Effective intervention strategies are required to reduce the immense burden of cardiometabolic disease. Standard behavioral interventions to reduce cardiometabolic disease risk are typically based on population averages [1,2]; however, variability in response to dietary interventions suggests the need for personalized approaches [3]. Precision approaches aim to tailor interventions to an individual’s unique characteristics and contexts [4,5]. For example, just-in-time adaptive interventions aim to facilitate behavior change by delivering an intervention at times and in contexts when individuals are more receptive to behavior change [6]. Identifying opportunities for change requires an understanding of the time-varying factors that influence behaviors.

The potential to measure time-varying factors has increased with the availability of wearable sensors. The use of continuous glucose monitors (CGMs) has resulted in the identification of glucose variability, that is, intraday fluctuations in blood glucose, as a risk factor for the development of cardiometabolic disease [7]. Greater glucose variability predicts increases in glycated hemoglobin (HbA_1c) in both diabetic and nondiabetic populations [8,9]. Among persons with type 2 diabetes who maintain good glycemic control (HbA_1c≤7.0%), greater glucose variability was an independent risk factor for coronary artery disease during a 10-year follow-up period [10]. Though evidence in nondiabetic populations is sparse, mechanistically, it has been proposed that glucose variability increases cardiovascular disease risk through increased oxidative stress and inflammation [11], an effect that has been observed among individuals with normal insulin function [12].

While several factors influence glucose levels (eg, stress, physical activity, etc), glucose variability is primarily driven by increases in postprandial glucose [13]. Controlled laboratory studies have demonstrated that several dietary factors impact postprandial glucose, such as meal composition [14,15], meal timing [16], and meal frequency [17]. In real-world settings, eating behavior, and thus glucose variability, is influenced by several individual-level factors. One example of an individual factor that influences eating behavior is affect [18-21]. The relationship between eating behavior and affect appears to be bidirectional as fatigue, stress, and negative mood are associated with increased cravings and food consumption [19-22], while variations in glucose in response to eating may relate to affective and physical feeling states [23-26]. However, research in this area has primarily relied on dietary assessment recall methodologies, which may be prone to bias and recall error [27]. Pairing data from CGMs with data on affect using ecological momentary assessment (EMA), which involves repeated sampling of affect in real time and in an individual’s natural environment, improves the validity of responses [28] and dramatically decreases the time between measurement of mood and glucose.

Combining CGM and EMA data to investigate the relationship between affective states and glucose variability presents an opportunity to support the development of precision interventions to prevent chronic disease by identifying situations that are likely to lead to large glucose increases (ie, an opportunity for change). A recent review [29] of mood and glucose variability in type 1 and 2 diabetes identified only 1 study that used CGMs and EMA to evaluate mood and glucose [25] and called for more research evaluating the relationships between mood or affect and glucose variability. To date, only a small (n=15) pilot study has combined EMA responses with CGM data to evaluate its association with affective and physical feeling states in nondiabetic samples [18]. Thus, the aims of this study are to (1) evaluate the feasibility of conducting a study using CGMs and EMA and (2) examine the bidirectional relationship between affect and glucose in adults using CGM and EMA data.

This is a 14-day observational study. Enrolled participants (N=100) will wear a CGM and receive EMA prompts to assess affect 6 times per day over the 14-day period (detailed below).

Recruitment started in October 2022. As of December 12, 2022, a total of 39 participants have been enrolled. We expect to enroll 100 participants by a target completion date of March 2023.

This project will generate data to elucidate the relationship between affective states and glucose variability, with an aim of ultimately informing intervention development to reduce glucose variability. We hypothesize that a bidirectional relationship exists by which greater negative affect will predict greater glucose variability, while greater glucose variability will predict greater negative affect.

Technological innovations hold promise for improving approaches to understand the relationship between affective states and glucose variability in real-world settings. CGMs allow for real-time collection of data related to postprandial response, while EMA allows for real-time collection of mood states. Pairing these approaches will facilitate a more valid evaluation of the relationships between affective states and eating behavior, as well as provide opportunities for real-time precision nutrition interventions that consider individuals’ physiological and cognitive states. Precision intervention approaches, such as just-in-time interventions, aim to intervene during moments of opportunity [6]. Compliance data, along with the postparticipation questionnaire, will inform changes to study procedures to address the barriers experienced by participants. Evaluation of the relationship between glucose level and affect will aim to identify situations in which glucose excursions occur (opportunity) and can inform interventions to be delivered during those opportunities (ie, meditation or physical activity to improve affect and dietary suggestions that minimize glucose excursions).