JMIR Publications

ABSTRACT

Background:

Weight regain following discontinuation of anti-obesity medications is a recognised clinical concern, yet real-world data on post-discontinuation weight trajectories following structured digital weight loss programs remain scarce.

Objective:

To characterise weight trajectories at 6, 9, and 12 months following discontinuation of a medicated digital weight loss program, and to identify baseline and in-program behavioural predictors of post-exit weight change.

Methods:

This retrospective cohort study analysed data from 8,408 adults who completed the Juniper medicated digital weight loss program in Australia (N=1254) or the United Kingdom (N=7154) and subsequently discontinued. Participants received semaglutide or tirzepatide and voluntarily recorded post-exit weights via the Juniper application, to which all users retain access following discontinuation. Post-exit weight change from program exit was categorised into five bands (>5% additional loss to >5% regain). Linear regression models identified predictors of weight change at 6 months (primary; N=6181) and 12 months (primary; N=1889), with 9-month analyses as supplementary (N=3349). Two sequential model blocks tested baseline predictors alone, then baseline plus in-program behavioural predictors.

Results:

Among the 6736 participants with 6-month follow-up weight data, mean post-exit weight change was −0.4% (SD 8.2); 28.7% achieved >5% additional weight loss, 14.0% lost 2–5%, 19.4% maintained within 2% of exit weight, 13.7% regained 2–5%, and 24.2% regained >5%. This heterogeneous distribution persisted at 9 months (N=3788; mean −0.3%, SD 10.1; >5% additional loss: 32.6%; >5% regain: 29.2%) and 12 months (N=2,165; mean −0.1%, SD 11.6; >5% additional loss: 32.7%; >5% regain: 33.9%). In linear regression, higher baseline BMI predicted continued post-exit weight loss at both primary timepoints (6m: β=−0.33, SE=0.02; 12m: β=−0.54, SE=0.04; both P<0.001), while UK participants showed significantly greater post-exit weight gain than Australian participants (6m: β=1.97, SE=0.28; 12m: β=4.40, SE=0.82; both P<0.001). Greater in-program weight loss was associated with more post-exit weight regain relative to exit weight in a dose-response pattern (≥20% vs <5% in-program loss: 6m β=3.93, SE=0.48; 12m β=8.99, SE=1.27; both P<0.001). Higher health coaching and digital support engagement was independently associated with less post-exit regain (25–50% of program weeks vs 0%: 6m β=−0.68, SE=0.31, P=0.03; 12m β=−2.66, SE=0.82, P<0.001). In-program behavioural predictors improved model fit beyond baseline predictors alone at 6 months (R²: 0.090 to 0.116) and 12 months (R²: 0.136 to 0.185).

Conclusions:

Post-discontinuation weight trajectories following a medicated digital weight loss program appear to be highly variable, with approximately one-third of participants continuing to lose weight and nearly one-quarter experiencing substantial regain. Paradoxically, greater in-program weight loss predicted more post-exit regain relative to exit weight, consistent with the known rebound effect following Semaglutide and Tirzepatide withdrawal. Digital health coaching engagement was an independently modifiable predictor of better post-exit outcomes, supporting the development of targeted off-ramp support to reduce post-discontinuation regain.