The hidden variable in a “neutral” upgrade: why parameter re-tuning and perceived reliability matter after G6 → G7 in Control-IQ

Dear Editor,

Nattero-Chávez et al. prospectively followed 92 adults with type 1 diabetes using t:slim X2 with Control-IQ who transitioned from Dexcom G6 to G7. Overall, TIR70–180 did not change materially over 3 months, while mean glucose/(GMI) Glucose Management Indicator decreased slightly, and patient-reported distress and quality of life improved; the most notable glycemic gains occurred in participants starting with time in range (TIR) <70%. ¹ The safety reassurance is valuable, but the two reports’ details merit targeted analysis to make the results more actionable at the point of care.

The subgroup improvement below the consensus TIR target ² may reflect therapy re-optimization rather than a pure sensor effect. Nearly one-third of participants (27/92) reported Control-IQ setting changes during follow-up, yet adjustments were recorded only as a final-visit yes/no variable and missingness used last-observation-carried-forward. ¹ Because real-world Control-IQ performance is sensitive to parameter tuning, ³ the authors could re-estimate the longitudinal models stratified by whether settings changed and, in parallel, fit a mixed-effects model across all study visits to minimize imputation bias. Reporting the proportion who newly achieve TIR ⩾70% and the proportion with a clinically meaningful TIR gain (e.g., ⩾5% absolute) would further translate the subgroup signal. Finally, because auto-correction boluses increased from 14.1% to 16.4%, ¹ testing whether this shift is confined to those with parameter changes would help distinguish sensor-driven effects from re-tuning, in line with published G7 performance data. ⁴

Equally important for counseling, 29.3% reported G7-related issues—most commonly inadequate accuracy (22%) and pump connectivity loss (22%)—yet none discontinued use. ¹ Linking these (PREMs) Patient-Reported Experience Measures to objective logs (CGM) Continuous Glucose Monitoring gaps, time out of automated mode) and reporting effect sizes for distress/quality-of-life instruments^5,6 would clarify whether perceived problems correspond to measurable workflow burden and whether (PROM) Patient-Reported Outcome Measures improvements differ in those reporting issues.

Finally, internal inconsistencies (baseline TIR <70% noted as n = 19 in text but n = 20 in Table 1, and an apparent inversion in glycated hemoglobin comparisons by baseline-TIR stratum) should be corrected because they directly affect subgroup interpretation. ¹

Table 1.

Reporting inconsistencies in the source study relevant to subgroup interpretation.

Item	Location in source article	Reported text/value	Why correction matters
Subgroup size at baseline TIR70-180 <70%	Results text vs Table 1	Results text states “individuals with baseline TIR70-180 <70% (n = 19),” whereas Table 1 presents the subgroup header as “<70% (n = 20).”	The subgroup denominator should be internally consistent because it directly affects interpretation of the subgroup signal.
Baseline HbA1c by TIR stratum	Results text, Table 1, abstract, and Table 3	The Results text says the <70% group had higher HbA1c, but the printed pair is “6.3% ± 0.5% vs 7.3% ± 0.9%,” which reverses the stated direction. By contrast, Table 1 lists HbA1c as 7.3% ± 0.9% for TIR < 70% and 6.3% ± 0.5% for TIR ⩾70%; the abstract and Table 3 are concordant with that ordering.	This affects the directionality of the baseline comparison and, therefore, the clinical interpretation of the lower-TIR subgroup.

HbA1c, glycated hemoglobin; TIR, time in range.