Authors: Sarah S. Jaser (aDepartment of Pediatrics, Vanderbilt University Medical Center, Nashville, TN, USA), Jill Simmons (aDepartment of Pediatrics, Vanderbilt University Medical Center, Nashville, TN, USA), Lauren L. Milner (aDepartment of Pediatrics, Vanderbilt University Medical Center, Nashville, TN, USA), Charity E. Davis (aDepartment of Pediatrics, Vanderbilt University Medical Center, Nashville, TN, USA), Samantha M. Davis (aDepartment of Pediatrics, Vanderbilt University Medical Center, Nashville, TN, USA), Tabitha C. McCarty (aDepartment of Pediatrics, Vanderbilt University Medical Center, Nashville, TN, USA), Lauren LeStourgeon (bDepartment of Medicine, Division of General Internal Medicine & Public Health, Vanderbilt University Medical Center, Nashville, TN, USA), Beth A. Malow (cDepartment of Neurology, Sleep Division, Vanderbilt University Medical Center, Nashville, TN, USA), James C. Slaughter (dDepartment of Biostatistics, Vanderbilt University Medical Center, Nashville, TN, USA), Kashope Anifowoshe (aDepartment of Pediatrics, Vanderbilt University Medical Center, Nashville, TN, USA), Lori C. Jordan (aDepartment of Pediatrics, Vanderbilt University Medical Center, Nashville, TN, USA)
Categories: Article, Type 1 diabetes, Sleep, Sleep intervention, Executive function, Behavior, Adolescents
Source: Contemporary clinical trials
Authors: Sarah S. Jaser, Jill Simmons, Lauren L. Milner, Charity E. Davis, Samantha M. Davis, Tabitha C. McCarty, Lauren LeStourgeon, Beth A. Malow, James C. Slaughter, Kashope Anifowoshe, Lori C. Jordan
Despite improvements in hemoglobin A1c among adolescents with type 1 diabetes (T1D) over the past several years, most are still not meeting recommended glycemic targets, placing them at elevated risk for acute and long-term health complications. Thus, there is a critical need for novel approaches to improve diabetes management in adolescents with T1D. Insufficient and inconsistent sleep affects glycemic outcomes directly through decreased insulin sensitivity, and indirectly via compromised executive function in adolescents, reducing their ability to effectively manage T1D. Via a randomized controlled trial of 150 adolescents (age 11–17 years) with T1D, we will evaluate the effects of a sleep-promoting behavioral intervention that includes individual coaching on sleep duration and timing. We hypothesize that adolescents randomized to the Sleep Coach intervention will exhibit significantly longer sleep duration and reduced sleep variability as compared to those who receive enhanced usual care, consisting of additional diabetes education materials and text messages. Effects of the sleep-promoting intervention on executive function, glycemic outcomes (hemoglobin A1c, Time in Range) and diabetes management will be evaluated.
Primary outcomes (sleep duration and timing) and secondary outcomes (executive function, assessed with NIH Toolbox measures and caregiver reports, and glycemic outcomes) are assessed over 12 months. Intent-to-treat analysis will be used to evaluate efficacy of the intervention.
If efficacious, Sleep Coach has the potential to improve both cognitive and glycemic outcomes in adolescents with T1D. This individualized, interactive, manualized behavioral intervention can be delivered remotely by trained staff, with the potential for wide dissemination.
Type 1 diabetes (T1D) is one of the most common chronic health conditions in youth, with increasing prevalence [1]. The recommended regimen for managing diabetes is complex and demanding, including frequent glucose monitoring, insulin administration (via injections or pump) multiple times per day, and careful tracking of diet and activity levels [2]. Maintaining glucose levels in the target range (70–180 mg/dL) reduces the risk for acute and long-term medical complications [3], but the intensive level of self-management behaviors required to achieve this target is difficult to maintain. Problems with diabetes management are reported by the majority of adolescents with T1D (up to 93 %) [4], and diabetes distress (the emotional burden of living with diabetes) is commonly experienced by youth with T1D [5]. Despite advances in technology (e.g., insulin pumps, continuous glucose monitoring, automated insulin delivery systems), challenges with T1D management in adolescence persist; a national study found significant improvements in hemoglobin A1c (HbA1c) in recent years, but most people with T1D were still not meeting glycemic targets [6]. Higher HbA1c has also been associated with neurocognitive dysfunction and reduced brain white matter integrity seen via diffusion tensor imaging [7]. Novel interventions are needed to improve diabetes management and reduce the risk of cognitive dysfunction in adolescents.
Sleep problems in T1D have recently gained attention as a potential risk factor for poor T1D outcomes, and emerging evidence indicates that the associations between sleep and diabetes outcomes are complex and bidirectional – poor sleep contributes to worse glycemic outcomes, and greater glycemic variability likely causes sleep disruptions [8,9]. In adults with T1D, reduced sleep duration increased evening cortisol and growth hormone levels (hormones counter-regulatory to insulin) [10] and increased insulin resistance [11]. Results from a recent national survey indicate that over 77 % of adolescents in the general population obtain insufficient sleep (defined as <8 h/night for those 13–17 years of age) [12], and a meta-analysis revealed that youth with T1D sleep significantly less than those without diabetes [13]. Yet, sleep characteristics, such as total sleep time, sleep/wake times, and sleep quality, are not routinely addressed in youth with T1D.
In addition to its direct effects on glycemic outcomes, insufficient sleep may affect outcomes indirectly through poor diabetes management [9]. Thus, by improving sleep quality and duration with a targeted behavioral intervention termed “Sleep Coach,” we believe we can break this negative cycle, improving outcomes in adolescents with T1D through both an indirect, behavioral pathway (i.e., diabetes management) and a direct pathway (i.e., glycemic outcomes) (Fig. 1).
While sleep disturbances have biological, circadian, and neuro-developmental causes, they are also influenced by modifiable environmental and psychosocial factors [14]. The biological shift in sleep/wake times associated with a delay in melatonin secretion, combined with early school start times, contribute to shorter sleep duration in adolescents [15]. In addition, elevated use of electronic media, extracurricular activities, work, academics, and social interaction (e.g., texting, social media) are linked with later bedtimes and difficulty falling asleep [16]. Caffeine use, which increases in adolescence, is also strongly associated with sleep disturbances [17]. Finally, social jetlag, which occurs when individuals sleep less on week nights and compensate by sleeping more on weekend nights [18], is common among adolescents. In our published data, variability in sleep duration (related to longer sleep duration on weekends) was more strongly associated with HbA1c and diabetes management than sleep duration or quality [19], and therefore may represent an important target. Many barriers to obtaining sufficient sleep can be addressed by a sleep-promoting intervention, as shown in our prior pilot study [20].
Advances in diabetes technology, including the use of automated insulin delivery systems, hold the potential to improve glycemic outcomes, increasing time in range and reducing glycemic variability, especially overnight [6]. However, the effects of automated insulin delivery systems on sleep are mixed; while children and adolescents using these systems report improvements in sleep quality, objective sleep data indicate no improvement in sleep duration [21].
Studies on sleep disturbances in children and adolescents demonstrate negative effects on cognitive function, especially executive function skills (e.g., problem-solving, working memory) [22,23], that are essential for T1D management [24]. T1D is one of the most complex of all chronic medical conditions to manage [2], and by late adolescence, most patients are managing their T1D independently, which requires complex executive function skills, including planning, problem-solving, and organization. Thus, the impact of insufficient sleep on brain function may be more significant for adolescents with T1D than in the general population.
The study will evaluate the efficacy of a sleep-promoting intervention via a randomized controlled trial (RCT). Adolescents will be randomized to parallel groups in a 1 allocation. The primary outcome is sleep duration and consistency, and the secondary outcomes are executive functioning and diabetes outcomes.
Prior to finalizing the protocol, we updated intervention materials from the pilot study with relevant information on new diabetes technology and feedback from community members (adolescents with T1D and their caregivers). We conducted a Community Engagement Studio, a consultative model [25], to enhance our plans for recruitment and retention. The Meharry-Vanderbilt Community Engagement team, which is part of the Vanderbilt Institute for Clinical and Translational Research, led the Studio with six participants (three adult parents/caregivers and three adolescents with T1D). Participants provided recommendations and feedback about various aspects of the study protocol related to recruitment and retention, including the best way to communicate with adolescents and their caregivers, the type of compensation the study should provide, the level of parent/caregiver engagement with study participants, and what study-branded retention items they preferred. Participants suggested sharing recruitment updates during the study, updating parents on their child’s progress, promoting retention by providing journals and pens (to keep the study upfront in participants’ minds), and communicating to participants that the study involves minimal risks.
We plan to enroll a sample of 150 adolescents with T1D and their caregivers. Adolescents will be eligible if they (1) are 11–17 years of age; (2) have been diagnosed with T1D for ≥12 months; (3) report insufficient sleep on most nights (< 8 h/night for ages 13–17, <9 h/night for ages 11–12) [26] but have no other sleep disorders or sleep apnea; (4) are not meeting the target for HbA1c (<7 %); (5) use a Continuous Glucose Monitor (CGM) and (6) read/speak English. The eligible age range captures a key developmental stage, when youth are at greater risk for problems with diabetes management and insufficient sleep. Adolescents diagnosed with T1D for at least 12 months will be studied to avoid confounding the effects of the initial adjustment period after diagnosis [27]. We plan to enroll a sample that reflects the clinic demographics with approximately equal numbers of males and females and 75 % white Non-Hispanic youth.
We will recruit from a pediatric diabetes clinic in an academic medical center during adolescents’ regularly scheduled clinic visits. Interested adolescent and caregiver dyads will be screened using questions from the Pediatric Sleep Questionnaire [28] to assess for obstructive sleep apnea and questions about typical sleep duration (completed by caregivers on behalf of their adolescents). Those that screen as at high risk for obstructive sleep apnea will be advised to discuss referral for polysomnography with their pediatrician and will not be study eligible. After providing informed e-consent/assent, eligible participants will complete the baseline questionnaires in REDCap (Research Electronic Data Capture). REDCap is a secure, web-based application designed to support data capture for research studies [29] and has been used effectively in other studies with adolescents with T1D.
As an objective measure of sleep characteristics, adolescents will be asked to use a wrist-worn accelerometer (the Fitbit Charge 6) continuously for seven days. Fitbits have been widely used for sleep tracking in studies with adolescents [30,31]. The Fitabase platform will be used to analyze participants’ synced Fitbit data, including minutes asleep, minutes in bed, and sleep stage data [32,33]. Fitabase uses encryption and logging software and is password-protected to maintain the security and privacy of the database. Participants will be deidentified in the Fitabase database (only study ID will be used). Concurrently, adolescents will complete daily sleep diaries in the evenings to track daytime behaviors (e.g., caffeine use, screen time, physical activity) and nocturnal sleep characteristics (e.g., bedtime, awakenings due to high or low blood glucose levels). Sleep diary data will be collected via MyCap, a participant-facing mobile application for survey data collection [34]. The combination of Fitbit activity tracking and MyCap surveys will provide more accurate information about sleep duration and characteristics. To reduce challenges with the digital aspects of the study, participants will complete most of the tasks in person, and the study staff will help participants set up their watches and download the Fitbit and MyCAP apps during the initial study visit. Study staff will be in regular communication with participants to troubleshoot technological difficulties.
After sharing sleep data and completing the sleep diary for seven days, participants will be randomly assigned 1 to the Sleep Promotion Intervention (n = 75) or Enhanced Usual Care (n = 75) in blocks stratified by age (11–13 vs. 14–17 years old) and insulin administration at baseline (insulin injections vs. insulin pump (not closed loop) vs. automated insulin delivery (AID) system) to reduce the possibility of confounding by insulin delivery system. The study biostatistician will generate a random allocation table which will be imported into the REDCap randomization module.
The intervention consists of 4 phone call sessions (Table 1), with content tailored to meet individual participant’s needs. The duration of the intervention will be 3 months. Adolescents will be mailed the Sleep Coach manual, and a trained interventionist (post-bachelor’s level) will conduct sessions over the telephone. Phone calls will be scheduled at times convenient to adolescents, and we will use text message reminders to confirm sessions. Each session will begin by reviewing the adolescent’s personalized sleep goals for the program, followed by a discussion of skills, and ending with a homework assignment reinforcing newly learned concepts. Interventionists will complete a minimum of 40 h of training on T1D management and training on strategies to improve sleep. A licensed clinical psychologist will provide regular supervision for interventionists.
Session 1 includes education regarding healthy sleep habits and sleep timing. The interventionist discusses potential changes to improve sleep hygiene, and the adolescent is asked to identify goals (e.g., open blinds in the morning, reduce caffeine intake after lunch). In Session 2, the interventionist reviews the goals from the first session and uses motivational interviewing techniques to address barriers to change. In this session, the interventionist also discusses strategies to reduce sleep variability. In Session 3, the interventionist will review goals, and the participant will choose a module to facilitate meeting their sleep goal (Relaxation Training, Coping Thoughts, Diabetes Technology). Finally, the Booster Session focuses on anticipating challenges to obtaining consistent sleep (e.g., studying for exams, vacation), reviews content and helps adolescents problem solve barriers to sufficient sleep. Caregivers receive the same materials and will be instructed to support positive changes in their adolescents’ sleep habits. Based on caregiver suggestions from the Community Engagement Studio, the study team will send weekly text messages during the intervention reminding caregivers to give their child supportive messages.
Participants randomized to the EUC condition will continue with regular outpatient diabetes clinic visits. In addition, they will be mailed a 15-page packet with developmentally relevant diabetes education materials based on publicly available information (diabetes.org). Adolescents in the EUC group will also receive educational messages (i.e., Did You Know? facts) on the same schedule as the intervention sessions, via text message. Engagement will be evaluated based on responses to text messages, as adolescents are prompted to respond with their favorite emoji acknowledging receipt of the educational message. They will have the opportunity to receive the Sleep Coach intervention materials after the 12-month data collection (Fig. 2).
Clinical data (HbA1c, glucose data), sleep and executive function outcomes, and potential moderators will be collected at baseline, 3, 6, and 12 months to evaluate efficacy of the sleep-promoting intervention. Fitbit data and sleep diaries will only be completed at baseline and 3 months. All survey measures have established reliability and validity and take about 20–30 min to complete. The selected NIH Toolbox cognitive function measures take 10–12 min to administer.
Our primary outcome will be sleep duration (total sleep time, average minutes per night, assessed over seven consecutive nights) measured at 3-month follow-up using a Fitbit watch synced with the Fitabase platform (see above for description of sleep data collection). Sleep diaries collected via MyCap will be used to corroborate watch data, and any scoring discrepancies will be discussed with the study team and resolved by our co-investigator who is board-certified in sleep medicine. Data will be scored if there are a minimum of 4 nights of usable data.
Secondary assessments of sleep outcomes include (i) a self-report questionnaire, the Pittsburgh Sleep Quality Index (PSQI) [35], which captures sleep duration and disturbances, as well as overall sleep quality. A global score > 5 indicates clinically significant sleep disturbances; and (ii) the Pediatric Sleep Questionnaire completed by caregivers [28].
Glycosylated Hemoglobin (HbA1c) is an objective measure of blood glucose over the prior 8–12 weeks [36]. Analyses will be performed using the Bayer Diagnostics DCA2000^®^ machine. The recommended goal for HbA1c in adolescents with T1D is <7 % [2]. However, as HbA1c typically peaks in adolescence [37], we are aiming to prevent deterioration over time. Point-of-care values will be extracted from the medical record at baseline, 3, 6, and 12 months. If a clinic appointment is missed, we will send the participant a home HbA1c kit. This process has been used in previous trials, with the increase in telehealth visits during and after COVID-19 [38]. Home kit HbA1c shows strong concordance to Point-of-Care HbA1c in youth (0.98) [39].
Time in Range (TIR), the percentage of time glucose is between 70 mg/dL and 180 mg/dL, is recommended as a more precise measure of glycemia than HbA1c, which reflects average glucose and may obscure frequent episodes of hypoglycemia and hyperglycemia. The target for people with T1D is >70 % TIR [3]. Glucose data is collected at regular diabetes clinic visits from device uploads. If participants are missing data in the medical record, research staff will be able to obtain glucose data from the clinic portal for continuous glucose monitors. While CGM use is required for participation, data will be analyzed if participants have ≥70 % use in the prior 14 days [40].
Diabetes Management Outcomes: The updated Self-Care Inventory (SCI-U) assesses key elements in contemporary diabetes self-management behavior. The SCI-U items apply to various insulin regimens (i.e., pump and injections) and has demonstrated good reliability and validity [41]. The Parent-proxy and Self-report versions will be used to measure diabetes management behaviors, and the mean score will be used in data analyses.
Two NIH Toolbox Dimensional Change Card Sort Test and the List Sorting Working Memory Test will be used to assess adolescents’ executive function skills. These are a common data element that has been used in research with adolescent cohorts [42,43]. The parent-report Behavior Rating Inventory of Executive Function, 2nd Edition (BRIEF2) will also be used to assess adolescents’ EF in daily life [44].
Adolescents’ caregivers will provide information on demographic variables, including child race/ethnicity, parental education, family income, and subjective social status, which has been linked to glycemic outcomes [45]. Adolescents will report on gender and race/ethnicity. Diabetes device use (% time using insulin pump, % time CGM active, % time using AID) will be obtained from adolescents’ medical records and confirmed with caregivers.
Adolescents will complete the Morningness-Eveningness Scale for Children (MESC) [46], a measure of circadian preference, the Pubertal Development Scale to assess pubertal stage [47], the Children’s Hypoglycemic Fear Survey (HFS-C) [48], and the Problem Areas in Diabetes–Teen Version (PAID-T) [49] to capture potential covariates that may affect sleep. Caregivers will complete the Pediatric Sleep Questionnaire (PSQ) sleep-disordered breathing scale [28], a measure of obstructive sleep apnea, and the Hypoglycemic Fear Survey for Parents (HFS-P) [48].
Based on effect sizes from our pilot study on total sleep time after 3 months (d = 0.69) [20], a sample of 150 adolescents with T1D randomized 1 to the Sleep Coach intervention, allowing for 15 % attrition, provides >90 % power to detect a treatment difference of 30 min in sleep time, a clinically significant difference in sleep duration [50].
We will conduct unadjusted tests of the association of demographic covariates including sex, race/ethnicity, income [51], fear of hypoglycemia, and circadian preference with sleep habits. While we anticipate our randomized design will mitigate confounding, adjusting for covariates associated with the sleep outcomes will improve the precision of our estimate of the treatment effect. We will also evaluate fidelity to the intervention content prior to conducting analyses.
A linear regression model will be used to estimate the effect of the intervention over time on adolescents’ total sleep time and variability from baseline to 3 months. All primary analyses will be based on intention to treat with the intervention. Separate linear mixed effects regression models will be used to estimate the effect of the intervention over time on executive function (BRIEF2, NIH Toolbox measures), glycemic outcomes (HbA1c, Time in Range), and diabetes management (parent-SCI-U, SCI-U). Outcomes will be measured at up to four time-points per subject (baseline, 3 months, 6 months, and 12 months), so we will use time, treatment group and the interaction of treatment with time, as the main explanatory variables. Models will allow us to test for any differences between treatment and control groups over time while accounting for baseline values. We will assess the initial intervention effects on executive function, glycemia, and diabetes management at 6 months, and durable effects will be assessed at 12 months.
Missing data will be minimized by pairing research and clinical visits, enabling remote study assessments for most measures, and increasing participant incentives for the 12-month follow-up. However, attrition will be considered in the we will analyze differences between participants with 12-month follow-up and those lost to follow-up and use imputation and carry-forward methods when appropriate.
Based on other studies of behavioral interventions to improve outcomes in youth with T1D [52], our benchmark for retention is ≥85 %. To evaluate acceptability of the intervention, we will examine session attendance, satisfaction ratings, and exit interviews. We chose a feasibility benchmark of ≥75 % completion of sessions, and the sleep diary and Fitabase data will assess protocol adherence (changes in sleep timing and duration). Data from participant satisfaction surveys and exit interviews will be utilized to evaluate the acceptability of the intervention. These semi-structured interviews will consist of open-ended questions asking about participants’ experiences with the study and suggestions for improvements. Interviews will be transcribed and coded using NVivo qualitative data analysis software with a thematic analytic approach to organize patterns (themes) [53]. In addition, fidelity checks will be conducted by audiotaping the telephone sessions, 20 % of which will be randomly selected to be coded for content by an independent rater, with a benchmark of ≥90 % adherence to session content.
Adolescents with T1D are at high risk for problems with diabetes management and most do not meet glycemic targets [6]. Sleep disturbances and insufficient sleep are potentially modifiable risk factors for poor diabetes outcomes in adolescents with T1D. This clinical trial in adolescents with T1D will evaluate the effects of a sleep-promoting intervention on adolescents’ sleep duration and consistency and assess whether this leads to improvements in diabetes management (SCI-U, CGM and other technology use), glycemia (HbA1c, TIR), and executive function (NIH Toolbox, BRIEF2). If effective, this low-cost, sleep-promoting intervention could be rapidly implemented into clinical practice.
The cause of insufficient sleep may be difficult to determine. However, by including measures of circadian preference, pubertal status, sleep apnea, fear of hypoglycemia, and insulin administration we will have information about potential covariates. Further, by using multiple measures of sleep, we will be able to determine whether sleep duration or sleep variability (or both) are most strongly related to diabetes outcomes. In addition, the ways that diabetes devices and technology may impact sleep are unknown. While the use of insulin pumps and AID systems decrease the frequency of hypoglycemia, especially overnight [54], frequent alarms also disrupt sleep, and the use of these systems will be important to assess.
Participants will be randomly assigned to sleep promotion intervention or enhanced usual care in blocks by treatment type (AID vs. pump vs. injections) to reduce the possibility of confounding by treatment type (i.e., adolescents who are using injections may differ from those using AID in terms of caregiver education, household income and race/ethnicity). However, given that the EUC condition is less intensive, it may be that changes among participants who receive the active intervention are related to greater interaction with the study team. Finally, our target sample is predominantly non-Hispanic White, which is consistent with our clinic population, but may not allow us to generalize findings to other racial or ethnic groups.
If the Sleep Coach intervention increases the duration and consistency of sleep, adolescents with T1D may experience improvements through both an indirect, behavioral pathway (i.e., diabetes management as assessed by the validated self-care inventory) and a direct, physiological pathway (glycemia). Our low-cost, sleep-promoting intervention could be rapidly implemented into clinical practice. This trial is innovative in targeting sleep promotion in adolescents with T1D to improve diabetes management, and to understand the mechanisms of and reduce the risk for cognitive dysfunction.