Quantifying Input Behaviors That Influence Clinical Outcomes in Diabetes and Other Chronic Illnesses

With new digital and wearable technology, it is now possible to quantify not only overall health outcomes but also the contribution of individual factors such as varying patient behaviors that contribute toward those outcomes. In chronic illnesses such as type 1 diabetes, for example, infrequent measurements like HbA1c are insufficient to avoid frustration for patients asked to perform dozens of self-management tasks on a daily basis with no clear indicator as to whether those tasks are having the desired impact on outcomes or which tasks are the most impactful. In the absence of such feedback, both the required frequency and perceived burden of such self-management behaviors can influence clinical outcomes. ¹ Some clinicians have begun assessing time in target glycemic range (TIR) instead of solely assessing A1c, ² which is beneficial for those who can access continuous glucose monitoring, but there is still significant opportunity to help patients identify the specific impact of discrete behavior changes on clinical outcomes.

A recent research article in JAMA Network Open by Lee et al ³ illustrates that it is now possible to use electronic health record (EHR) assessment of diabetes self-management behaviors to quantify specific groups of behaviors’ associations with glycemic outcomes. Lee et al hypothesize that quantifying the effect of different types and groups of behaviors could play a role in helping improve disparities in glycemic outcomes. Translated into clinical practice, such methods could enable health care providers to show patients when their recent behavior changes successfully improve glycemic outcomes, which may motivate willingness to proceed with additional changes in self-management. Such improved glycemic outcomes may also result in improved quality of life. ⁴

The work of Lee et al focused on clinicians documenting diabetes self-management behaviors within an EHR, resulting in quantifiable associations between behaviors and outcomes. There are other opportunities to quantify and demonstrate association between diabetes self-management behaviors and clinical outcomes. New research has shown that open-source automated insulin delivery (AID) systems are as safe and effective as commercially developed AID systems for people with insulin-requiring diabetes. ⁵ With these open-source AID systems, lead users have shown it is possible to quantify the impact of individual input behaviors (such as meal announcement, carb counting, meal boluses, and consuming medium to higher carbohydrate meals) and their cumulative effect on preferred outcomes (~80% TIR 70-180 mg/dL and little to no hypoglycemia). ⁶ Additional research should be done on open-source and commercial AID systems alike to further quantify the effects of discrete behaviors of diabetes self-management while using this technology.

By quantifying and separating the impact of different behaviors on clinical outcomes for diabetes and many other chronic diseases, with a variety of diabetes self-management technologies, clinicians and patients can and should work together to prioritize the behaviors and technology choices that best balance patient preferences and feasibility with improved clinical outcomes and increased quality of life.