Efficacy of a Personalized Mobile Health Intervention (BedTime) to Increase Sleep Duration Among Short-Sleeping Patients With Type 2 Diabetes: Protocol for a Pilot Randomized Controlled Trial

Introduction

Worldwide, approximately 537 million people aged 20-79 years have diabetes [1]. Diabetes is a serious public health concern, significantly impacting health care expenditures and human life, causing 4.2 million deaths annually [2,3]. Type 2 diabetes (T2D) is difficult to cure once it develops and, if left untreated, can lead to complications such as cardiovascular disease and microvascular damage [4].

T2D treatment guidelines include both pharmacological and lifestyle modification elements. Traditionally, lifestyle modifications have focused on diet and exercise, but recently, improving sleep in patients with T2D has gained prominence [5,6]. Some patients with T2D suffer from sleep disorders, including obstructive sleep apnea, restless leg syndrome, and insomnia, for which treating the underlying disorder is the appropriate approach [7]. Sleep quality is associated with T2D risk [8], and interventions to improve sleep quality have shown effectiveness [9]. Many patients with T2D do not get sufficient sleep, experiencing what is termed short sleep. Definitions of short sleep vary across studies, ranging from 6 or fewer hours per night to less than 4.5 hours per night [10]. This issue is widespread, with data indicating that 39.4% of patients with T2D in Japan sleep fewer than 6.4 hours per night [11].

There is strong evidence that short sleep is associated with an increased risk of T2D and poorer glycemic control. In Japanese patients with T2D, short sleep is linked to higher hemoglobin A_1c (HbA_1c) levels [11], with a meta-analysis indicating a 0.23% difference in HbA_1c due to short sleep [12]. Prospective studies have also shown an increased risk of T2D associated with short sleep [8]. Plausible biological mechanisms support these findings, as sleep extension has been shown to reduce energy intake [13], while sleep restriction worsens insulin sensitivity [14].

Surprisingly, there is a lack of strong evidence for causality—specifically, increasing sleep duration in short-sleeping patients with T2D without sleep disorders leads to improvements in HbA_1c levels. Recent meta-analyses [15,16] found limited and conflicting evidence on the effect of sleep interventions on glycemic control. Some studies have investigated interventions for patients with T2D with insomnia [17,18]. One study examined an educational intervention for patients with T2D, including those with short sleep, but results from a published abstract [19] are unclear, and no full paper has been published. Two related studies used cognitive behavioral therapy to improve HbA_1c by enhancing sleep quality rather than duration [20,21]. Overall, strong evidence is lacking to support the idea that increasing sleep duration in short-sleeping patients with T2D without sleep disorders leads to improved HbA_1c levels.

Numerous studies have successfully increased sleep duration in various populations. A recent meta-analysis of 42 studies found that extending time in bed (TIB) has the greatest potential to modify sleep, with direct interventions (ie, scheduling a longer period in bed) resulting in an average increase of 1 hour and 23 minutes in sleep duration [22].

Mobile health (mHealth) interventions, such as mobile phone apps supporting self-management, have been shown to be effective in increasing physical activity, reducing body weight [23,24], and improving glycemic control [25,26]. However, to our knowledge, no mHealth apps have been designed specifically to improve sleep duration in short-sleeping patients with T2D without sleep disorders. Personalization of intervention components has been shown to enhance patient motivation and engagement with the app [27]. Several studies have found that tailored interventions are more effective at changing health behaviors than nontailored ones [28,29]. Moreover, interventions based on behavior change theory have been shown to be more effective than those lacking a theoretical foundation [30], with those based on the Theory of Planned Behavior (TPB) being particularly effective [31]. TPB emphasizes intention and ABC as key determinants of behavior, with perceived behavioral control (PBC)—similar or identical to Bandura’s self-efficacy—playing a central role in shaping intention [32,33].

To address the lack of evidence on whether improvements in sleep duration lead to better glycemic control, we developed a pilot study of a personalized mHealth intervention, BedTime, designed to increase sleep duration and assess its impact on glycemic control in short-sleeping patients with T2D. The intervention is based on the TPB framework to promote behavior change toward earlier bedtimes. This trial aims to evaluate the efficacy of the BedTime approach in a 12-week, single-blind, randomized controlled pilot trial. The specific research objectives and hypotheses are as follows:

• Objective 1: To evaluate the efficacy of the BedTime approach in shifting patients to an earlier bedtime, as measured by daily bedtime reporting, and increasing sleep duration, as measured by daily actigraphy.
• We hypothesize that, by the end of the intervention, the intervention group will show statistically significant changes in bedtime and sleep duration compared with the control group.
• Objective 2: To evaluate the efficacy of the BedTime approach in reducing HbA_1c levels.
• We hypothesize that, by the end of the intervention, the intervention group will show a statistically significant reduction in HbA_1c levels compared with the control group.

Methods

Study Overview

This single-blinded, multicenter pilot study is a 2-arm randomized controlled trial designed to evaluate the efficacy of the BedTime mHealth intervention in promoting earlier bedtime behavior and increasing sleep duration among short-sleeping patients with T2D. The trial will be conducted and reported in accordance with CONSORT (Consolidated Standards of Reporting Trails) guidelines. As outlined in Figure 1, the study begins with recruitment (E0), followed by a baseline measurement period (nominally 2 weeks, but at least 7 days), and then randomized assignment to intervention and control groups (E1). Patients in the intervention group will record their daily bedtime, sleep duration, and step count using an actigraphy device, with TIB assessed via sleep diaries. They will receive weekly bedtime goals, daily feedback on their bedtime performance relative to these goals, and guidance to identify personal barriers to an earlier bedtime along with strategies to overcome them. The control group will also record daily bedtime, sleep duration, and step count using an actigraphy device, with TIB assessed via sleep diaries, but they will receive no feedback and will not be assigned goals. During week 10 of the intervention period (E2), the intervention group will complete a TPB questionnaire. The intervention will last for 12 weeks, followed by end-of-intervention assessments of the primary and secondary outcomes (E3). After the intervention period, there will be an observation phase, nominally 12 weeks in length, but its duration will depend on the date of the final visit (E4), which may occur between 10 and 14 weeks after the intervention ends. The study will be conducted over 1 year, from February 8, 2024, to January 22, 2025.

‎

Intervention Design

Some features of the intervention are shared by both the control and intervention groups, while others are unique to the intervention group (Table 1). This mHealth intervention incorporates a physical activity monitoring system (iAide2; Tokai) and a researcher-designed paper diary for patients in both groups. The iAide2 measures activity intensity levels, skin temperature, and pulse rate [34,35] to assess step count and sleep, providing minute-by-minute wake/sleep assessments [36,37]. Additionally, the iAide2 enables timely behavioral feedback via a wirelessly connected Event button and near-real-time data access for researchers through an administrative screen. During the 12-week intervention period, patients in both the control and intervention groups wear the iAide2 device at all times except when bathing and press its Event button at bedtime each day. Patients record their bedtime and risetime daily in the paper diary. For the intervention group, the diary includes additional pages implementing specific aspects of the intervention. Bedtime is defined as the time when a patient begins trying to sleep, and risetime as the time when they stop trying to sleep. Diary entries are used to calculate TIB, the duration between bedtime and risetime. The intervention is designed to encourage an earlier bedtime, with the expectation that this will extend TIB and thereby increase sleep duration (ie, the portion of the TIB interval during which the patient is asleep).

For the intervention group, daily and weekly messaging will deliver a TPB-based intervention targeting PBC and ABC (Figure 2). A researcher or clinical research coordinator (CRC), under the supervision of the researcher, will establish a baseline bedtime and set an initial goal for an earlier bedtime. They will also assist patients in identifying barriers to an earlier bedtime and developing coping strategies. Intervention group patients will receive a daily SMS text message assessment of their performance relative to their target bedtime and will record whether they implemented their coping strategies each day. They will also receive weekly assessments of their progress toward the goal, along with an updated target bedtime, via weekly SMS text messages. Additionally, they will determine coping strategies for the upcoming week.

‎

The intervention is based on the TPB framework (Figure 3) [32,33]. According to TPB, behavior is driven by an individual’s intention to perform the behavior and their ABC, or ability to do so. A behavior is more likely to occur when both intention and actual ability are high. Intention itself is influenced by 3 factors: personal attitude toward the behavior, subjective norms or the perceived expectations of important others, and PBC, which reflects one’s confidence in their ability to perform the behavior. Each of these 3 drivers of intention (attitude toward the behavior, subjective norms, and PBC) is, in turn, influenced by the person’s beliefs about them. The stronger the positive behavioral, normative, and control beliefs, the more likely an individual is to intend to perform the behavior. Actually carrying out the behavior depends on both intention and ABC. As ABC is difficult to measure, researchers generally use PBC as a proxy for ABC in predicting behavior [32,38].

‎

The BedTime intervention is designed to influence PBC by modifying control beliefs. The primary strategy for improving control beliefs—and thus PBC—is to set achievable goals, enabling patients to experience success and build positive self-mastery. We implement progressively challenging but attainable goals using a simple algorithm based on our prior work [39]. This algorithm adjusts the suggested target bedtime each week based on the number of days the user met their goal in the previous week. Adjustments include increases of 5, 10, or 15 minutes and decreases of 5 or 10 minutes, depending on past target bedtime achievement levels.

The secondary mechanism for improving control beliefs involves daily self-reflection on barriers and coping strategies, reinforcing patients’ sense of control over their behavior. These mechanisms align with those used to enhance self-efficacy in social cognitive theory [32]. Beyond influencing PBC, the barrier and coping strategy intervention may also improve ABC by providing useful coping strategies. However, its impact on ABC is expected to be minor, as the suggested strategies may not be entirely novel or groundbreaking for patients. Each week, participants select or write down 1-3 barriers to achieving their target bedtime and propose a coping strategy for each (Table 2). Participants can select coping strategies from a literature-based list of solutions to common barriers to early bedtime [40-60] or describe their own. At the end of each day, they use the diary to report the number of coping strategies they implemented.

The intervention provides daily individualized feedback via an SMS text message (Table 3). These messages, typically sent between 12 PM and 6 PM, inform the patient whether they went to bed earlier or later than their target bedtime the previous night, using an encouraging tone. This timing ensures that participants receive feedback well before their usual bedtime, allowing them sufficient time to adjust their nightly activities accordingly. The exact delivery time of the SMS text messages varies slightly due to the manual nature of the sending process.

At the end of each week, the intervention provides feedback via an SMS text message reporting the number of days the goal was achieved.

Patient Recruitment

Patients will be recruited from 9 medical institutions in Japan (Textbox 1). Attending physicians will conduct recruitment during patients’ regular outpatient consultations. The recruitment period will run from February 8, 2024, to July 30, 2024.

To detect a 0.3% difference in the primary outcome (change in HbA_1c), assuming an SD of 0.41% (the average observed in 2 prior studies [17,61]), and to achieve a 2-sided significance level of .05 with 80% statistical power, the minimum required sample size is 31 patients per group. Considering an estimated dropout rate of approximately 10%, we aim to recruit a total of 70 patients (35 in the intervention group and 35 in the control group).

All participants will receive a comprehensive written and verbal explanation before providing written informed consent. Eligibility screening will be conducted only after consent is obtained. If any findings related to efficacy or safety arise during the trial that may impact patient consent, we will promptly disclose this information to participants and obtain renewed consent.

We will use survey instruments as part of the inclusion and exclusion criteria (Textbox 2). To recruit patients who are both likely to benefit and capable of participating, we will include only those meeting all inclusion criteria (Textbox 3) while ensuring they do not meet any exclusion criteria (Textbox 4), following the methods of a prior study [62]. We anticipate a roughly equal sex distribution and an average participant age in the late 50s. Women undergoing menopausal transition will be included if they meet the study’s criteria.

We will use a Transtheoretical Model (TTM) questionnaire to categorize participants into the contemplation stage (willing to change health behavior within the next 6 months), preparation stage (willing to change health behavior within the next month), or action stage (has already made modifications to health behavior) [63]. In our previous study, patients in the contemplation and preparation stages were more likely to prefer smartphone-based self-management tools than those in the precontemplation stage [64]. Therefore, we will target patients in the contemplation, preparation, or action stage of increasing their sleep duration by 1 hour per day.

We will provisionally register patients who meet the eligibility criteria at E0. At this stage, each participant will receive an actigraph (iAide2) and will be instructed to wear it 24 hours a day, except when bathing, to press the Event button at bedtime, and to record their bedtimes and risetimes in the diary. Patients who fail to press the Event button for 7 or more days or fail to complete the diary for 7 or more days during the run-in period will be excluded to ensure adequate participation. Data collected during the run-in period will be used to confirm eligibility and establish baseline sleep and step counts. We will randomize eligible participants in a 1:1 ratio to either the intervention or control group using covariate-adaptive randomization by minimization to ensure balance across age (<65 and ≥65 years), sex, HbA_1c (<8.5% and ≥8.5%), and institution [72]. Random allocation will be conducted via the Internet Data and Information Center for Medical Research cloud, provided by the University Hospital Medical Information Network, an internet-based central randomization system. A research co-investigator will access the University Hospital Medical Information Network Internet Data and Information Center for Medical Research cloud and input the required baseline information. The system will then securely generate the allocation result immediately before assigning each participant to a group. This process ensures that researchers cannot access or predict allocation results in advance, minimizing potential bias.

Intervention Process

The study (Table 4) consists of 5 events (E0-E4), a 2-week run-in period, a 12-week intervention period, and a 12-week observation period. Initial measurements will be taken at event E0. Questionnaires will be used to assess sleep behavior and gather patient feedback (Textbox 5).

Following randomization, the intervention begins. Before explaining the intervention, the researcher or CRC will administer the TPB Sleep Questionnaire and the Sleep-Related Behavior (SRB) Questionnaire to the intervention group. The researcher or CRC will then provide a diary customized for the intervention group. To establish a baseline bedtime, the researcher or CRC will average bedtimes recorded during the run-in period, considering any exclusions due to unusual circumstances in consultation with the patient. Patients will be informed that their ultimate goal is to shift their bedtime 1 hour earlier than the baseline, with an initial target of going to bed 10 minutes earlier during the first week. The researcher or CRC will help patients identify barriers preventing them from going to bed early and assist them in selecting up to 3 coping strategies tailored to these barriers for the first week. During this process, the researcher or CRC will facilitate discussions using the list in the diary as a reference. Patients may either choose strategies from the list or write their own in the diary. For the control group, the researcher or CRC will provide a diary similar to the one used during the run-in period but covering 12 weeks. Control group patients will be instructed to continue recording their bedtime and risetime as they did during the run-in period. The intervention period lasts 12 weeks, from E1 to E3. We will send reminder SMS text messages to both groups at the end of weeks 1 and 10 to encourage diary use. All equipment, including the diary, will be collected at E3 before the follow-up observation period begins.

For the intervention group, we will provide both daily and weekly feedback. Researchers will monitor the receipt of the Event time, and if it is missing, they will send a reminder asking the patient to report their bedtime via an SMS text message. If no bedtime is reported through either the Event record or SMS text message, the night will be considered a failure to achieve the target bedtime. At E2, 10 weeks after E1, we will send a reminder for participants to complete the TPB Sleep Questionnaire in the diary.

During the study, we will not restrict the use of supplements or drinks that may affect blood glucose and blood pressure. In both the intervention and control groups, T2D treatment—including oral T2D medications, insulin, glucagon-like peptide 1 receptor agonists, and medication dosage—may be adjusted at the discretion of the attending physician. We will collect information on T2D-related medication and medications affecting sleep at E1, E3, and E4.

Discontinuation is defined as a patient expressing a decision to withdraw or failing to record data for a continuous period of 7 or more days. If a participant discontinues during the intervention period, we will attempt to collect E3 measurement items. If discontinuation occurs during the observation period, we will attempt to collect E4 measurement items.

Monitoring, Quality Control, and Data Management

An auditor, independent of the departments involved in the trial, including those responsible for monitoring, will inspect the medical institution and other facilities to ensure the trial is conducted appropriately.

Ethical Approval

This trial will be conducted in compliance with the Declaration of Helsinki, the Pharmaceutical and Medical Device Act, the Ministerial Ordinance on Good Clinical Practice for Medical Devices, and all other relevant guidelines (jRCT1030230650). The study was approved by the Institutional Review Board of the University of Tokyo School of Medicine (approval number 2023336NI). We will obtain written informed consent from all study participants, and all patient data will be pseudonymized. Patients will receive compensation of JPY 30,000 (approximately US $200) at the end of E4.

The participant information materials and informed consent form are available from the corresponding author (KW) upon request. Any amendments to the protocol will be submitted for review and approval by the institutional review board. Study findings will be disseminated through peer-reviewed publications and conference presentations.

Outcome Measures

Primary Outcome

The primary outcome of this study is the between-group difference in the change in HbA_1c from baseline (E0) to either the end of the intervention period at week 12 (E3) or, for patients who discontinue before E3, the point of discontinuation. A reduction of 0.3 percentage points in HbA_1c is considered clinically significant [75,76].

Secondary Outcomes

Secondary outcomes (Table 5) include health measures, bedtime and sleep measurements, other behavior change assessments, and evaluations of actigraph and diary usage. We will also assess safety outcomes, including hypertension requiring medical assistance (evaluated from system records), subjective hypoglycemia, lower back pain, and pain in the lower extremities (tarsus, thighs, knees, calves, shins, ankles, and feet), as well as any other adverse events (all assessed through patient interviews).

Statistical Analysis

In this study, we define 3 analysis populations: the full analysis set (FAS), the per-protocol set (PPS), and the safety analysis set (SAS). The FAS includes all patients for whom the primary outcome was obtained. Following the intention-to-treat principle, we will analyze data based on the assigned group in the FAS analysis. The PPS consists of the FAS population, excluding patients who violated eligibility criteria after randomization, had 7 or more contiguous days of nonuse of the iAide2, or took medications affecting sleep. The SAS includes all patients who used the iAide2 at least once after randomization. In the safety analysis using the SAS, we will analyze data based on the intervention actually used by patients, regardless of allocation. The FAS will serve as the primary analysis population, while the PPS will provide supportive results. All safety analyses will be conducted using the SAS.

Patient characteristics will be presented as mean, SD, minimum, 25th percentile, median, 75th percentile, and maximum for continuous variables, and as frequency and proportion for categorical variables.

In the primary analysis, we will compare the change in HbA_1c from baseline (E0) to week 12 (E3; or the point of discontinuation, if earlier) between groups using analysis of covariance, with baseline HbA_1c (ie, HbA_1c at E0) included as a covariate in the FAS. For participants who discontinue the study, the most recent HbA_1c measurement before E3 will be used.

We will also conduct 2 subgroup analyses: one based on baseline HbA_1c at E0 (<8.5% or ≥8.5%) and the other based on age at initial registration (E0; <65 or ≥65 years). Subgroup analyses of the primary outcome will be performed in each analysis set to assess whether the efficacy of BedTime is consistent across these subgroups.

In the secondary analysis, changes in HbA_1c at week 12 (E3) and week 24 (E4) will be analyzed using the same approach as in the primary analysis. The proportion of patients with HbA_1c below 7% at E3 and E4 (or at the point of discontinuation, if earlier) will be compared between groups using the Fisher exact test. Changes from baseline in step counts, various laboratory test values, and questionnaire scores at E3 and E4 (or at the point of discontinuation, if earlier) will also be analyzed as key endpoints. Changes in T2D medications, categorized as weakened, unchanged, or strengthened, will be compared between groups using the Cochran-Mantel-Haenszel test. The proportion of newly added medications will be compared between groups using the Fisher exact test. A paired (2-sample) t test will be used for within-group comparisons of changes in TPB Questionnaire measurements over weeks 0 (E1), 10 (E2), 12 (E3), and 24 (E4). Changes in TPB Questionnaire and SRB Questionnaire measurements between week 12 (E3) and week 24 (E4) will be compared using a t test. A paired t test will also be used for within-group comparisons from week 0 (E1) to week 12 (E3) for goal achievement rate, goal increment rate, goal reduction rate, average number of coping strategies implemented per day, and the number of identified barriers.

As an exploratory analysis, we will use linear regression to examine the relationship between the average number of coping strategies implemented per day and improvement in HbA_1c within the intervention group. The model will include the average number of coping strategies implemented per day and baseline HbA_1c as explanatory variables, with the change in HbA_1c as the response variable. Changes from baseline to week 12 (E3) and week 24 (E4) (or the point of discontinuation, if earlier) will be assessed.

We will perform a statistical analysis of safety outcomes by calculating the frequency, number of cases, and percentage of individuals experiencing adverse events during the trial, intervention, and observation periods. For the intervention and observation periods, data will be compiled separately for each group, and the percentages will be compared between groups using the Fisher exact test.

Results

Recruitment began on February 8, 2024, and ended on August 29, 2024. The intervention period concluded on December 6, 2024, and the observation period ended on February 26, 2025, with 70 participants completing the observation period (Figure 4). The analysis is currently underway, and we anticipate publishing the results in July 2025. Before commencing analysis, we will conduct a thorough review of all collected data and data entry.

‎

Discussion

We designed this study based on the foundational concept that it is a behavioral change intervention, despite measuring medical outcomes. The intervention is grounded in a specific theory of behavioral change (TPB) to provide a structured framework, ensure clarity on intervention targets, and maintain focus, avoiding a scattershot approach. We identified a specific, actionable target behavior—going to bed earlier—and developed a measurement and feedback system centered on this behavior. This focused approach provides clarity for both researchers and patients. By maintaining this focus, we can determine whether the intervention effectively changes sleep behavior while also evaluating its impact on sleep patterns and, ultimately, health outcomes. Demonstrating that a behavioral intervention improves HbA_1c is not enough, we need to understand the intermediate steps, particularly the behavioral response, to refine the intervention by enhancing effective components and eliminating ineffective ones. This study is designed to generate evidence on these intermediate stages.

Our study utilizes objectively measured data for all primary outcomes and many secondary outcomes. The intervention leverages wireless technology, enabling automatic synchronization of sleep and step count data to the server. These simple, passive features facilitate easy-to-use objective measurements, particularly for older patients.

Participants will be blinded to group assignments to minimize social desirability bias. To achieve this, they will not be informed that the study involves random assignment to 2 groups or that the intervention specifically targets advancing bedtime. Instead, patients will be told that the study examines the relationship between self-management of lifestyle habits (including exercise and sleep) and blood glucose control in individuals with T2D. They will also be informed that various lifestyle factors, including sleep, will be monitored. This approach minimizes the risk of unblinding in the control group by preventing participants from identifying the absence of an intervention. We believe that maintaining blinding offers significant scientific benefits by reducing potential bias and enhancing the validity of the study findings.

We selected the HbA_1c level as our primary outcome, as it is recognized as a global standard for assessing glycemic control and an appropriate indicator of T2D treatment effectiveness [77]. According to the Japan Diabetes Society’s treatment guidelines, patients with T2D should aim to lower their HbA_1c level below 7% to prevent complications [78]. Therefore, in addition to analyzing absolute changes in HbA_1c levels, we examined the proportion of patients achieving an HbA_1c level below 7%. We acknowledge that T2D duration may influence the primary outcome. Although our recruitment process does not explicitly stratify participants by disease duration, this information is collected at registration and will be incorporated into the analysis to account for its potential impact on study outcomes.

In this study, we have chosen a “formidable” control [79], using the same devices as the intervention group. As an initial proof-of-concept study, our goal is to determine whether our specific behavior change intervention is more effective than simply having patients record their bedtime and risetime. Evidence suggests that providing patients with a pedometer can increase their activity levels [80], raising concerns that bedtime and risetime recording in the control group may similarly trigger behavioral changes, potentially making it more challenging to demonstrate the intervention’s superiority statistically. By providing the control group with a diary but withholding the intervention functionality, we have designed the study to offer strong evidence of the intervention’s effectiveness—should it indeed be effective.

Many behavior interventions demonstrate solid short-term gains that later fade [39]. This study includes a 12-week observation period following the intervention to assess whether the benefits persist or diminish. Behavior change maintenance is theorized to result from active, ongoing self-regulation, with habit formation emerging after a period of successfully regulating a new behavior [81]. Our goal is to foster sustainable behavior change, ensuring that the new pattern becomes an enduring habit.

A key aspect of our approach is providing patients with a strong introduction to the mechanics of the intervention. Given their busy lives, we strive to make participation clear and manageable. By offering thorough training and ensuring that all intervention messaging remains simple and focused, we maximize the likelihood that patients will engage with the tools as intended.

Another key aspect of our approach is prioritizing the needs of the patients the intervention is designed to serve. Rather than adopting a one-size-fits-all approach, we tailored the intervention for a specific target population, ensuring it aligns with their needs and preferences. Our previous study [39] highlighted the importance of considering cultural background and patient preferences in intervention design. Additionally, we recognize that socioeconomic background may influence sleep patterns. While our recruitment strategy does not explicitly target specific socioeconomic groups, the participating institutions serve patient populations whose demographics are representative of the general community.

Another key aspect of our approach is applying the intervention only to patients who are already motivated to change their behavior. We use the TTM stage as a filter, an approach that has proven highly effective in previous research [39,63]. Our intervention is specifically designed for highly motivated patients who need support in implementing their desired behavior changes, whereas low-motivation patients require a different intervention focused on enhancing motivation. Notably, a large proportion of patients with T2D in Japan exhibit high motivation, with one study indicating that 92% are in the contemplation-through-action stages of TTM for dietary behavior change [82]. There is a substantial pool of patients who could benefit from this targeted intervention. As a multicenter randomized controlled trial, this study enhances the generalizability of our findings to patients with T2D across Japan. Moreover, our techniques are broadly applicable, and there is strong reason to believe that these methods could be adapted to increase sleep duration for patients with other conditions and in diverse health care settings worldwide.

This study has several limitations. The findings may be specific to the population studied and may not fully generalize to other groups. Our participants are likely to be older individuals of Japanese ethnicity, and differences in lifestyle and T2D pathophysiology between Japanese and other populations may limit broader applicability. Additionally, the study includes only patients capable of using mobile phones, introducing potential biases related to digital literacy. Furthermore, our baseline sleep measurements rely on actigraphy and sleep diaries provided to participants, which may overestimate true prestudy sleep levels. This effect is mitigated by the 2-arm design and the duration of our trial. However, it may lead to an underestimation of increases in sleep duration, potentially resulting in an overestimation of the per-hour-of-sleep impact on health outcomes.

This trial will provide important evidence on the efficacy of a TPB-based mHealth intervention in improving sleep duration and glycemic control in patients with T2D. If the intervention is proven effective and safe, this study could serve as a key step toward integrating mHealth into standard T2D treatment in Japan.