Digital Technology for Diabetes

Abstract

This review article presents three true-life clinical vignettes that illustrate how digital health technology can aid providers caring for patients with diabetes. Specific information that would identify real patients was removed or altered. Each vignette is followed by a discussion of how these methods were used in the care of the patient.

Figure 1. System Configurations for Automated Insulin Delivery (AID).

Panel A shows a system using a “tethered” insulin pump, and Panel B shows a system using a tubeless or “patch” insulin pump. PDM denotes personal diabetes manager.

Figure 2. Trends in Leo’s Glycated Hemoglobin Levels and an Example of Hourly Glucose and Insulin Tracing on AID.

During the 18-month study, Leo’s glycated hemoglobin levels improved, as shown in Panel A. Switching from multiple-daily-injection (MDI) insulin therapy to a sensor-augmented pump (SAP), which provided communication between the glucose sensor and insulin pump but without AID, led to an initial improvement during the first 6 months of the study, which was further enhanced when the two systems were allowed to communicate as an AID system. Panel B shows an example of a single-day tracing while Leo was using AID. Only one meal bolus was given, at approximately 9 p.m. (red bar). However, the algorithm called for six automatic corrective doses (blue bars) to help provide insulin for other meals, including lunch and afternoon snacks. To convert the values for glucose to millimoles per liter, multiply by 0.05551.

Figure 3. Trend in Islet Recovery and Medication Adjustments Informed by Glucose Sensor Data.

This figure shows Eli’s glucose trends during the 8 weeks subsequent to his emergency department (ED) encounter and also includes the medication changes made during the month after his diabetes clinic visit, which took place 4 weeks after he was discharged home from the ED. For consistency, the data are labeled according to weeks since the ED visit, such that weeks 5–8 represent the month immediately after Eli’s visit in the diabetes clinic. Eli’s glucose sensor readings in the 4 weeks after his ED visit (Panel A) showed rapid and progressive reductions in his glucose levels, indicative of islet cell recovery. The glucose management indicator (GMI, calculated from 2 weeks of glucose data; see below) at week 2 was 10.0% and at week 4 was 6.7%. At Eli’s in-person clinic visit, 4 weeks after his discharge from the ED, the 4 weeks of glucose data were reviewed with him, and a plan for how to progressively increase metformin and simultaneously decrease insulin glargine was made. Two weeks after the clinic visit (after weeks 5–6) (Panel B), the glucose sensor data were reviewed remotely by the physician with Eli. Together, they agreed on a plan to add empagliflozin while continuing to decrease the insulin. The results of this plan are shown in the subsequent 2 weeks (weeks 7–8), during which time empagliflozin was increased to the full dose and insulin was discontinued. The medication adjustments made in the month after his clinic visit are summarized (Panel C). These adjustments correspond to the continuous glucose monitor (CGM) data shown in Panel B from the same period, which allowed Eli and his physician to be comfortable with the swift dose adjustments. The GMI approximates glycated hemoglobin values on the basis of the mean sensor glucose using at least 2 weeks of CGM data (GMI %=3.31+0.02392×[mean glucose level in milligrams per deciliter]; GMI millimoles per mole=12.71+4.70587×[mean glucose level in millimoles per liter]). CV denotes coefficient of variation.

Figure 4. Trend in Daily Hospital Glucose Levels during Inpatient Diabetes Management.

When Gia was receiving MDI insulin therapy during her first 5 days in the hospital, she had fluctuating glucose levels that frequently exceeded 250 mg per deciliter. AID therapy was initiated on day 5, which led to improved glucose control. MDI insulin therapy was reinstituted on day 15 in order to facilitate Gia’s discharge to a rehabilitation facility, and her glycemic control again showed greater fluctuation with higher glucose levels.