JMIR Publications

ABSTRACT

Background:

The rising popularity of apps that sync with continuous glucose monitors (CGMs) reflects growing interest in on-demand, personalized care. These platforms combine real-time glucose biofeedback with self-monitored behaviours to optimize metabolic health among individuals with and without diabetes. However, little is known about user characteristics, engagement patterns, or factors that predict sustained use of CGM-integrated digital health applications in real-world settings.

Objective:

This study aimed to describe user demographics, CGM usage patterns, and food logging behaviours among users of the Vively app, and to identify predictors of sustained engagement with CGM wear and food tracking.

Methods:

We conducted a retrospective observational study of Vively app users between August 2021 and February 2025. Vively is a commercial digital health application that integrates with Abbott FreeStyle Libre CGMs to deliver personalized nutrition guidance. Users with at least one day of CGM wear were included. Primary outcomes were CGM wear duration (total days) and food logging engagement (total days and binary ever-logged). User characteristics, CGM usage patterns, and self-monitored behaviours were analysed descriptively. Predictors of engagement were identified using negative binomial regression for CGM wear and hurdle negative binomial models for food logging, adjusting for age, sex, BMI, baseline glucose, and device connectivity.

Results:

The analytical sample included 7,647 users (62.5% female, mean age 44.4 years, mean BMI 27.8 kg/m²). Users wore CGMs for a median of 15 days (IQR 14-30), with 42.7% completing one full wear period (13-15 days) and 30.3% completing two or more wear periods (≥28 days). Most users (91.7%) logged food at least once, with a median of 47 food entries over 12 days. Food logging remained high during CGM wear (90.8% daily participation) but declined sharply after sensor removal (1.5% daily participation). In multivariate models, higher baseline glucose predicted longer CGM wear (IRR 1.15, 95% CI 1.13-1.17) but fewer food logging days (IRR 0.96, 95% CI 0.94-0.98). Connected device syncing was the strongest predictor of engagement for both CGM wear (IRR 1.32, 95% CI 1.28-1.37) and food logging (IRR 1.45, 95% CI 1.39-1.51). Older age and female sex were associated with higher engagement in both behaviours.

Conclusions:

This large-scale analysis reveals distinct patterns of engagement with CGM-integrated digital health applications. While users demonstrate high engagement during active CGM wear, participation in self-monitoring behaviours declines dramatically after sensor removal. The divergent effects of baseline glucose levels—predicting longer CGM wear but reduced food logging—suggest different motivational drivers for passive monitoring versus active behaviour tracking. These findings have important implications for designing sustainable digital health interventions that maintain user engagement beyond periods of biological feedback.