Exercise is recommended for type 2 diabetes, which is a global health problem. However, during a moderate-intensity exercise training there is an unpredictable risk of hypoglycemia for patients with type 2 diabetes. Therefore, the investigators aim to explore trends in blood glucose levels in response to a 12-week moderate-intensity exercise training in patients with type 2 diabetes and to explore the predictors of post-exercise blood glucose (PEBG) and exercise-induced glucose response (EIGR). A descriptive and longitudinal design was conducted. Eligible type 2 diabetes patients were recruited from outpatient clinics of a medical center in Taiwan and invited to participate in a 12-week moderate-intensity exercise-training program. Each participant received 36 repeated measures of blood glucose during the exercise training program. Participants were randomly assigned to one of three exercise times (morning/afternoon/evening). Each exercise session was took place 1 to 2 hours after a meal. Capillary blood glucose levels were measured pre- and post-exercise. The EIGR was calculated from subtracting the PEBG from the before-exercise blood glucose (BEBG). Generalized estimating equation was used to examine the trends and predictors of PEBG and EIGR.

A prospective and longitudinal design using purposive sampling was used. All eligible Taiwanese participants with T2DM at a medical centre were invited to participate in the exercise-training programme that included three sessions per week, 30 minutes per session for 12 weeks (total 36 sessions). Each participant's exercise sessions were equally randomised into three times of day (morning/08:00-10:00, afternoon/14:00-16:00 evening/18:00-20:00) using permuted-block randomisation. Capillary blood glucose samples were obtained before and after each 30-minute moderate-intensity (60% VO2max) exercise session. The EIGR, described as the difference between BEBG and PEBG values, was calculated.

Participants and setting: Participants with T2DM being seen at an outpatient clinic in a northern Taiwan medical centre were recruited by an endocrinology/metabolism physician. Participants, 40-60 years of age, diagnosed as having T2DM based on the criterion of the American Diabetes Association, and were being treated with oral antidiabetic medications only were eligible. Other inclusion criteria were (1) able to speak and understand Mandarin; (2) able to walk without assistance; (3) have no regular exercise habit; and (4) agree to join in a 12-week moderate-intensity exercise training after passing a graded exercise test (GXT), described below. Exclusion criteria included insulin therapy, a history of cancer, end-stage renal disease with dialysis, an inability to participate in exercise training due to comorbid neurological and musculoskeletal conditions, severe comorbidity or complications such as heart failure, autonomic neuropathy, and recent stroke within 6 months.

Under an effect size of 0.2, thirty-six numbers of measurements with a correlation efficient of 0.32 among repeated measures, an expected power of 0.7 to 0.8, and an alpha of 0.05, the required sample size was between 12 and 14 based on repeat measures (within factors). Therefore, initially 52 T2DM patients were approached and seventeen eligible participants were invited by a research nurse to receive GXT. Fifteen passed the exercise test based on the guidelines of the American College of Sports Medicine (ACSM) for exercise testing. A second research nurse assessed the baseline characteristics (eg, self-reported demographics and lifestyle patterns, blood analyses and anthropometric measures) of the final 15 participants. Participants were then invited to perform 36 exercise sessions (3 sessions/week for 12 weeks), which were randomly assigned to morning, afternoon, or evening using permuted-block randomization, except two dropped out after exercise test.

Graded exercise test: The GXT was performed using a motorised treadmill (Trackmaster 400, JAS Fitness System, USA) for identification of maximum oxygen uptake (VO2max), maximum heart rate (HRmax) and the risk for potentially life-threatening cardiovascular disease (eg, myocardial infarction). GXT procedures followed a modified Balke protocol.20 Two observers (a medical/rehabilitation doctor and a research nurse) were involved with each patient throughout the GXT. The GXT was immediately stopped if the patient complained of exhaustion; reached maximum heart rate or VO2max, determined by the maximal amount of blood the heart pumps per minute (cardiac output) and the amount of oxygen utilised by the exercising muscles (arterial-venous oxygen difference); had a respiratory exchange ratio >1.15; developed symptoms such as chest pain, dyspnea, pallor, diaphoresis or dizziness; had systolic pressure greater than 250 mm Hg or diastolic pressure greater than 120 mm Hg; had a decrease in systolic pressure of more than 10 mm Hg compared to the systolic pressure at rest; or requested stopping the test.

Ethical consideration: Institutional review board approval (TSGHIRB: 097-05-157) was obtained from Tri-Service General Hospital in Taiwan. All participants gave written informed consent when invited to join the study. Participants were assured that their participation was entirely voluntary and that they could withdraw at any time.

Data analysis: Statistical analyses were performed by SPSS version 16.0 (SPSS Inc., Chicago, IL). Descriptive data are presented as mean/standard deviation (SD) and numbers/percentage (%). Before evaluating the trend of blood glucose response during the 12-week/36-session exercise-training programme, two variables including "training month (1st, 2nd and 3rd month of the training)" and "timing of exercise sessions (1st to 36th session of exercise)" were coded. With generalised estimating equations (GEE), the investigators examined the change patterns of BEBG, PEBG and EIGR over time during the training programme. The trend analyes of BEBG, PEBG and EIGR were adjusted for covariates (i.e, age, gender, baseline body mass index, VO2max, HgbA1c, antidiabetic medication (metformin, sulfonylureas or repaglinide) and exercise time of day (morning, afternoon or evening). When evaluating the predictors of EIGR and PEBG, univariable analysis was applied first and, following the multivariable analysis, incorporated with those significant covariates from univariable analysis. All of the statistical analyses were two-tailed and considered significant at p < .05.

Inclusion Criteria:

• 40-60 years of age
• diagnosed as having T2DM based on the criterion of the American Diabetes Association
• being treated with oral antidiabetic medications
• able to speak and understand Mandarin
• able to walk without assistance
• have no regular exercise habit
• agree to join in a 12-week moderate-intensity exercise training after passing a graded exercise test.

Exclusion Criteria:

• receiving insulin therapy
• a history of cancer, end-stage renal disease with dialysis
• inability to participate in exercise training due to comorbid neurological and musculoskeletal conditions
• severe comorbidity or complications such as heart failure, autonomic neuropathy, and recent stroke within 6 months.

After our research results were published. We will share our Study Protocol, Statistical Analysis Plan (SAP), and Clinical Study Report (CSR).