The study was conducted by the Diabetes Research in Children Network (DirecNet) at five clinical centers as an ancillary study to a study evaluating the use of the FreeStyle Navigator™
Continuous Glucose Monitoring System (“Navigator”, Abbott Diabetes Care, Alameda, CA) (13
). A Data and Safety Monitoring Board and the Institutional Review Boards a each center approved the study protocol, consent form and assent form. A parent or guardian and each subject 7 years and older gave written consent and assent, respectively.
The primary eligibility requirements were: 1) age between 3 and <18 years, 2) clinical diagnosis of T1D of ≥1 year duration, 3) use of insulin pump for at least 6 months and 4) home computer with e-mail access. There was no pre-specified HbA1c requirement for eligibility. After completing the first 6 months of the Navigator use study, subjects were given the opportunity to participate in this study concurrently with continued use of the Navigator. The 10 subjects included in this study was a convenience sample size not based on statistical principles.
The protocol consisted of a minimum of 12 nights during which each subject consumed a low fat snack on 6 nights and a high fat snack on the other 6 nights. Protein and carbohydrate content was similar between the two snacks. A minimization algorithm was used to determine the ordering of the snack types, balancing on the pre-snack meter glucose level and self-reported amount of activity during that day (each defined as dichotomous variables). Each night prior to the bedtime snack, the subject checked his/her glucose level with the FreeStyle meter built into the Navigator. If the glucose level was <80 mg/dL, carbohydrate (such as juice) was taken and the glucose level rechecked until the glucose level was ≥80 mg/dL. The last value was considered the pre-snack glucose level. On the DirecNet website, the subject entered the pre-snack glucose level, the level of activity during that day, the amount of carbohydrates to be taken for the bedtime snack, and the amount of insulin to be taken to cover the bedtime snack. Prior to being informed as to whether to have the high-fat or low-fat snack each night, the subject used his/her usual practices for determining the amount of carbohydrate (in 15 gram increments) in the bedtime snack and amount of insulin to be given. The amount of insulin delivered each night was not verified in this study. Upon submission of the data, the website instructed the subject on which snack (high-fat or low-fat) should be eaten on that night.
The high-fat snack consisted of potato chips which for each 30 grams of carbohydrate contained 20 grams of total fat, 5 grams of saturated fat, 2 grams of protein, and 320 calories. The low fat snack consisted of pretzels which for each 30 grams of carbohydrate contained 1.3 grams of total fat, no saturated fat, 2.5 grams of protein, and 138 calories.
The Navigator was used during each of the nights of the study. After the 12 nights were completed, the Navigator data were reviewed. If there are not at least 12 nights with at least 5 hours of sensor data, data were collected for additional nights.
The primary outcome was hypoglycemia, defined as at least one glucose value ≤70 mg/dL on either the Navigator or the FreeStyle meter. Hyperglycemia was defined as at least one value ≥200 mg/dL and at least 50 mg/dL above pre-snack Navigator glucose value on either the Navigator or the FreeStyle meter.
Only nights with at least 5 hours of Navigator values between the snack and 6am were used in the analysis. This included 11 nights where pre-snack meter glucose was initially <80 mg/dL and followed by the carbohydrate treatment. Results were similar when we excluded those 11 nights (data not shown). Glycemic indices based on the Navigator data (mean glucose over night, percentage of glucose values ≤70 mg/dL and percentage of glucose values ≥200 mg/dL) were calculated.
Hypoglycemia and hyperglycemia binary outcomes were analyzed using repeated measures regression models controlling for the pre-snack glucose level, reported activity level during that day (high, low), and the hours of glucose readings. The reported adjusted mean differences were from this model. A permutation test was used to account for the correlated data from the same subject for the comparisons of time to first hypoglycemic event and time to first hyperglycemic event.