9.9%; P = .04) in the intensive training group, while no change was found in the regular training group. Total serious adverse events were comparable between the 2 groups (intensive group, 2.6% [26 of 1009] vs regular group, 2.4% [24 of 1008]) (eTable 6 in Supplement 2).

Changes in Physicians’ Knowledge and Lifestyle Behaviors

Physicians in the intensive training group showed greater improvements in diabetes-related knowledge than those in the regular training group. Total questionnaire accuracy increased by 7.9% (95% CI, 5.6%-10.2%; P < .001), with improvements in clinical target management and lifestyle intervention knowledge (eTable 7 in Supplement 2).

Physicians in the intensive training group also had a greater reduction in BMI than those in the regular training group (−0.5 [95% CI, −0.7 to −0.2] vs 0.2 [95% CI, −0.1 to 0.5]; P = .01) and showed improved dietary behaviors (eg, increase in vegetable and fruit consumption by 225 g/week [95% CI, 50-400 g/week]; P = .03), whereas no corresponding improvements were observed in the regular training group (eTable 8 in Supplement 2). These findings paralleled the changes found in their patients at the 12-month follow-up (eTable 5 in Supplement 2).

Discussion

This multicenter, cluster randomized clinical trial demonstrated that implementing a role-playing–based intensive training for physicians significantly improved management of type 2 diabetes. Compared with regular training, the intensive training group achieved a greater proportion of HbA_1c control (<7.0%) with absolute increase by 16.6% (relative increase by 35.2%) at 6 months and 17.0% (relative increase by 36.5%) at 12 months, accompanied by significantly greater improvement in multiple metabolic parameters, including BMI, waist circumference, fasting blood glucose, and systolic blood pressure. Moreover, both patients’ lifestyle modifications and physicians’ guideline-related knowledge and personal behaviors improved significantly in the intensive training group. These findings suggest that the role-playing–based on-site experiential training effectively bridges the guideline-implementation gap in diabetes care and reaffirms the importance of human-centered offline training in digital health care systems for medical education.

This trial extends previous research on physician education and experiential learning in several ways. Physician education is an important strategy for narrowing the gap between evidence-based guidelines and routine clinical practice, particularly in chronic disease care, in which long-term self-management, treatment adherence, and patient-physician communication are central to clinical outcomes.^9,10,11 Because changes in one physician’s practice may influence the care of many patients, physician-focused training may have broad downstream effects on patient care. Prior medical education studies have suggested that learner-centered and practice-oriented approaches, including role-play, feedback, standardized patient encounters, and simulation-based communication training, can improve physicians’ communication behaviors, empathy, and training-related outcomes.^22,23,24,25 Our study extends this literature by applying a physician-as-patient experiential model to diabetes management and by linking physician training with objective patient metabolic outcomes.

Unlike procedural simulation studies that focus mainly on technical skills or acute clinical scenarios,^14,26,27,28 this intervention exposed physicians to selected aspects of patients’ daily experiences with diabetes and the practical and psychosocial challenges of long-term care. In addition, whereas previous diabetes-related role-playing or peer-support interventions primarily targeted patients’ self-management or trained patients as peer supporters,^3,29 our physician-centered intervention combined clinicians’ medical knowledge with experiential understanding of patients’ perspectives. Moreover, whereas many prior training programs relied on physician-reported, questionnaire-based, or short-term educational outcomes,^{12,13,22,23,24,25} this trial prospectively evaluated patient-level metabolic and behavioral outcomes, with each physician managing a prespecified number of patients with comparable baseline characteristics. This design strengthens the clinical relevance of the intervention and supports the credibility of between-group comparisons.

Both groups achieved statistically significant improvements in HbA_1c level, BMI, and blood pressure from baseline, with nuanced between-group differences. The approximately 1% within-group HbA_1c reduction was clinically meaningful, helping patients achieve an HbA_1c level of less than 7.0% and reducing diabetes-related complication risks.³⁰

Strengths and Limitations

The overall design of this trial was characterized by several unique strengths. First, this study addresses low guideline adherence by developing an immersive, role-playing training program aligned with current diabetes management guidelines.^7,8,31 Physicians’ answers to questionnaires revealed superior knowledge in the intensive training group after training and at 12 months, alongside greater BMI reduction and healthier dietary habits. This finding indicated that the intensive training induced meaningful behavioral changes among physicians, consistent with patient improvements. Second, with prevalent online education in continuing medical education, our study highlighted the role of offline training. Despite ample online resources, supplementary offline training yielded better patient outcomes, suggesting that digital learning alone cannot fully cultivate complex clinical skills. In-person interaction fosters emotional empathy and reflective practice unachievable via virtual modes. Third, the study used the nationwide MMC platform with a unified workflow, electronic database, and follow-up infrastructure.^10,18 It ensured reliable implementation and scalable, replicable chronic disease management patterns. With accumulated practical experience, the intervention can be broadly rolled out nationwide via more than 2000 MMCs and further adapted globally,¹⁰ with training tailored to local medical contexts.

This study also has several limitations. First, each physician enrolled approximately 10 patients, which limited the ability to assess physician-level variation in intervention effects, although the overall sample size provided adequate statistical power. Second, the intervention did not significantly improve patients’ exercise adherence, highlighting the difficulty in modifying patient exercise habits. Third, because patients were enrolled after center randomization, potential selection bias was considered. This risk was mitigated by consecutive patient screening and enrollment, together with assessments of sample representativeness. Future multicenter trials may further reduce this risk through independent oversight of patient enrollment. Fourth, the open-label design may have introduced performance bias and Hawthorne effects, potentially influencing the observed intervention effects.

Conclusions

In this cluster randomized clinical trial of a role-playing–based intensive training program for physicians managing patients with type 2 diabetes, the intervention significantly enhanced physicians’ clinical practice and sustainably improved glycemic control and other metabolic outcomes among their patients. This approach may help narrow the guideline-implementation gap in diabetes care and represents a potentially scalable strategy for improving chronic metabolic disease management.

Supplement 1.

Trial Protocol and Statistical Analysis Plan

jamanetwopen-e2627376-s001.pdf^{(3.2MB, pdf)}

Supplement 2.

eTable 1. Contents of 1-Week Role-Playing Based Intensive Training Program for Physicians

eTable 2. Primary and Secondary Outcomes in Participants With Type 2 Diabetes Mellitus in LOCF Set

eTable 3. Primary and Secondary Outcomes in Participants With Type 2 Diabetes Mellitus (Excluding Centers With Physician Changed)

eTable 4. Primary and Secondary Outcomes in Multiple Imputation and Non-Responder Imputation Analyses

eTable 5. Lifestyle Changes of Participants With Type 2 Diabetes Mellitus

eTable 6. Serious Adverse Events and Hypoglycemia Events During 12-Month Study Period

eTable 7. Accuracy of Diabetes Knowledge Tests of Physicians

eTable 8. Body Weight Management and Lifestyle Changes of Physicians

eTable 9. Reasons for Patients Excluded in the Study

eFigure. Study Design

jamanetwopen-e2627376-s002.pdf^{(512.3KB, pdf)}

Supplement 3.

Nonauthor Collaborators. Chinese Endocrinologists Health Education Study (CREATION) group

jamanetwopen-e2627376-s003.pdf^{(541.8KB, pdf)}

Supplement 4.

Data Sharing Statement

jamanetwopen-e2627376-s004.pdf^{(16.5KB, pdf)}

References

1.GBD 2021 Diabetes Collaborators . Global, regional, and national burden of diabetes from 1990 to 2021, with projections of prevalence to 2050: a systematic analysis for the Global Burden of Disease Study 2021. Lancet. 2023;402(10397):203-234. doi: 10.1016/S0140-6736(23)01301-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Ahmad E, Lim S, Lamptey R, Webb DR, Davies MJ. Type 2 diabetes. Lancet. 2022;400(10365):1803-1820. doi: 10.1016/S0140-6736(22)01655-5 [DOI] [PubMed] [Google Scholar]
3.Chan JCN, Sui Y, Oldenburg B, et al. ; JADE and PEARL Project Team . Effects of telephone-based peer support in patients with type 2 diabetes mellitus receiving integrated care: a randomized clinical trial. JAMA Intern Med. 2014;174(6):972-981. doi: 10.1001/jamainternmed.2014.655 [DOI] [PubMed] [Google Scholar]
4.Xu Y, Lu J, Li M, et al. Diabetes in China part 1: epidemiology and risk factors. Lancet Public Health. 2024;9(12):e1089-e1097. doi: 10.1016/S2468-2667(24)00250-0 [DOI] [PubMed] [Google Scholar]
5.Wang L, Peng W, Zhao Z, et al. Prevalence and treatment of diabetes in China, 2013-2018. JAMA. 2021;326(24):2498-2506. doi: 10.1001/jama.2021.22208 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.NCD Risk Factor Collaboration (NCD-RisC) . Worldwide trends in diabetes prevalence and treatment from 1990 to 2022: a pooled analysis of 1108 population-representative studies with 141 million participants. Lancet. 2024;404(10467):2077-2093. doi: 10.1016/S0140-6736(24)02317-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Flood D, Seiglie JA, Dunn M, et al. The state of diabetes treatment coverage in 55 low-income and middle-income countries: a cross-sectional study of nationally representative, individual-level data in 680 102 adults. Lancet Healthy Longev. 2021;2(6):e340-e351. doi: 10.1016/S2666-7568(21)00089-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Guariguata L, Sobers N. Rising diabetes, lagging treatment, and the need for better systems. Lancet. 2024;404(10467):2026-2028. doi: 10.1016/S0140-6736(24)02422-X [DOI] [PubMed] [Google Scholar]
9.Chan JCN, Lim LL, Wareham NJ, et al. The Lancet Commission on diabetes: using data to transform diabetes care and patient lives. Lancet. 2021;396(10267):2019-2082. doi: 10.1016/S0140-6736(20)32374-6 [DOI] [PubMed] [Google Scholar]
10.Ning G. Medical education in diabetes management on the new horizon: insights from metabolic management center. J Diabetes. 2025;17(3):e70075. doi: 10.1111/1753-0407.70075 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Siminerio LM, Albright A, Fradkin J, et al. The National Diabetes Education Program at 20 years: lessons learned and plans for the future. Diabetes Care. 2018;41(2):209-218. doi: 10.2337/dc17-0976 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Barteit S, Guzek D, Jahn A, Bärnighausen T, Jorge MM, Neuhann F. Evaluation of e-learning for medical education in low- and middle-income countries: a systematic review. Comput Educ. 2020;145:103726. doi: 10.1016/j.compedu.2019.103726 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Cheng C, Papadakos J, Umakanthan B, et al. On the advantages and disadvantages of virtual continuing medical education: a scoping review. Can Med Educ J. 2023;14(3):41-74. doi: 10.36834/cmej.75681 [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Cook DA, Hatala R, Brydges R, et al. Technology-enhanced simulation for health professions education: a systematic review and meta-analysis. JAMA. 2011;306(9):978-988. doi: 10.1001/jama.2011.1234 [DOI] [PubMed] [Google Scholar]
15.Jiao Z, Yang Y, Zhang S, Xu J. A comparative study about attitudes towards the efficiency, effectiveness, and atmosphere of offline and online learning among medical students. Ann Transl Med. 2022;10(23):1270. doi: 10.21037/atm-22-5112 [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Yardley S, Teunissen PW, Dornan T. Experiential learning: transforming theory into practice. Med Teach. 2012;34(2):161-164. doi: 10.3109/0142159X.2012.643264 [DOI] [PubMed] [Google Scholar]
17.Yardley S, Teunissen PW, Dornan T. Experiential learning: AMEE Guide No. 63. Med Teach. 2012;34(2):e102-e115. doi: 10.3109/0142159X.2012.650741 [DOI] [PubMed] [Google Scholar]
18.Zhang Y, Wang W, Ning G. Metabolic Management Center: an innovation project for the management of metabolic diseases and complications in China. J Diabetes. 2019;11(1):11-13. doi: 10.1111/1753-0407.12847 [DOI] [PubMed] [Google Scholar]
19.World Medical Association . World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA. 2013;310(20):2191-2194. doi: 10.1001/jama.2013.281053 [DOI] [PubMed] [Google Scholar]
20.Wang W, Wang G, Wang Y, et al. National standardized guidelines for the construction and management of Metabolic Disease Management Centers. Chin J Endocrinol Metab. 2019;35(11):907-926. doi: 10.3760/cma.j.issn.1000-6699.2019.11.001 [DOI] [Google Scholar]
21.Chinese Diabetes Society . Guideline for the prevention and treatment of type 2 diabetes mellitus in China (2020 edition). Chin J Diabetes Mellitus. 2021;13(4):315-409. doi: 10.3760/cma.j.cn115791-20210221-00095 [DOI] [Google Scholar]
22.Pilnick A, Trusson D, Beeke S, O’Brien R, Goldberg S, Harwood RH. Using conversation analysis to inform role play and simulated interaction in communications skills training for healthcare professionals: identifying avenues for further development through a scoping review. BMC Med Educ. 2018;18(1):267. doi: 10.1186/s12909-018-1381-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Gilligan C, Powell M, Lynagh MC, et al. Interventions for improving medical students’ interpersonal communication in medical consultations. Cochrane Database Syst Rev. 2021;2(2):CD012418. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Berkhof M, van Rijssen HJ, Schellart AJM, Anema JR, van der Beek AJ. Effective training strategies for teaching communication skills to physicians: an overview of systematic reviews. Patient Educ Couns. 2011;84(2):152-162. doi: 10.1016/j.pec.2010.06.010 [DOI] [PubMed] [Google Scholar]
25.Lane C, Rollnick S. The use of simulated patients and role-play in communication skills training: a review of the literature to August 2005. Patient Educ Couns. 2007;67(1-2):13-20. doi: 10.1016/j.pec.2007.02.011 [DOI] [PubMed] [Google Scholar]
26.Gordon M, Hill E, Stojan JN, Daniel M. Educational interventions to improve handover in health care: an updated systematic review. Acad Med. 2018;93(8):1234-1244. doi: 10.1097/ACM.0000000000002236 [DOI] [PMC free article] [PubMed] [Google Scholar]
27.El Khamali R, Mouaci A, Valera S, et al. ; SISTRESSREA Study Group . Effects of a multimodal program including simulation on job strain among nurses working in intensive care units: a randomized clinical trial. JAMA. 2018;320(19):1988-1997. doi: 10.1001/jama.2018.14284 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Starmer AJ, Spector ND, Srivastava R, et al. ; I-PASS Study Group . Changes in medical errors after implementation of a handoff program. N Engl J Med. 2014;371(19):1803-1812. doi: 10.1056/NEJMsa1405556 [DOI] [PubMed] [Google Scholar]
29.Kaplan RM, Chadwick MW, Schimmel LE. Social learning intervention to promote metabolic control in type I diabetes mellitus: pilot experiment results. Diabetes Care. 1985;8(2):152-155. doi: 10.2337/diacare.8.2.152 [DOI] [PubMed] [Google Scholar]
30.UK Prospective Diabetes Study (UKPDS) Group . Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). Lancet. 1998;352(9131):837-853. doi: 10.1016/S0140-6736(98)07019-6 [DOI] [PubMed] [Google Scholar]
31.American Diabetes Association Professional Practice Committee . 6. Glycemic goals and hypoglycemia: standards of care in diabetes—2024. Diabetes Care. 2024;47(suppl 1):S111-S125. doi: 10.2337/dc24-S006 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials