Background: Continuous glucose monitoring (CGM) is increasing in insulin-treated type 2 diabetes (T2D). Yet, standardized CGM-based insulin titration is lacking. This study evaluated the feasibility of algorithmic CGM-based titration compared with self-monitoring blood glucose (SMBG) titration. Methods: We conducted a 16-week, two-site, randomized controlled trial in adults with T2D (glycated hemoglobin 7%–9%) using degludec and adjunctive noninsulin agents, without rapid-acting insulin. Participants were assigned (2:1) to weekly algorithmic CGM-based dose changes with open CGM (EXP) or weekly SMBG-based titration with blinded CGM (CTR). Both groups received dose notifications via phone. The primary endpoint was the change in CGM-measured time in range (TIR, 70–180 mg/dL) from baseline to week 16, tested for noninferiority (−5%-percentage points %-pt). The trial is registered at ClinicalTrials.gov: NCT06111508. Results: A total of 30 participants were randomized. Mean (standard deviation) TIR increased from 54.1% (22.5%) to 75.3% (19.3%) in EXP and from 50.2% (22.1%) to 55.3% (22.7%) in CTR. Mean change was +20.3%-pt versus +8.3%-pt, yielding an estimated treatment difference (EXP − CTR) of +14.6%-pt; one-sided 95% confidence interval (CI) lower bound was +4.0%-pt, exceeding the noninferiority margin ( P < 0.005). Exploratory superiority analysis showed two-sided 95% CI: 1.3–27.8 ( P = 0.03). CGM-measured hypoglycemia (<70 mg/dL) was low (median interquartile range: 0.34% 0.09–0.90 vs. 0.00% 0.00–0.41), and level 2 episodes (SMBG <54 mg/dL) were rare (1.1 vs. 2.2 patient-year of exposure). No severe hypoglycemia or serious adverse events occurred. Conclusions: Using CGM and receiving algorithmic CGM-based titrations were feasible, safe, and had favorable overall glycemic metrics. Long-term impact should be confirmed in broader populations.
Fathi et al. (Fri,) studied this question.