Classification and regression tree analysis to understand theory-aligned barriers to glucose self-monitoring...

Glucose self-monitoring is critical for type 2 diabetes management among patients using insulin. Combining behavior change theory with classification and regression tree (CART) is a novel data-driven approach to identifying barriers to self-monitoring. Baseline data from non-CGM-using patients in The GluCoCare Study were used to describe barriers to self-monitoring and use of data to inform diabetes self-care (N = 360). Patterns of barriers that predicted self-monitoring or use of self-monitoring data were also identified. Results showed barriers to self-monitoring were present in each of three Behavior Change Wheel categories: capability (e.g. procedural knowledge), opportunity (e.g. social support), and motivation (e.g. guilt). One example of a CART-identified pattern that predicted high self-monitoring was low guilt, high belief about importance of self-monitoring, low skin irritation, and high procedural knowledge. Future studies should test whether promoting these determinant patterns by addressing specific barriers leads to more self-monitoring and better outcomes.