This study was performed using a crossover design in which each participant underwent a total of three overnight study visits which included a 60 minute exercise session. For all visits, basal insulin was discontinued during the exercise followed by a 50% decrease in basal rate for 45 minutes after the exercise. In random sequence, each participant completed three intervention conditions: one visit with a bedtime dose of oral terbutaline, one visit with an overnight reduction in basal insulin rate, and one visit with no overnight intervention. The study was approved by both the Colorado Multiple Institutional Review Board and the Clinical and Translational Research Center (CTRC) at The Children’s Hospital in Aurora, Colorado. The trial is registered at (#NCT00974051). All study visits occurred at the CTRC.
Patients were eligible for inclusion in the study if they were aged 10-17 years at enrollment with a duration of diabetes greater than 12 months, on an insulin pump for greater than six months and with a HbA1c level < 10%. Patients were excluded if they had a severe hypoglycemic event in the last six months, a BMI > 95th percentile or < 5th percentile, a history of hypertension, recent use of ß-agonist or ß-blockade therapy, a known history of cardiovascular disease, a structural heart defect, or an inability to perform exercise due to physical limitation.
Eighteen patients with T1DM who had previously expressed interest in research studies were approached by study staff and enrolled after written informed consent was obtained. One patient came for the first study visit but then withdrew from the study due to acute illness at the time of the scheduled second visit, and another patient was ill at the time of the first study visit, withdrew as a result and thus did not complete any study visits. Thus 16 patients completed the study. At screening, all participants had a complete history and physical exam including direct Tanner staging, and blood for HbA1c analyzed using a DCA 2000+ Analyzer (Bayer, Tarrytown, NY). Eligible patients underwent randomization by computer program to determine the sequence of the three crossover arms.
Each of the three overnight visits in the CTRC had to be at least 72 hours apart and consisted of the same protocol. Patients were asked to abstain from exercise for the 24 hours prior to the study visit, and to perform an infusion set change the day before. Admission to the CTRC occurred between 11 a.m. and noon. Lunch occurred at noon and consisted of a frozen meal with known carbohydrate content. A peripheral intravenous catheter for sampling was inserted and a BG level was checked. An insulin bolus via the pump was given according to the patient’s home insulin to carbohydrate ratio and correction factor. At each subsequent visit, the lunch meal was exactly the same as the first visit.
The BG level was then checked hourly between 2 and 4 p.m. BG levels above 200 mg/dL were corrected with an intravenous regular insulin bolus of 0.05 units/kg to achieve a target glucose between 120-200 mg/dL at the beginning of the exercise session at 4 pm. BG levels were rechecked 15 and 30 minutes after treatment. If necessary the dose of insulin was repeated until the BG was in range. If BG was less than 120 mg/dL, 15-30 grams of oral, fast-acting carbohydrate was given at the discretion of the investigators to achieve the target BG.
The exercise protocol was specifically chosen to be similar to previous exercise studies (12
) and was performed in the same way for all three study visits for each participant. At 4 p.m., with the BG in range, the insulin pump was disconnected to minimize the risk of hypoglycemia during exercise, and a heart rate monitor was attached (Polar USA Inc). After an adequate heart rate signal was confirmed, the patient began exercising on a running treadmill. To standardize the exercise for individual variations in cardiovascular fitness, participants exercised to a target heart rate of 140 beats per minute, with study staff continuously monitoring the heart rate and adjusting the speed and incline of the treadmill to achieve this heart rate. Each session consisted of four 15-minute intervals on the treadmill for a total of one hour of aerobic exercise. A five minute rest occurred between each interval, at which time a BG was checked and hypoglycemia treated with 15-30 grams of juice if the BG was < 70 mg/dL.
At the conclusion of the fourth 15-minute interval, each patient had a BG checked, reconnected their pump under supervision and restarted their basal insulin rate at 50% of their usual rate for 45 minutes. After this period, the usual basal rate was recommenced and dinner was ordered. Upon arrival of dinner at 7 p.m., a BG was performed and an insulin bolus given for carbohydrates and any necessary correction as per their usual home regimen using the pump’s bolus calculator.
A bedtime snack was allowed if part of the patient’s usual home regimen, and insulin was dosed according to the patient’s home bolus ratios. At 9 p.m., study conditions commenced. With the order of interventions randomized, each patient received either a 2.5 mg oral dose of terbutaline (“terbutaline visit”), had their pump basal insulin infusion rate decreased by 20% from 9 p.m. to 3 a.m. (“basal reduction visit”) or received no intervention with 100% of their home basal insulin infusion rate continued through the night (“control visit”). The BG was checked at the bedside via a free flowing sample from the IV every 30 minutes. Blood pressure and pulse rate were measured every 30 minutes.
If the patient experienced a BG < 70 mg/dL, treatment with 15 grams of juice was given and the BG was rechecked 15 minutes later. Repeat doses of juice were given until the BG was > 70 mg/dL, and then the protocol continued as above. At 6 a.m. study conditions ceased and breakfast was ordered for the patient. After a breakfast bolus using the patient’s standard home carbohydrate and correction ratios was given and breakfast was completed, the participant was discharged from the CTRC.
We defined hypoglycemia as a blood glucose level < 70 mg/dL in accordance with the ADA definition.(16
) Our sample size estimation was based on an expected 48% frequency of nocturnal hypoglycemia as observed by DirecNet following afternoon exercise (12
Our pre-determined outcome measures were the overnight BG nadir and the reduction of nocturnal hypoglycemia. In secondary analysis, we also looked at frequency of BG levels < 80 and ≥ 250 mg/dL.
This crossover study was analyzed using a mixed effect model with subjects within sequence as a random effect, period and treatment as fixed effects for the continuous endpoints of BG nadir, BG max and BG over time. The Chi Square test of independence or Fisher’s Exact was used for the analysis of categorical variables. Data are presented as mean ± SD, least square mean ± SE or frequency and percent, as indicated.