Effect of 6 months’ flash glucose monitoring in adolescents and young adults with type 1 diabetes and suboptimal glycaemic control: managing diabetes in a ‘flash’ randomised controlled trial protocol

Adolescence is a high-risk period where glycaemic control is at its worst [18], treatment adherence is a particular challenge [19] and the strain of self-management takes a toll on adolescents’ and their parents’ quality of life [20]. Type 1 diabetes exchange data indicate only a minority of young people meet international guidelines for glycaemic control (HbA1c < 58 mmol/mol [< 7.5%]) [21]. During this life stage, barriers to treatment adherence include the major physical and cognitive changes and increasing independence regarding eating behaviour, physical activity and other aspects of lifestyle, such as sleep, increase the burden of managing T1D [22]. Most adolescents with T1D do not adhere to SMBG recommendations [19], with some adolescents undertaking SMBG only when experiencing symptoms of low or high blood glucose levels. Frequent misreporting and fabrication of SMBG data has also been described [23, 24].

Flash glucose monitoring (FGM) technology, sometimes referred to as intermittent continuous glucose monitoring (iCGM), is an accurate, safe, and acceptable approach to monitoring interstitial glucose levels in children (≥4 years) and adults [25‐27]. FGM provides an up-to-date interstitial glucose level, a graph of retrospective data and predicted glucose trend when the user scans the sensor with a handheld receiver [26]. These interstitial glucose measurements have been found to be accurate compared with capillary blood glucose reference values [26]. Higher rates of scanning are associated with indicators of better glycaemic control, such as increased time in range (defined as glucose levels between 3.9 and 10.0 mmol/L [70–180 mg/dL]), reduced time in hyperglycaemia (> 10 mmol/L [> 180 mg/dL]), and improved HbA1c [28, 29]. FGM is less expensive than a similar technology, traditional continuous glucose monitoring (CGM), and does not require regular calibration, like the majority of current CGM systems. However, FGM is retrospective, and therefore does not have the CGM feature of automatic alerts in response to pre-defined low or high glucose levels. Adverse events associated with FGM use are typically limited to skin issues associated with sensor insertion and reactions to sensor adhesive [18, 25, 27, 30].

Evidence of the potential to increase glucose monitoring among adolescents by using FGM technology is emerging [29‐32]. FGM may provide an important opportunity to engage adolescents in their diabetes care by reducing their disease burden and facilitating access to glucose data more frequently, which in turns enables adolescents to make better informed management decisions. The easy instant access to glucose levels that FGM affords is particularly suited for young people where motivation is lacking and the burden of disease is high [33]. However, to our knowledge, there is no evidence demonstrating the superiority of FGM over SMBG for improving clinical outcomes in this challenging patient population. A 6-month randomized controlled trial evaluating FGM among adults with well controlled T1D reported reduced time in hypoglycaemia in the intervention group compared to the SMBG group, but no significant change in HbA1c [31]. Short-term improvements in glycaemic outcomes, including time in range and reduced time in hyperglycaemia, as well as improvements in HbA1c, have been observed in a single-arm study among children and teenagers using FGM for 8 weeks [30]. There is evidence of clinically significant improvements in HbA1c (by ≥0.5% in HbA1c) among T1D patients aged 1–25 years being sustained at 12 months [32] in a real-world setting. This finding is similar to some other diabetes technology, with a recent insulin pump study among adolescents with unhealthy glycaemic control showing sustained improvements in HbA1c from baseline out to 12 months [34]. The same study reported short-term improvements in diabetes-specific and generic health- related quality of life (HRQoL) related to insulin pump use that were not sustained [34]. Research is clearly warranted to investigate the effectiveness of FGM on glycaemic control among young people, particularly those with the unhealthiest glycaemic control (who are usually excluded from manufacturer-funded studies and are plausibly the group with the greatest potential to benefit if FGM is effective). Broader impacts of FGM on salient outcomes such as fear of hypoglycaemia, diabetes treatment satisfaction, and HRQoL also warrant investigation.

The overall study aim is to evaluate, in a randomised controlled trial, the clinical effectiveness of FGM compared to SMBG for adolescents with T1D and a history of suboptimal glycaemic control. The primary objective is to investigate the effectiveness of FGM compared to SMBG in reducing HbA1c (i.e., the primary analysis will be on a modified intention to treat basis, using all available data).

The secondary objectives are to investigate:

The tertiary objectives are to explore:

It is hypothesized that compared to SMBG, FGM will result in a clinically relevant (at least 1.0%; 10 mmol/mol) and statistically significant reduction in mean HbA1c at 6 months’ post-baseline. Additionally, it is hypothesized that there will be statistically significant improvements in other important outcome measures in health, wellbeing, and diabetes management in the FGM intervention group at 6 months, compared to the SMBG group.

Managing Diabetes is a ‘Flash’ is a two phase study programme consisting of a multi-centre randomised, parallel-group study consisting of a 6-month comparison between SMBG and FGM, including a 6-month continuation phase (see Fig. 1). Once enrolled, the total study period is 12 months for each participant (12 months of FGM for the intervention group will enable an investigation of sustained FGM outcomes). During the continuation phase, the FGM group will continue the intervention for a further 6 months and the control group will cross over and receive a 6-month FGM intervention. The primary purpose of offering FGM to the control group at 6 months is to encourage retention and engagement.

Scientific peer review was undertaken both internally, by the Department of Women’s and Children’s Health, University of Otago, and externally, by the Cure Kids Medical and Scientific Advisory Committee. The Southern Health and Disability Ethics Committee approved this study (17/STH/240). The protocol underwent Māori (indigenous New Zealanders) consultation, which fostered input into this study. All district health boards approved recruitment and conduct of the study at their site. The trial was officially registered with the Australian New Zealand Clinical Trials Registry (ACTRN12618000320257p; http://​www.​ANZCTR.​org.​au/​ACTRN12618000320​257p.​aspx) on 5 March 2018 and was issued a Universal Trial Number (U1111–1205-5784) by the World Health Organization International Clinical Trials Registry Platform. The University of Otago is assuming overall responsibility for the initiation and management of the trial (i.e., is the Trial Sponsor). The study is in no way funded or affiliated with the manufacturer of the FGM system under investigation or any manufacturer of a competing system.

Adolescents with T1D meeting eligibility criteria for age, duration since diabetes diagnosis, and absent CGM/FGM use in the previous 4 months will be invited to participate in the study during routine diabetes clinic visits. The diabetes clinics were selected where an investigator was affiliated with both the University of Otago Medical School and a district health board (New Zealand’s regional organisations responsible for delivering publicly funded healthcare services to both urban and rural patients). Combined, these district health boards (DHBs) serve a population of approximately 1.2 million people [36] and provide care to approximately 619 people aged 13–20 with T1D [37]. It is assumed all patients with diabetes in these catchments would attend these clinics, as minimal private practice in diabetes is provided within New Zealand’s predominantly public health system.

After a potential participant is identified by diabetes clinic staff, they will receive a brief overview of the study and their interest in the study will be ascertained. Patients who are interested in the study will then be pre-screened over the phone by research staff and sent an information sheet, which explains the study objectives and participant responsibilities in full. Research staff will respond to all study-related questions and schedule a screening visit to confirm eligibility. The inclusion and exclusion criteria for assessing eligibility to participate in the study (Table 1) will be used during a screening visit conducted with the potential participant, any accompanying participant support people, family, or whānau (extended family) and research staff. Each centre will pre-screen and refer patients until the target sample size is achieved. Trial investigators will not receive financial or non-financial incentives for enrolment.

Based on our data from an earlier study in this population, the standard deviation of HbA1c amongst those with values 75 mmol/mol (≥9%) or greater (and irrespective of self-monitoring blood glucose frequency) was 18.6 mmol/mol [38, 39]. Assuming a correlation between repeated measures of 0.7, a sample size of 58 (29 per group) would provide 80% power to detect a difference in mean changes for HbA1c of 1% (10.93 mmol/mol) using a two-sided test at the 0.05 level. This would be considered a clinically important difference, of a similar magnitude to other proven technologies such as insulin pumps or CGM [40]. To account for a small amount of missing data and loss to follow-up (anticipated to be < 10%), we will recruit a sample size of 64 (32 participants per group) at baseline. Reasons for non-retention of participants (i.e., consent withdrawn, moving outside of the study DHBs, or lost to follow-up) will be recorded and shown in a CONSORT flow diagram (Fig. 1).

The baseline measurements will be collected approximately 7 days from the screening visit. Follow-up assessment visits will be scheduled within a 14-day window, defined as 7 days before and 7 days after the due date. Assessment visits will be undertaken by data collectors trained centrally on the study requirements and measurement procedures. Blinding participants and research staff to group allocation will not be feasible. Participants will receive a $20 gift voucher in recognition of participation at each of the baseline and follow-up assessment visits.

The primary outcome measure is the difference in change in HbA1c between groups at 6-months.

All other measures are for secondary and tertiary (exploratory) outcomes. The timing of all assessments is presented in Table 2.

All participants will be sent a link to an electronic questionnaire fortnightly, via both text and email, to report any episodes of cutaneous problems (e.g., infection, itching, rash, pain, subcutaneous haemorrhage), severe hypoglycaemia (i.e., child experiences altered mental status and as a result is unable to assist in their care, or is semiconscious or unconscious) [41], diabetic ketoacidosis (DKA), or missing school/work due to their diabetes. Up to four contact attempts will be made to non-responders. Data from safety surveys will be reviewed by research staff upon receipt by the study team. Participants will be referred to their general practitioner, diabetes team, or emergency department, as appropriate, for management of medical events. Adverse events (events which result in harm to a participant) will be reported to an internal safety monitoring committee made up of co-investigators with a clinical background and expertise in trials and diabetes.

The adverse events questionnaire will also collect data on FGM performance (i.e., sensor problems, reader problems), FGM adherence (i.e., duration of time each sensor is worn) and blood glucose meter performance (i.e., device malfunctions).

All study participants will be assigned a non-informative study identification number. Data will be recorded and stored electronically in REDCap, which is hosted at the University of Otago. REDCap features will help ensure adherence to time-frames, compliance to measurement procedures, and completeness of data. Data will be routinely checked for missing and/or erroneous values by the study coordinator. Only research staff and investigators will have access to the electronic study records. The ethics committee did not require auditing for this study.

At the end of the study, original data collection sheets and written informed consent will be stored securely at the lead site along with copies of all data collected electronically. At the end of the study, the lead investigator will retain an electronic copy of the cleaned data set, with all identifying information removed.

Means and standard deviations (SDs) will be reported for continuous variables (i.e., age, diabetes duration, HbA1c, glucose monitoring behaviour, questionnaire scores) that are approximately normally distributed, geometric means and SDs for those that are approximately log-normally distributed, and medians and IQRs otherwise. Categorical variables will be described as the number and percent of participants in each category (i.e., gender, ethnic group, insulin regimen). Adverse events will be categorised by the type of event (e.g., severe hypoglycaemia, DKA, sensor insertion issues, sensor wear issues). The frequency of events will be reported by event category.

A mixed linear model will be used to test the difference in HbA1c, with the primary outcome being at 6-months, between the intervention and control groups, using data from all time points with a random participant effect used to accommodate the repeated measures and a group-time interaction to model differences in longitudinal changes. Changes in number of glucose measurements performed and patient-reported outcome measures (i.e., PedsQL generic core scale, PedsQL diabetes module, HFS, and DTSQs scores) from baseline through to 6 months will be calculated by comparing scores from control and intervention group participants using Poisson (or negative binomial regression if there is evidence of over-dispersion) for count outcomes and linear mixed models for continuous outcomes with all collected data included. Model residuals will be examined along with the distribution of random effects. If residuals are positively skewed, natural logarithmic transformations will be investigated and retained if these improve the satisfaction of model assumptions. Mixed quantile regression will be investigated for continuous outcomes if issues with model residuals remain following such transformations. If HbA1c values of 130 are observed, mixed Tobit models with right censoring at 130 will be used for this outcome instead. All mixed models will use restricted maximum likelihood (REML) estimation for the variance components.

Some secondary analyses, such as those involving the wait-list control after they cross over, will only include participants from one group and so group (and group-time interaction) terms will not be included in these models.

The main analyses will follow a modified intention- to-treat principle with all participants analysed in the group to which they were randomised, regardless of actual sensor wear. Per protocol analyses will be also be investigated by including only those from the FGM group who scanned at least 12 out of 14 days in the two weeks prior to the 6-month study visit and excluding any participants from the SMBG group who used FGM at any time during the study. All models will include variables used in minimization [59].

Any missing HbA1c values will be imputed using multiple imputation with chained eqs. (100 imputations after at least 100 burn-ins) and with the imputation model including: gender, HbA1c status at baseline as well as values of that variable from other time points, insulin regimen, frequency of self-monitored glucose (capillary or interstitial) and socioeconomic status. If missing data exceeds 10%, plausible scenarios involving informative missing data will be investigated to explore the robustness of findings.