BEHAVE-HTN: subanalysis of a randomized trial evaluating timing adherence with a multifaceted mHealth intervention and financial incentives
A Vischer, J Strebel, A Raja Beharelle, M Mayr, V Nemtsova, J Zahnd, G Wuerzner, G Ehret, M Stiefel, E Rexhaj, T Meister, T BurkardAbstract
Background/Introduction
Medication non-adherence substantially contributes to premature mortality and increased healthcare utilization, particularly in hypertension, with limited sustained success of existing interventions. While most studies focus on overall dose-taking adherence, less is known about adherence to the prescribed dosing time.
Purpose
Purpose of this subanalysis was to determine whether a multifaceted mobile health application (mApp), with or without financial incentives, improves electronically monitored timing adherence in hypertensive patients and to assess whether baseline self-reported adherence modifies intervention effects.
Methods
BEHAVE-HTN was a multicenter, block-randomized, controlled, parallel, open-label trial with 1:1:1 allocation to mApp plus financial incentives (Group A), mApp alone (Group B), or standard care (control). The mApp combined reminders, contingent rewards, positive reinforcement, gamification, and patient-centered educational content. The active intervention period lasted 90 days (end of active period, EP), followed by a 90-day extended follow-up without mApp or incentives (eFU). This predefined subanalysis focuses on timing adherence, defined as electronically monitored medication intake within a predefined two-hour dosing window. Adherence was assessed using electronic medication monitoring devices and, additionally, by self-reported adherence using the Hill-Bone Scale. Electronically monitored adherence was analyzed both per 90-day period (EP and eFU) and per 30-day intervals (M1–M6). The primary endpoint of this subanalysis was timing adherence at EP comparing Group A with control. Secondary endpoints included comparisons between Groups A and B, changes between M1 and M3, and adherence trajectories over time, and assessment of baseline adherence as an effect modifier.
Results
Among 176 participants completing the EP, median timing adherence was numerically higher in Group A (81%) compared with control (68%) and Group B (66%), without reaching statistical significance (p=0.086 and p=0.148). The median change between M3 and M1 was 0% in Group A, −2% in the control group, and −3% in Group B (p=0.219 and p=0.009, respectively). Overall adherence declined over time across groups (−1.6 percentage points/month, p=0.005), without significant differences in overall adherence trajectories. At eFU, median timing adherence remained numerically higher in Group A (72%) compared with control (55%) and was significantly higher compared with Group B (59%) (p=0.064 and p=0.020). Baseline self-reported adherence did not predict electronically monitored outcomes (interaction p=0.793 and p=0.321).
Conclusions
The mHealth intervention did not significantly improve electronically monitored timing adherence compared with standard care. A signal toward greater adherence stability in the incentivized group suggests that financial incentives may support sustained adherence behavior.