Real-World Clinical and Economic Outcomes of Time in Range in Association With Diabetic Retinopathy Among Patients With Type I Diabetes
Author(s)
Nai-Chia Chen, MS1, Eric J. Gutierrez, MPH1, Viral N. Shah, MD2, Robert Brett McQueen, BA, MA, PhD1.
1University of Colorado Skaggs School of Pharmacy and Pharmaceutical Science, Aurora, CO, USA, 2Indiana University School of Medicine, Indianapolis, IN, USA.
1University of Colorado Skaggs School of Pharmacy and Pharmaceutical Science, Aurora, CO, USA, 2Indiana University School of Medicine, Indianapolis, IN, USA.
OBJECTIVES: Assessing continuous glucose monitoring (CGM) metrics in relation to diabetes-related complications is of growing interest in regulatory approval clinical studies. Most studies on CGM metrics report snapshots of CGM use during routine clinical care. However, measures of relative risk rates over time (e.g., hazard ratios) are crucial for linking longitudinal use of CGM to complications, particularly given unequal follow-up and censoring. We aimed to estimate longitudinal changes in time in range (TIR) on progression to diabetic retinopathy among patients with type 1 diabetes (T1D).
METHODS: We used linked data from a retrospective case-cohort study of T1D patients with diabetic retinopathy follow-up diagnoses from the University of Colorado Barbara Davis Center for Diabetes. Longitudinal data with an average of 7-year follow-up in CGM metrics and electronic health records were extracted and analyzed. TIR was defined as blood glucose control within 70 - 180 mg/dL. A multilevel mixed-effects parametric survival models was applied to obtain hazard ratios and parameters for the economic model with consideration of correlated CGM metrics. A Markov model was constructed to project lifetime retinopathy progression.
RESULTS: We analyzed 161 T1D patients with 70 diagnosed with diabetic retinopathy during follow-up. After adjusting for the fixed-effects (age at the last eye exam, baseline hemoglobin, BMI, diagnosis of proteinuria, antihypertensive drug use) and the random-effect of baseline A1c level, the Weibull model suggested a modest decrease in retinopathy risk for a 10% increase in TIR (HR = 0.93, 95% CI: 0.86, 1.00). Scale parameters for different TIR ratios were estimated with a fixed shape parameter from the Weibull model for time-varying probabilities. The blindness cases per 1000 over life increased by lowering TIR (at 70%: 22 cases, at 60%: 24 cases, at 50%: 22 cases).
CONCLUSIONS: Effects from this study can inform diabetes simulation models for predicting future treatment outcomes using CGM metrics.
METHODS: We used linked data from a retrospective case-cohort study of T1D patients with diabetic retinopathy follow-up diagnoses from the University of Colorado Barbara Davis Center for Diabetes. Longitudinal data with an average of 7-year follow-up in CGM metrics and electronic health records were extracted and analyzed. TIR was defined as blood glucose control within 70 - 180 mg/dL. A multilevel mixed-effects parametric survival models was applied to obtain hazard ratios and parameters for the economic model with consideration of correlated CGM metrics. A Markov model was constructed to project lifetime retinopathy progression.
RESULTS: We analyzed 161 T1D patients with 70 diagnosed with diabetic retinopathy during follow-up. After adjusting for the fixed-effects (age at the last eye exam, baseline hemoglobin, BMI, diagnosis of proteinuria, antihypertensive drug use) and the random-effect of baseline A1c level, the Weibull model suggested a modest decrease in retinopathy risk for a 10% increase in TIR (HR = 0.93, 95% CI: 0.86, 1.00). Scale parameters for different TIR ratios were estimated with a fixed shape parameter from the Weibull model for time-varying probabilities. The blindness cases per 1000 over life increased by lowering TIR (at 70%: 22 cases, at 60%: 24 cases, at 50%: 22 cases).
CONCLUSIONS: Effects from this study can inform diabetes simulation models for predicting future treatment outcomes using CGM metrics.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
CO201
Topic
Clinical Outcomes, Economic Evaluation, Methodological & Statistical Research
Topic Subcategory
Relating Intermediate to Long-term Outcomes
Disease
Diabetes/Endocrine/Metabolic Disorders (including obesity)