The trial was conducted by the Diabetes Research in Children Network (DirecNet). The protocol was approved by the institutional review boards of the five participating sites. Written informed consent was obtained from the parents/guardians, and the child’s assent was obtained when appropriate. The study is listed on www.clinicaltrials.gov
Study participants had a clinical diagnosis of type 1 diabetes and were using daily insulin therapy for at least 12 months. Eligibility criteria included age 4.0 to <10.0 years, HbA1c ≥7.0%, and basal-bolus therapy using either an insulin pump or at least three multiple daily injections (MDIs) of insulin for the prior 3 months with no plans to switch the insulin modality within the next 6 months. Exclusion criteria included 1) diagnosis of diabetes prior to 6 months of age; 2) use of a medication that could affect glycemic control, the performance of the CGM sensor, or completion of any aspect of the protocol; and 3) use of CGM during the prior 6 months.
After enrollment, participants had a run-in period for a minimum of 6 weeks to optimize glycemic control prior to CGM use. During the run-in period, a blinded CGM device was then used for 2 to 4 weeks to familiarize the participant and parent with its use and to obtain CGM data as a baseline assessment of glycemic control. To be randomized, participants had to wear the CGM for a minimum of 7 of 14 days; have no severe skin reaction at the insertion site; have at least 96 h of CGM values, including at least 24 h during 10 p.m. to 6 a.m.; and have performed a minimum of three blood glucose meter measurements per day. Participants meeting these criteria were randomly assigned to either the CGM group or the usual care control group, using a permuted-blocks design stratified by clinical center.
Participants randomized to the CGM group were provided with an unblinded CGM device, sensors, and FreeStyle Flash (Abbott Diabetes Care, Inc., Alameda, CA) blood glucose meter and test strips. A FreeStyle Navigator (Abbott Diabetes Care, Inc.) was provided unless the participant was already using a Medtronic Paradigm insulin pump (Medtronic MiniMed, Inc., Northridge, CA), in which case a MiniMed MiniLink REAL-Time Transmitter (Medtronic MiniMed, Inc.) could be used. Parents were instructed on the use of the device and encouraged to use the sensor on a daily basis. They were instructed to continue testing with the home blood glucose meter ≥4 times each day and to verify the accuracy of the CGM glucose measurement with the home blood glucose meter before making management decisions. Participants in the control group were given a FreeStyle Flash blood glucose meter and test strips and asked to perform blood glucose monitoring at least four times daily.
Parents of participants in both the CGM and control groups were provided with detailed verbal and written instructions on how to use CGM and blood glucose meter data, respectively, to make real-time insulin dose adjustments and on using computer software to retrospectively review the glucose data to alter insulin dosing (if a computer was available at home for downloading). The insulin dose adjustment algorithms are available on the DirecNet public Web site (http://direcnet.jaeb.org/Studies.aspx?RecID=162
). Target glucose values were 80–150 mg/dL before meals, <200 mg/dL after meals, 100–150 mg/dL at bedtime, and 80–150 mg/dL overnight.
The number of scheduled contacts was identical for both treatment groups. Visits were conducted at 1, 4, 8, 13, 19, and 26 weeks (± 1 week) postrandomization, with one scheduled phone contact between each visit, to review glucose data and adjust diabetes management, as indicated. After the 13- and 26-week visits, the control group wore a blinded CGM device to collect a minimum of 96 h of glucose values overall, with at least 24 h overnight.
Bayer DCA (Tarrytown, NY) point-of-care devices were used for HbA1c
measurements at each visit with the exception of the 1-week visit. A blood sample was collected at baseline, 13 weeks, and 26 weeks for measurement of HbA1c
at the University of Minnesota using the Tosoh A1c 2.2 Plus Glycohemoglobin Analyzer (Tosoh Medics, Foster City, CA) method (12
). The parent completed the following questionnaires at baseline (prior to initiating use of the blinded CGM device) and at 26 weeks: Glucose Monitoring Survey (13
), Pediatric Assessment In Diabetes Survey–Parent Version (PAID) (14
), and Hypoglycemia Fear Survey (15
). In addition, the CGM Satisfaction Scale (13
) was completed by the parent for those in the CGM group at 26 weeks. Severe hypoglycemia was defined as an event requiring assistance of another person, as a result of altered consciousness, to administer carbohydrate, glucagon, or other resuscitative actions. For those developmentally too young to independently recognize and treat hypoglycemia, hypoglycemia was considered severe if there were associated signs or symptoms of neuroglycopenia, including temporary impairment of cognition; incoherent, disoriented, and/or combative behavior; seizure; or coma.
The primary outcome was a binary variable, defined as a decrease in HbA1c
of ≥0.5% from baseline to 26 weeks and no severe hypoglycemic events. A sample size of 140 was planned to have 90% power to detect an absolute difference in this outcome between treatment groups of 25%, assuming a control group rate of 10% (based on control group data from the JDRF CGM RCT) (2
), an α-level of 0.05, and ≤7.5% losses to follow-up.
Analyses followed the intent-to-treat principle, with all participants analyzed in the group to which they were randomized regardless of actual sensor use. Treatment group comparisons of binary outcomes were performed with logistic regression models, adjusted for baseline HbA1c level and clinical center. The comparisons of continuous outcomes, including HbA1c, questionnaire data, and CGM glucose data, were made using ANCOVA models adjusted for their corresponding baseline values and clinical center. There was one outlier for HbA1c (changed from 10.5% at baseline to 7.9% at 26 weeks), and the results were similar using a rank transformation (data not shown). In addition, CGM glucose data were transformed using van der Waerden normal scores, and the comparisons were adjusted for baseline HbA1c level and type of CGM device. The percentages of participants with at least one severe hypoglycemic event in the two treatment groups were compared using Fisher exact test, and the incidences were compared using a permutation test.
Among participants in the CGM group, change in the amount of CGM use over time was assessed using a repeated-measures regression model based on van der Waerden transformed scores. Spearman rank correlations between the amount of CGM use and age, baseline HbA1c, and change in HbA1c were computed. Changes in HbA1c from baseline in the participants who wore CGM ≥6 days/week vs. <6 days/week were compared using least squares regression model adjusted for baseline HbA1c level and clinical center.
Analyses were conducted using SAS version 9.2 (SAS Institute, Cary, NC). All P values are two-sided.