All-encompassing Program Led by Diabetes Specialist Team

Aim: To examine the effects of 1-year team-based multi-component intervention care (MIC) program in obese T2D with poor glycemic control

Hypothesis: The change from a traditional clinic consultation to an individualized, multicomponent care program will improve cardiometabolic parameters in suboptimally treated obese type 2 diabetic participants.

Study design: This is a pilot translational study aiming to compare the effects of a multi-component care program with personalized drug regimen augmented by behavioural therapy with psychological support and peer influence, led by a diabetes specialist team on metabolic and psychobehavioural health in obese type 2 diabetic participants versus usual care (UC).

Settings: The participants would be identified from the Prince of Wales Hospital (PWH). The recruitment period would be 2 years. Since 1995, the PWH runs a comprehensive complication screening program in the ambulatory Diabetes Centre. On a weekly basis, 50-100 patients undergo assessments for risk factors and complications by nurses and healthcare assistants using standard protocols. On average, 8-12 patients attend each session. Since 2007, these data are entered into a web-based Joint Asia Diabetes Evaluation (JADE) Program which incorporates a validated risk engine to generate an integrated report with risk categories, care protocols, trends of risk factor control and decision support to empower both doctors and patients to make informed decisions. After the assessments, During the first two months of intervention, each group would have three 3.5-hour face-to-face workshops with the nurses (week 0, 4 and 8). This would be reduced to 2-hour per workshop 4-monthly thereafter in the following ten months (week 24, 40, 56) i.e. a total of six sessions per group in 1-year intervention. At each workshop, the nurses would use role plays, quizzes, reflective stories and on-site fitness classes to instill these key principles. Each workshop would use similar strategies including goal-setting, health behaviors tracking and peer sharing on personal progress. During this 1-year MIC program, 4 out of 6 workshops would be conducted when participants would be waiting to see endocrinologists (at week 24, 40 and 56). The consultations would be around 15-minute for each participant per visit. In the first two months of intervention, each MIC group also would receive three 20-minute telephone reminders from HCA at 4-weekly intervals (week 0, 4, 8). The call frequency would be reduced to 4-monthly intervals in the following ten months (week 24, 40, 56) i.e. a total of six calls per group. Group members would also be acted as peer supporters for one another in groups of 3 to 4 patients with at least one telephone reminder between workshops. The calls would be guided by structured log sheets focusing on adherence to lifestyle modification, medications and action plan, as well as to identify patients who would require ad-hoc support from the nurses.

The MIC group would return to UC after exit from the 1-year intervention. During the ensuing two years, the MIC group would receive mailing of three simplified JADE follow-up reports at month 18, month 24 and month 30 where the HCA entered BP, HbA1c, body weight and LDL-cholesterol retrieved from the clinic notes or laboratory measurements to provide feedback to the patients. They would be invited to an annual 4-hour group gathering where doctors and nurses would invite patients who achieve and/or maintain risk factor control to share their success stories and provide motivational support to the group.

The UC group would have traditional clinic-based consultation at 3 to 4-monthly intervals in either general medical or diabetes clinic. They would have consultation by different doctors who might or might not refer them to dietitians or nurses for education. There would be no workshops or peer supporters in the UC group.

For this EMERALD project, eligible participants will be randomized to either the MIC or UC group, which will be done independently by a statistician. The group allocation of each participant will be assigned by having a staff independent of the study. Computer generated random codes will be prepared by the statistician prior to participant recruitment and the codes will be stored in a computer. Irrespective of the assignment group, all participants will receive a 2-hour session on how to interpret the JADE report and their risk profiles and will be reinforced on the importance of achieving targets and optimizing self-care.

Statistical Analysis: Descriptive data would be presented as mean ± standard deviation (SD), median (interquartile range) or number (percentage), as appropriate. Triglyceride and urinary albumin-to-creatinine ratio (ACR) would be logarithmically-transformed due to skewed distribution. We would compare baseline characteristics between two groups using either independent t or Wilcoxon rank-sum test for continuous variables, and either χ2 or Fisher's exact test for categorical variables. For continuous and binary outcomes obtained over time, we would perform generalized estimating equations (GEE) analysis in all patients who would have baseline and at least one follow-up measurement. Using UC as referent, we would introduce a group × time (baseline, year 1 and year 3) interaction term in the GEE models to examine the between-group differences from baseline to year 1 and year 3, adjusting for sex and time-varying covariates namely age, disease duration, BMI, use of insulin, oral glucose-, BP- and lipid-lowering agents. When either body weight or waist circumference would be the dependent variable, BMI would be excluded from the model due to collinearity (Pearson correlation 0.84-0.86). The between-group difference for each continuous or binary outcome would be expressed in mean or adjusted odds ratio (aOR) with 95% confidence interval (CI), respectively. GEE analysis can account for correlations between repeated measurements for any individual patient and accommodate missing data caused by dropout, provided these data are missing at random.

Because HbA1c<7% and KPIs would be null values at baseline in both groups, we would run separate multiple logistic regression analysis that incorporate these data at year 1 and year 3, with group allocation as an independent variable, adjusting for the aforementioned covariates. Data would be analysed using SPSS version 24.0 (Chicago, IL). A two-tailed P-value<0.05 would be considered significant.

Sample size determination: In this quasi-experimental real-world experiment with comparative group, we estimate that 100 patients in each group would give 80% power at α=0.05 to yield a 0.5% difference at one year, with comparable baseline HbA1c levels and a standard deviation (SD) of 1.2% and 10% attrition rate.