Continuous Glucose Monitoring in Adolescents With Obesity: Monitoring of Glucose Profiles, Glycemic Excursions, and Adherence to Time Restricted Eating Programs

This study tested whether adolescents with obesity (without diabetes) would consistently wear a continuous glucose monitor (CGM) for 12 weeks, and whether an 8-hour time-restricted eating (TRE) schedule would improve glucose variability compared to a standard 12-hour eating window.

The three intervention arms were:

**Arm 1 (Control):** 12-hour eating window + blinded CGM (participants could not see their glucose data)

**Arm 2 (TRE + blinded CGM):** 8-hour eating window with 16-hour fasting, 5 days per week + blinded CGM

**Arm 3 (TRE + unblinded CGM):** 8-hour eating window with 16-hour fasting, 5 days per week + real-time glucose feedback visible to the participant

The primary outcomes were:

1. Feasibility of CGM wear (percentage of days worn, satisfaction questionnaire scores)

2. Baseline glycemic profiles in adolescents with obesity

3. Change in glycemic variability between TRE groups vs control over 12 weeks

Secondary outcomes included: mean glucose, fasting glucose, post-prandial glucose excursions, glucose management indicator (GMI), and various measures of within-day and between-day glucose variability.

**Sample size:** 50 adolescents enrolled; 43 had complete CGM and dietary recall data available for analysis

**Age:** Mean 16.4 years (SD 1.3 years, range 14–18)

**Sex:** 64% female

**Race/ethnicity:** 64% Hispanic

**Socioeconomic status:** 74% on public insurance

**Health status:** All had obesity (BMI ≥ 95th percentile for age and sex), none had diabetes

**Setting:** Outpatient, recruited from a pediatric weight management clinic at Children's Hospital Los Angeles

**Exclusions:** Prior diagnosis of Prader-Willi syndrome, brain tumor, hypothalamic obesity, developmental or intellectual disability, eating disorder, prior bariatric surgery, current use of weight-affecting medications (antipsychotics, sedatives, hypnotics, insulin, off-label obesity drugs), current participation in weight-related psychotherapy or other interventional studies

**Continuous Glucose Monitor:** Dexcom G6, which measures interstitial glucose every 5 minutes, worn continuously for 13 weeks (1 week run-in + 12 weeks intervention)

**CGM-derived metrics** calculated using EasyGV software (University of Oxford):

- Mean glucose (mg/dL)

- Standard deviation (SD) of glucose

- Coefficient of variation (CV)

- Mean amplitude of glycemic excursion (MAGE)

- Area under the curve (AUC) for glucose

- Glucose Management Indicator (GMI)

- J-index

- Continuous overlapping net glycemic action (CONGA)

- Mean of daily differences (MODD)

- Low and high blood glucose indices

- Daytime and nighttime mean glucose and SD

**Feasibility:** Percent wear time recorded by Dexcom Clarity platform; satisfaction measured via adapted CGM satisfaction scale administered at weeks 4, 8, and 12

**Dietary adherence:** Self-reported eating windows reported weekly to study team; 24-hour dietary recalls used to cross-reference meal timing with CGM tracings

**Weight:** Height and weight measured at home via Bluetooth scale with research coordinator monitoring via HIPAA-compliant video at weeks 1, 4, 8, and 12; BMI z-score and percent of 95th percentile calculated using CDC growth charts

**Study design:** This was a three-arm, parallel-group, randomized controlled pilot trial (ClinicalTrials.gov NCT03500835). Participants were randomized 1:1:1 using blocked randomization to one of three groups: 12-hour eating control, 8-hour TRE with blinded CGM, or 8-hour TRE with unblinded CGM.

**Randomization:** Blocked randomization was used, though the authors do not specify block sizes or stratification factors. All participants completed a 7–10 day run-in period wearing a blinded CGM before randomization to capture baseline glycemic profiles.

**Blinding:** The control group and the blinded TRE group wore CGMs that did not display glucose data to the participant. The unblinded TRE group could see their real-time glucose readings. The study team was not blinded to group assignment (this is an open-label trial design). There is no mention of blinding of outcome assessors.

**Duration:** The intervention lasted 12 weeks. CGM wear was continuous for 13 weeks total (1 week run-in + 12 weeks intervention). Participants wore each sensor for 10 days before replacement.

**Statistical approach:** Repeat measures analysis was conducted to assess change in glycemic variability over time between groups. The authors used EasyGV software to compute summary measures of glycemic data. The paper reports results as mean ± SD or mean ± SE where applicable, with p-values for between-group comparisons.

**What this design can and cannot prove:**

*What it CAN prove:*

Feasibility and acceptability of CGM wear in adolescents with obesity over 12 weeks

Descriptive baseline glycemic profiles in this population

Preliminary estimates of effect size for TRE on glycemic variability (for powering future trials)

Whether real-time glucose feedback changes adherence or outcomes compared to blinded CGM

*What it CANNOT prove:*

Efficacy of TRE for improving glucose regulation (the study was a pilot trial, not powered for efficacy—the null result may reflect insufficient sample size rather than true lack of effect)

Long-term effects beyond 12 weeks

Whether TRE works differently in adolescents compared to adults (no direct comparison group)

Causal mechanisms—the design cannot distinguish whether any observed changes were due to TRE itself, calorie reduction, weight loss, or behavioral changes

Generalizability to other adolescent populations (predominantly Hispanic, female, low-income sample from a single center)

**Major methodological weaknesses:**

1. **Pilot trial, not powered for efficacy** – The null finding for glycemic variability may be a Type II error (false negative). The authors explicitly state this was a feasibility study.

2. **No blinding of participants or study team** – Open-label design introduces potential for differential behavior, expectation effects, and measurement bias

3. **Self-reported eating windows** – Adherence to TRE was based on participant report, not objective verification (though CGM tracings were used to cross-reference meal timing with glucose excursions)

4. **No calorie restriction** – TRE was implemented without calorie counting, so it's unclear whether participants actually reduced total energy intake

5. **Home-based weight measurements** – Height and weight were self-measured at home under remote supervision, introducing measurement error

6. **Short duration for metabolic outcomes** – 12 weeks may be insufficient to see changes in glucose variability, especially in adolescents without diabetes who already have relatively normal glucose regulation

7. **No washout period** – Not applicable to a parallel-group design, but there was no crossover element to control for individual differences

**Feasibility of CGM wear:**

96.4% of prescribed days had CGM data available across all participants

Participants reported no negative impacts on daily functioning from wearing the CGM

Satisfaction questionnaire responses were positive (specific scores not reported in the abstract)

**Baseline glycemic profiles (pre-randomization, all participants):**

Mean glucose: Not explicitly reported in the abstract (full text would contain these values)

Glycemic variability measures (SD, MAGE, AUC) were computed but specific baseline values are not in the abstract

**Effect of TRE on glycemic variability (primary analysis):**

**No significant change** in standard deviation of glucose between groups over the study period

**No significant change** in mean amplitude of glycemic excursion (MAGE) between groups

**No significant change** in glucose area under the curve (AUC) between groups

**No significant change** in any other glycemic variability measure (CV, J-index, CONGA, MODD, low/high blood glucose indices)

Specific p-values and effect sizes are not reported in the abstract; the paper states "no significant change" without providing exact statistics

**Comparison between blinded and unblinded CGM groups:**

No significant differences in glycemic outcomes between the two TRE groups (blinded vs unblinded CGM), suggesting that real-time glucose feedback did not alter eating behavior enough to change glucose profiles

**Adherence to TRE:**

Participants in TRE groups reported adhering to their 8-hour eating window on the prescribed 5 days per week

CGM tracings were used to confirm meal timing by identifying post-prandial glucose excursions, though specific concordance rates are not in the abstract

The study found **no statistically significant effect** of TRE on any measure of glucose variability. This means the effect size was either zero or too small to detect with 43 participants.

To put this in context: In adult TRE studies that have found positive results, improvements in glucose variability (measured by SD or MAGE) are typically in the range of 5–15% reduction. This study would have needed to detect a similar or larger effect to reach statistical significance with their sample size. The fact that they found nothing suggests that either:

TRE does not meaningfully change glucose variability in adolescents with obesity (true null)

The effect is smaller than in adults and requires a larger sample to detect

The 8-hour window (with 5 days/week adherence) is not a strong enough intervention in this population

Adolescents without diabetes already have relatively tight glucose regulation, leaving little room for improvement

The 96.4% CGM wear rate is a strong positive finding—this is comparable to or better than adherence rates in adult CGM studies and suggests that adolescents are willing and able to wear these devices consistently.

**Acknowledged by authors:**

This was a pilot feasibility trial, not powered to detect efficacy

Further research is needed to investigate how TRE impacts glycemic variability in adolescents

Need to explore if timing of the eating window (early vs late TRE) affects outcomes

**Critical reader observations:**

**No effect sizes or confidence intervals reported** – The abstract states "no significant change" without providing the actual numbers, making it impossible to assess whether the null result reflects a truly zero effect or simply an underpowered study

**Small sample size** – 43 participants with complete data across three groups means roughly 14 per group, which is far too few to detect small-to-moderate effects

**No correction for multiple comparisons** – The study computed dozens of glycemic variability metrics but did not adjust for multiple testing, increasing the risk of false positives (though none were found)

**No objective adherence verification** – Self-reported eating windows are unreliable; adolescents may overreport adherence. CGM tracings can show post-prandial glucose spikes but cannot confirm the exact start/end of eating windows

**No calorie or macronutrient data** – Without knowing whether TRE participants actually ate less or changed what they ate, it's impossible to interpret the null glucose result

**No control for physical activity** – Exercise affects glucose regulation independently of diet; activity levels were not measured or controlled

**Single-center, homogeneous sample** – 64% Hispanic, 74% public insurance limits generalizability to other populations

**No long-term follow-up** – 12 weeks may be too short for metabolic adaptations to occur

**Industry funding not disclosed** – The paper does not explicitly state whether Dexcom provided devices or funding, though this is common in CGM research

**No mention of menstrual cycle phase** – In female participants, menstrual cycle phase significantly affects glucose regulation and was not controlled for

**No washout or crossover** – A crossover design would have been more powerful but was not used

For someone running their own n=1 experiment:

### What to test

**Intervention:** Time-restricted eating with an 8-hour eating window (e.g., eat only between 10:00 AM and 6:00 PM) for 5–7 days per week

**Dose:** 16-hour daily fast, with the eating window self-selected but consistent day-to-day

**Comparison:** Your usual eating pattern (which for most people is a 12–14 hour eating window)

### Minimum meaningful duration

**At least 4 weeks** to see any changes in glucose variability (this study found nothing at 12 weeks, but individual responses may vary)

**Ideally 8–12 weeks** to allow for metabolic adaptation

**CGM wear:** Minimum 7–10 days per condition to get reliable glycemic variability metrics (shorter periods are too noisy)

### What to measure (specific metrics)

**Primary metric:** Standard deviation of glucose (SD) – a measure of how much your glucose swings throughout the day. Lower = better.

**Secondary metrics:**

- Mean glucose (mg/dL) – your average over 24 hours

- Coefficient of variation (CV) – SD divided by mean, expressed as percentage. <36% is considered stable

- Mean amplitude of glycemic excursion (MAGE) – captures the size of glucose spikes after meals

- Time in range (70–140 mg/dL for non-diabetic individuals) – percentage of readings in this healthy range

- Fasting glucose (first reading upon waking)

- Post-prandial glucose peak and duration (how high and how long after meals)

**Adherence metric:** Actual eating window start/end times logged daily (use a time-stamped app or photo log)

### Key confounds to control for

**Total calorie intake** – TRE may cause you to eat less (which improves glucose) or eat more during the window (which worsens it). Log calories for at least the first 2 weeks of each condition.

**Carbohydrate composition** – If you eat more carbs during your eating window, glucose variability will increase regardless of timing. Keep macronutrient ratios consistent between conditions.

**Sleep duration and quality** – Poor sleep worsens glucose regulation. Track sleep (time in bed, wake time, quality) daily.

**Physical activity** – Exercise improves glucose disposal. Keep exercise type, duration, and intensity consistent between conditions.

**Menstrual cycle phase** – If female, glucose regulation worsens in the luteal phase (second half of cycle). Compare TRE vs control within the same phase, or track across at least two full cycles.

**Stress** – Acute and chronic stress raise glucose via cortisol. Log daily stress on a 1–10 scale.

**Hydration and caffeine** – Dehydration and caffeine both affect glucose. Keep intake consistent.

**Alcohol** – Alcohol initially lowers glucose then can cause rebound hyperglycemia. Avoid alcohol during measurement periods or standardize intake.

### What a positive result would look like

**SD decreases by 5–10 mg/dL** (e.g., from 25 mg/dL to 22 mg/dL) during the TRE period compared to your usual eating pattern

**CV drops below 30%** (from a baseline of 35% or higher)

**MAGE decreases by 10–15 mg/dL** (smaller post-meal spikes)

**Fasting glucose drops by 3–5 mg/dL** (e.g., from 95 to 90 mg/dL)

**Time in range (70–140 mg/dL) increases** from ~85% to ~92% or higher

**Post-prandial glucose peaks are lower and return to baseline faster** (e.g., peak at 140 mg/dL instead of 160 mg/dL, and back to baseline within 90 minutes instead of 120 minutes)

**Important caveat:** This study found NO effect of TRE on glucose variability in adolescents. If you are an adult, the evidence is more mixed—some adult studies show improvements, others don't. Your individual response will depend on your baseline glucose regulation, your actual adherence to the eating window, and