A Randomized Clinical Trial Assessing Continuous Glucose Monitoring (CGM) Use With Standardized Education With or Without a Family Behavioral Intervention Compared With Fingerstick Blood Glucose Monitoring in Very Young Children With Type 1 Diabetes

The study compared three approaches to managing type 1 diabetes in children aged 2 to under 8 years old:

**CGM + FBI:** Continuous glucose monitoring (Dexcom G5) with standardized training plus a family behavioral intervention (FBI) — five 30-minute sessions teaching relaxation, problem-solving, and communication strategies to overcome barriers to CGM use.

**Standard-CGM:** CGM with standardized training only (four 30-minute sessions on how to use the device, interpret data, and dose insulin).

**BGM (control):** Traditional fingerstick blood glucose monitoring (BGM) — no CGM, just standard care with a blood glucose meter and test strips.

The primary outcome was **percent time in range (TIR)** — the percentage of time glucose levels stayed between 70 and 180 mg/dL. Secondary outcomes included time spent in hypoglycemia (<70 mg/dL, <60 mg/dL, <54 mg/dL), time in hyperglycemia (>180 mg/dL, >250 mg/dL, >300 mg/dL), HbA1c (a 3-month average of blood sugar), and several parent-reported quality-of-life measures (diabetes burden, fear of hypoglycemia, well-being, family impact, and technology satisfaction).

**143 children** aged 2 to under 8 years (mean age ~5.5 years) with type 1 diabetes

Recruited from **14 pediatric endocrinology clinics** across the United States

All had type 1 diabetes for at least 3 months (later changed to 3 months from 6 months to speed recruitment)

HbA1c between 7.0% and <10.0% (53–86 mmol/mol) at screening

Total daily insulin requirement ≥0.3 units/kg/day

No use of real-time CGM in the 30 days before enrollment

52% were male, 82% white, mean diabetes duration ~2.5 years

57% used an insulin pump, 43% used multiple daily injections

Mean baseline HbA1c was ~8.2% (66 mmol/mol) — indicating suboptimal control (only 24% of young children in the T1D Exchange Registry achieve HbA1c <7.5%)

**CGM data:** Dexcom G5 Mobile Continuous Glucose Monitor (real-time for CGM groups; masked/ blinded for BGM group). For the BGM group, masked CGM was worn for 1-week periods at 6, 13, 19, and 26 weeks — the device recorded glucose but did not display it, so it couldn't influence behavior.

**HbA1c:** Measured at randomization, 13 weeks, and 26 weeks by central lab (University of Minnesota, Tosoh A1c 2.2 Plus Glycohemoglobin Analyzer).

**Parent-reported questionnaires** (completed at randomization and 26 weeks):

- **WHO-5 Well-Being Index** (5 items, 0–100 scale, higher = better well-being)

- **Problem Areas in Diabetes Survey – Parent Revised (PAID-PR)** (18 items, measures diabetes-specific burden; higher = more burden)

- **Diabetes Family Impact Scale (DFIS)** (measures impact of diabetes on family functioning)

- **Hypoglycemia Fear Survey for Parents of Young Children (HFS-PYC)** (measures fear of low blood sugar)

- **Diabetes Technology Questionnaire (DTQ)** (measures satisfaction with diabetes technology)

**Safety outcomes:** Severe hypoglycemia (requiring assistance from another person due to altered consciousness), diabetic ketoacidosis (DKA), device-related adverse events, and serious adverse events.

**Design:** Multicenter, 6-month, randomized controlled trial (RCT) with three parallel groups (1:1:1 allocation).

**Randomisation:** Block design stratified by clinical center and screening HbA1c (<8.5% vs. ≥8.5%). This means the randomization was done in blocks within each center, ensuring roughly equal numbers of high and low HbA1c participants in each group at each site.

**Blinding:** This was an **open-label** trial — participants, families, and clinicians knew which group they were in. This is a major limitation because CGM use is inherently visible (the device is worn on the body, alarms go off, data is displayed). The BGM group wore a masked CGM for data collection, but they didn't get real-time feedback. The FBI sessions were also obviously different from standard training. There was no placebo or sham intervention for the BGM group.

**Duration:** 6 months total. The prerandomization phase was 14–21 days of masked CGM to establish baseline glucose metrics. After randomization, visits occurred at 6, 13, 19, and 26 weeks, with phone calls at 10, 16, and 22 weeks. The two CGM groups had additional in-clinic training visits at 1 and 3 weeks.

**Washout:** Not applicable — this was a parallel-group design, not a crossover.

**Statistical approach:** Primary analysis used linear mixed-effects regression models adjusted for baseline TIR, baseline HbA1c (stratification factor), and clinical center. The primary analysis compared all three pairwise groups. Because there were three comparisons, the significance threshold was set to p < 0.0167 (Bonferroni correction for multiple comparisons). Analysis was intention-to-treat (all randomized participants included, regardless of adherence). Two sensitivity analyses were performed: one adjusting for any baseline imbalances, and one per-protocol (excluding participants who didn't meet minimum CGM data requirements).

**What this design can prove:** An RCT with intention-to-treat analysis can establish causal relationships — if the groups are well-balanced at baseline, any differences at follow-up can be attributed to the interventions. The stratification by center and HbA1c helps control for these variables. The 6-month duration is long enough to see sustained effects (not just a novelty effect).

**What this design cannot prove:** Open-label designs are vulnerable to **performance bias** (families in the CGM groups might change other behaviors because they know they're being monitored) and **detection bias** (outcome assessment is not blinded). The lack of blinding is particularly problematic for parent-reported outcomes (quality of life, fear of hypoglycemia) — parents who received the extra FBI sessions might report better well-being simply because they received more attention and support (a Hawthorne effect). The study also cannot tell us about long-term effects beyond 6 months, or about effects in older children or adults.

**Major methodological weaknesses:**

**Open-label design** — no blinding for participants or clinicians

**No sham control** for the FBI — the extra attention alone could explain the parent-reported benefits

**BGM group had limited CGM data** — only four 1-week periods of masked CGM, compared to continuous data for CGM groups. This could introduce measurement error for the BGM group's glycemic outcomes.

**Recruitment criteria were relaxed mid-study** (diabetes duration reduced from 6 to 3 months, insulin regimen stability reduced from 3 to 1 month) — this could introduce heterogeneity

**22 of 165 enrolled participants (13%) didn't make it to randomization** — mostly due to run-in failure or withdrawal, which could introduce selection bias (the most motivated families may have been more likely to complete the run-in)

**Primary outcome — Time in Range (TIR, 70–180 mg/dL):**

No significant differences between any groups:

- CGM+FBI vs. BGM: +3.2% (95% CI: -0.5% to +7.0%) — not statistically significant

- Standard-CGM vs. BGM: +0.5% (95% CI: -2.6% to +3.6%) — not significant

- CGM+FBI vs. Standard-CGM: +2.7% (95% CI: -0.6% to +6.1%) — not significant

Mean TIR at baseline was ~42% across all groups; at follow-up it was ~45–48% — a modest improvement that didn't differ by group.

**Secondary glycemic outcomes — Hypoglycemia (the big win):**

**Time <70 mg/dL** was reduced by roughly half in both CGM groups:

- CGM+FBI: from 5.2% to 2.6% (about 75 minutes/day to 37 minutes/day)

- Standard-CGM: from 5.8% to 2.5% (about 84 minutes/day to 36 minutes/day)

- BGM: from 5.4% to 5.8% (about 78 minutes/day to 84 minutes/day) — essentially unchanged

- CGM+FBI vs. BGM: p < 0.001

- Standard-CGM vs. BGM: p < 0.001

**Time <54 mg/dL** (dangerously low):

- CGM+FBI: from 1.1% to 0.4%

- Standard-CGM: from 1.4% to 0.4%

- BGM: from 1.0% to 1.1%

- Both CGM groups significantly better than BGM (p < 0.001)

**Severe hypoglycemic events** (requiring assistance):

- CGM+FBI: 0 events

- Standard-CGM: 1 event

- BGM: 5 events

- This difference was not statistically significant (p = 0.06), but the pattern is striking.

**Secondary glycemic outcomes — Hyperglycemia and HbA1c:**

No significant differences in time >180 mg/dL, >250 mg/dL, or >300 mg/dL between groups

No significant differences in HbA1c at 26 weeks (all groups ~8.0–8.2%)

No significant differences in mean glucose or glucose variability (coefficient of variation)

**Parent-reported outcomes:**

**Diabetes burden (PAID-PR):** CGM+FBI parents reported significantly greater reductions in diabetes-specific burden compared to Standard-CGM (p = 0.008) and BGM (p = 0.02)

**Fear of hypoglycemia (HFS-PYC):** CGM+FBI parents reported significantly greater reductions in fear compared to Standard-CGM (p = 0.04) and BGM (p = 0.002)

**Well-being (WHO-5):** No significant differences between groups

**Family impact (DFIS):** No significant differences

**Technology satisfaction (DTQ):** Both CGM groups reported high satisfaction, with no significant difference between CGM+FBI and Standard-CGM

**CGM adherence:**

~90% of participants in both CGM groups used CGM ≥6 days/week at 6 months — excellent adherence

**Hypoglycemia reduction:** Time spent with glucose <70 mg/dL was reduced from about 75–84 minutes per day to about 36–37 minutes per day — a reduction of roughly 45–48 minutes per day. This is clinically meaningful: it means a child spending over an hour per day with low blood sugar now spends about half an hour. For a parent, this translates to fewer alarms, fewer middle-of-the-night lows, and less constant vigilance.

**Severe hypoglycemia:** The BGM group had 5 severe events requiring assistance; the CGM+FBI group had zero. While not statistically significant (p = 0.06), the absolute risk reduction is 5 events in 48 children over 6 months — a number needed to treat (NNT) of roughly 10 to prevent one severe hypo.

**Parental burden:** The PAID-PR (diabetes burden) and HFS-PYC (fear of hypoglycemia) improvements in the CGM+FBI group were statistically significant but the paper doesn't report the raw score changes in a way that's easily translated to "how much better." However, the p-values (0.008 and 0.002) indicate these were robust effects, not flukes.

**Time in range:** The 3.2% improvement in TIR for CGM+FBI vs. BGM translates to about 46 minutes more per day in range — but this was not statistically significant, meaning we can't be confident it's a real effect (it could be due to chance).

**Acknowledged by authors:**

Open-label design — participants and clinicians knew treatment assignment

BGM group had limited CGM data (only four 1-week masked periods) compared to continuous data for CGM groups

The FBI was delivered by research assistants, not licensed mental health professionals — may have reduced potency

The study was not powered to detect differences in severe hypoglycemia (rare events)

Recruitment criteria were relaxed mid-study, potentially introducing heterogeneity

**Critical reader observations:**

**No blinding for parent-reported outcomes** — the FBI group received more attention (5 extra sessions), so the improved well-being could be a placebo/attention effect. A proper control would have been an equal number of "sham" educational sessions.

**Industry funding:** The study was funded by the Leona M. and Harry B. Helmsley Charitable Trust, but Dexcom provided the CGM devices. Device company involvement is common but can introduce subtle bias (e.g., in how training is delivered, which outcomes are emphasized).

**Generalizability:** 82% white, recruited from specialized pediatric endocrinology clinics. Results may not apply to more diverse populations, families with lower health literacy, or those without smartphone access (needed for CGM sharing features).

**Short duration:** 6 months is good but doesn't tell us about long-term adherence or effects on complications years down the line.

**No cost-effectiveness analysis:** CGM is expensive (~$300–$500/month for sensors). The study didn't address whether the benefits justify the cost.

**The FBI was bundled with CGM** — we can't separate the effect of the behavioral intervention from the effect of CGM itself. The comparison of CGM+FBI vs. Standard-CGM showed no difference in glycemic outcomes, only in parent-reported outcomes, which are the most susceptible to bias.

For someone running their own n=1 experiment (e.g., an adult with type 1 diabetes considering CGM, or a parent of a young child with type 1):

### What to test

**Intervention:** Continuous glucose monitoring (CGM) — specifically a real-time CGM with alerts (like Dexcom G6 or similar). If you want to test the "family behavioral intervention" component, that would involve structured sessions on problem-solving, communication about diabetes, and relaxation techniques to reduce fear and burden.

**Dose:** Wear the CGM continuously (replace sensor every 7–10 days depending on model). For the behavioral component, try 5 structured sessions over 6 months (roughly monthly) focused on specific barriers you identify.

### Minimum meaningful duration

**For glycemic outcomes:** At least 3–6 months. The SENCE trial saw effects on hypoglycemia emerge over 6 months. Shorter periods may capture novelty effects or temporary behavior changes.

**For quality-of-life outcomes:** 3 months may be sufficient to see changes in fear of hypoglycemia and diabetes burden, but 6 months is better to ensure the effect isn't just from the initial excitement of new technology.

### What to measure (specific metrics)

**Primary metric:** Time in range (TIR) 70–180 mg/dL — aim for >70% (the general target for most adults with type 1 diabetes; targets may differ for children)

**Key secondary metric:** Time below range <