UC San Diego
Continuous Glucose Monitoring and Diabetes Management Behaviors: A Secondary Data Analysis from the REPLACE-BG Trial
- Author(s): Crawford, Margaret
- Advisor(s): Pierce, John P
- Strong, David R
- et al.
Background: Continuous glucose monitors (CGM) are becoming a norm for type 1 diabetes management and provide the opportunity to describe hypoglycemic and hyperglycemic events experienced by people with type 1 diabetes (T1Ds). This dissertation had four objectives: 1) derive scales from the Hypoglycemia Fear Survey- Behavior (HFS-B) scale that represent unique constructs of hypoglycemia- related behavior, 2) describe the frequency and severity of CGM- measured hypoglycemic events, and assess how these relate to levels of hypoglycemia- related behaviors, 3) develop severity categories for CGM- measured hyperglycemic events and describe how the severity categories relate to A1C, and 4) collate insulin pump and CGM data to describe how participants at different levels of A1C use insulin boluses to manage their hyperglycemic events.
Methods: Four analyses were conducted using CGM, insulin pump, demographic, and HFS-B data collected over 26 weeks from 216 T1Ds in the REPLACE-BG trial. The first was a psychometric analysis of the HFS-B. The second identified and measured hypoglycemic events and assessed how these events related to hypoglycemia-related behaviors. The third identified hyperglycemic events, categorized them by severity, and assessed how measures of hyperglycemic event severity predict A1C, the standard measure of glucose control. The fourth identified hyperglycemic events in which insulin boluses were administered and assessed the association of proactive insulin bolusing with the occurrence of severe hyperglycemic events.
Results: Three scales were derived from the HFS-B, labelled hypoglycemia avoidance, reaction, and prevention behavior. Higher levels of hypoglycemia prevention behavior were associated with a lower percentage and shorter duration of moderate hypoglycemic events. Four categories of hyperglycemic event severity were developed, and those in the best glucose control (A1C < 7.1) had 1) a larger percent occurrence of non-severe hyperglycemic events and 2) a smaller percentage of time in the most severe event category. Finally, those in the best glucose control were more likely to practice proactive bolusing to prevent severe hyperglycemic events.
Conclusion. This analysis demonstrated the importance of CGM data in its continuous form. CGM identifies behaviors associated with prevention of both hypo- and hyperglycemia, which are preferentially performed by those with the best glucose control.