JMIR Publications

ABSTRACT

Background:

Glucagon-like peptide-1 receptor agonists (GLP-1 RAs), including semaglutide and tirzepatide, are highly effective for weight management; however, long-term outcomes depend on sustained behavioral engagement. Digital health tools may enhance these behavioral components, but evidence regarding their real-world benefit alongside pharmacotherapy remains limited.

Objective:

To evaluate whether augmenting a direct-to-consumer GLP-1 RA program with a behavioral support smartphone app improves weight loss outcomes compared with medication alone, and to identify which application features are associated with weight loss over time.

Methods:

A retrospective observational cohort study was conducted using data from 14,599 adults enrolled in an Australian direct-to-consumer weight-loss program, of whom 6,753 met inclusion criteria after data cleaning. Participants received semaglutide or tirzepatide, with optional access to a smartphone app and clinician support. Logistic regression models examined achievement of ≥5%, ≥10%, and ≥15% weight-loss thresholds drawn from established clinical guidelines and trial conventions at 60, 120, and 180 days. Linear mixed-effects models with linear and quadratic time terms assessed which application features impacted percentage weight loss trajectories, adjusting for medication type, gender, ethnicity, and baseline BMI.

Results:

Participants using the program plus medication were more likely to achieve clinically significant weight loss, with odds ratios ranging from 1.46 to 2.57 for ≥5% weight loss across timepoints (P< .001), 1.73 to 2.15 for ≥10% weight loss at later timepoints (P< .001), and 2.13 for ≥15% weight loss at 180 days (P=.001). Participants using the program plus medication were more likely to achieve clinically significant weight loss, with odds ratios ranging from 1.46 to 2.57 for ≥5% weight loss across timepoints (P< .001), 1.73 to 2.15 for ≥10% weight loss at later timepoints (P< .001), and 2.13 for ≥15% weight loss at 180 days (P=.001). Mixed-effects modelling demonstrated that progress logging was associated with greater weight loss over time (β = 0.28, 95% CI 0.05 to 0.52, P = .019), with no evidence of a quadratic effect. Recipe viewing was not associated with linear change over time, although the linear interaction approached significance (β = −0.18, 95% CI −0.37 to 0.02, P = .078), but demonstrated a small positive quadratic effect (β₂ = 0.05, 95% CI 0.01 to 0.08, P = .004), suggesting a delayed or increasing benefit over time. Goal setting and calorie target setting were not associated with changes in weight-loss trajectories.

Conclusions:

Augmenting GLP-1 RA treatment with a behavioral smartphone app was associated with higher likelihood of achieving clinically significant weight loss, although differences in overall weight-loss trajectories were modest. Medication type was the strongest determinant of weight loss over time, with tirzepatide producing greater early reductions. Among app features, progress logging emerged as the primary behavioral predictor of improved outcomes. These findings support the value of integrating digital tools with pharmacotherapy and highlight the importance of self-monitoring in weight management.