Issues and Solutions for Time-In-Range Analyses Based on Inpatient Continuous Glucose Monitoring Data

Continuous glucose monitoring (CGM) has been increasingly used in the hospital for the care of patients with hyperglycemia and diabetes. “Time in range” (TIR) has been spotlighted as a pivotal metric derived from CGM data to assess glycemic control. However, a prevailing data issue that is often ignored in TIR analysis is that the use of CGM in the hospital often has a shorter sampling duration due to the patient’s early discharge. As shown by our simulation results, ignoring this issue can lead to a considerably biased evaluation of TIR. We have developed rigorous statistical procedures that properly account for the limitations of the hospital CGM use, and confer valid estimation and inference of mean TIR. We also established the asymptotic properties of the proposed estimators. Results from our numerical studies demonstrate good finite sample performance of the proposed method as well as its advantages over the existing approach.