Real-Time CGM Linked to Slower eGFR Decline in Diabetes
Key Takeaways
- Among insulin-treated adults with type 1 or type 2 diabetes and CKD or risk for CKD, RT-CGM initiation was associated with a significantly slower eGFR decline than no CGM (adjusted difference 1.74 mL/min/1.73 m² over three years).
- The ACR comparison found no statistically significant between-group difference in change.
- HbA1c fell significantly more with RT-CGM than with no CGM.
Investigators conducting a propensity-matched retrospective study compared initiation of real-time continuous glucose monitoring (RT-CGM) with no CGM use. They used de-identified electronic medical records supplied by Truveta. Eligible adults had type 1 or type 2 diabetes, received insulin, and had a baseline estimated glomerular filtration rate (eGFR) of 15–89 mL/min/1.73 m². Adults in the RT-CGM group started a Dexcom G-series device between 2018 and 2023 and were required to have at least one refill within six months. Comparison patients had index dates set at their first outpatient visit with an eGFR lab order. Follow-up values were taken within six months of the three-year mark. Separate matched cohorts were used to assess changes from baseline to three-year follow-up in eGFR, albumin-to-creatinine ratio (ACR), and HbA1c; the participants contributing to each comparison were not assumed to be identical.
The eGFR comparison included 518 matched participants per group. From baseline to follow-up, adults who initiated RT-CGM had a significantly slower eGFR decline than adults who did not use CGM. The mean decline was 7.72 mL/min/1.73 m² with RT-CGM versus 9.46 without, an adjusted difference of 1.74 mL/min/1.73 m² over about three years (standard error 0.83; p = 0.04). This was a comparison of observed filtration trajectories, not a direct test of whether monitoring prevented kidney disease.
The ACR comparison included 138 matched participants per group and found no significant between-group difference in change. In the separate HbA1c comparison, 461 matched participants per group contributed to a finding of a significantly greater decrease with RT-CGM than with no CGM. HbA1c fell 0.33 percentage points with RT-CGM versus 0.13 without, an adjusted difference of 0.20 percentage points (p = 0.01).
Because the comparison was retrospective, propensity matching cannot establish that RT-CGM prevents CKD progression. Most participants did not have advanced kidney disease: in the matched groups, about 81% were at eGFR stage G2 (60–89) and about 1% at stage G4, and more than half had type 1 diabetes. The authors listed as limitations that acute kidney injury was not excluded, adherence was not assessed beyond one refill, baseline ACR was not matched in the eGFR and HbA1c models, and socioeconomic status and care quality were not captured. The absence of a significant ACR difference does not mirror the eGFR pattern, and the greater HbA1c decrease cannot be assumed to explain that pattern. These outcomes should not be treated as interchangeable evidence of a kidney benefit. It's also important to note that Dexcom, which makes the RT-CGM device studied, funded the work.
The researchers reported an association between RT-CGM initiation and slower kidney-function decline in this insulin-treated population, and stated that these findings support RT-CGM initiation in people with diabetes and CKD or at risk for CKD; in the Discussion they said it “could be considered.” They also stated that a causal relationship cannot be determined, and called for future prospective studies to assess whether CGM metrics correlate with delayed CKD progression. Whether monitoring itself accounts for the difference remains unresolved.
Clinician Questions
Did everyone eligible for the RT-CGM kidney-function comparison have established CKD?
No. Eligible insulin-treated adults with type 1 or type 2 diabetes were described as having CKD or being at risk for CKD; eligibility did not establish that every participant already had CKD.
Was the RT-CGM association with eGFR decline different in type 1 versus type 2 diabetes?
The study did not report results separately by diabetes type. Both types were eligible, and diabetes type was used in matching (about 57–59% of each matched group had type 1), so a difference between types cannot be determined.