1907-LB: Estimating Optimal CGM Wear Time for Individuals with Type 2 Diabetes Not on Insulin

Objective: Continuous glucose monitoring (CGM) consensus guidelines, developed using large CGM datasets in type 1 and insulin-treated type 2 diabetes, recommend 14 days of wear time with at least 70% (~10 days) complete data. We sought to examine whether fewer days could reliably capture glycemic variability measures such as time in range (TIR) in individuals with type 2 diabetes (T2D) not on insulin.

Methods: The NIH-funded DiaTeleMed study is a 6-month randomized controlled trial comparing three different dietary strategies in adults diagnosed with T2D managed using diet alone or diet and metformin. We analyzed blinded Freestyle Libre Pro CGM data collected for up to 14 days before the intervention started. For participants with at least 10 complete days of data, we computed their time in 70-140 mg/dL (TIR₁₄₀) and 70-180 mg/dL (TIR₁₈₀) range using a cumulative approach starting from the first day and successively adding each day of data until their final day. We then identified the minimum number of days it took for the cumulative TIR metrics to stabilize within a clinically meaningful ±2% of their final values.

Results: We analyzed 80 participants (age 60±12 years, 63% female, 47% non-White, 14% Hispanic/Latino, BMI 30±6 kg/m², HbA1c 6.7±0.6%, 66% on metformin) who had at least 10 days of complete CGM data. Participants’ cumulative TIR stabilized within ±2% of final values in 70±20% of days worn for TIR₁₄₀ and in 50±23% of days worn for TIR₁₈₀. Extrapolating these percentages to 14 days of wear, effectively 10±3 days and 7±3 days would be sufficient to estimate TIR₁₄₀ and TIR₁₈₀ respectively in this population.

Conclusion: Fewer than the 14 recommended days may be sufficient to estimate important CGM-derived metrics such as the TIR in individuals with T2D managed with diet alone or diet plus metformin. CGMs worn for less than the mandated 14-day period due to issues such as device fall-off, discomfort, or malfunction may still provide reliable glycemic variability information.