Continuous Glucose Monitoring and Dietary Self-Management in Adults in Namwon

This community-based, individually randomized, open-label, wait-list controlled study evaluates the incremental effect of adding a continuous glucose monitoring (CGM) device as a personal glucose biofeedback tool to a common digital health and education program during the first 6 weeks. The randomized comparison is device provision versus no device provision during this period; standardized education and behavioral goal setting are common to both groups. Approximately 3,000 residents are expected to undergo screening to identify 860 eligible participants who provide additional written consent for randomization. Screening participants who do not enter the randomized study are not part of its control group. Their separately consented optional week 12 follow-up will be analyzed descriptively and separately from the randomized comparison. For these nonrandomized participants, the optional follow-up window is Days 77-91 after the first screening visit.

The actual randomization date is Day 0. Randomization is preferably completed at the first screening visit after eligibility verification, required consent, and baseline assessments, but may occur later without a separate maximum screening-to-randomization interval. Before allocation, willingness to participate, treatment, safety status, and sensor-use changes are rechecked. The week 6 assessment window is Days 35-49 and the week 12 window is Days 77-91 after randomization. Sensor placement delays do not reset Day 0. Screening, consent, baseline assessment, randomization, and sensor placement dates are recorded separately. Scheduling must permit completion within the approved study period and the planned field-assessment end in September 2027.

If screening and randomization occur on different dates, the four dietary self-management items are collected again before allocation using the same wording, response options, and recall periods. The mean of the first three items collected before allocation on the randomization date is the primary baseline score. Same-day assessments are not duplicated. Other baseline questionnaires, AI Diet, and physical measurements use the pre-allocation values from the first screening visit, retaining the measurement dates and screening-to-randomization interval. Post-allocation values are not substituted for baseline. The interval is described by group and included in an additional adjusted sensitivity analysis.

Allocation will be 1:1 using a web-based system operated outside the field research team, with computer-generated variable blocks and three strata: documented type 2 diabetes diagnosis or treatment; no documented diagnosis or treatment with fasting glucose at least 126 mg/dL; and no documented diagnosis or treatment with fasting glucose from 100 through 125 mg/dL. The primary analyst will finalize the analysis plan and code using group labels A and B before the group code is revealed. Participants and field researchers will know the assigned timing.

Beginning at the first screening visit, both groups receive the PASTA application, the commercially available AI Diet dietary assessment and feedback provided through Kakao Healthcare's PASTA application, CancerFree prevention information, digital literacy support, and basic health counseling. Both groups also receive the same standardized education on glucose patterns, relationships of meals, activity and sleep with glucose responses, and safe behavioral goal setting during the first 6 weeks, independently of sensor provision. Education and goal setting are not deferred in the control group. The added intervention is access to the CGM device and linked personal glucose feedback. Sensor placement, connection, and device safety instructions are provided when each group actually starts using its sensor.

The primary comparison is the week 6 dietary self-management score, adjusted for its baseline value, diabetes stratum, and initial enrollment location. The study-specific score is based on the mean of the first three dietary self-management items adapted with revised wording and response guidance from the Korean version of the Summary of Diabetes Self-Care Activities. It is not presented as an independently validated instrument. The fourth item, concerning carbohydrate distribution, is reserved for a sensitivity analysis.

After the week 6 outcome assessment and data lock, the wait-list control group is offered the same sensor so that all eligible randomized participants have an opportunity to receive the device. The first 6 weeks constitute the primary controlled intervention period; each sensor is worn for approximately 10 days, not continuously for 6 weeks. Week 12 assessments are exploratory follow-up after both groups have been offered the sensor. The study is not primarily designed to establish superiority of early versus delayed provision. No on-site blood sampling is planned. Optional psychosocial assessments and planned administrative data linkage, conditional on later separate consent and approvals, are exploratory and do not replace the primary randomized comparison.

AI Diet uses the existing 10 dietary categories and their commercially provided risk-stage table without modification. For the prespecified exploratory analysis, the four category-assigned stages for obesity, diabetes, hypertension, and hypertriglyceridemia are summed. Each stage ranges from 1 (lowest) to 4 (highest); the sums across the 10 existing categories range from 4 to 13. A higher sum indicates a higher combined category-assigned risk stage. This sum is not an individual probability of disease. Category transitions are also examined.

With separate consent, PASTA records actually generated during the 24-month study period may be received for exploratory analyses beyond the 12-week field follow-up. The current National Health Insurance Service linkage-plan and follow-up-contact consent does not authorize actual linkage or collection of resident registration numbers. Actual linkage requires a confirmed lawful processing route, responsible entities, approvals, and later separate explanation and consent. Pseudonymized analysis is planned in the approved National Health Insurance Service environment, with only approved analytical outputs released; individual insurance records are not downloaded or stored by the research team. One linkage is planned after all week 12 visits, within the study period ending in September 2028. Up to five years of records before the first screening visit are used to assess prior history. Incident outcomes are observed from the actual randomization date to the actual linkage date for randomized participants, and from the first screening visit for the separately consented nonrandomized follow-up cohort, subject to record availability. The overall study end is not treated as each participant's observed follow-up end.

Protocol version 8 received ethics approval. The version 9 amendment will undergo ethics review before implementation of the amended study plan.