Background: Hypertension (HTN) remote monitoring programs (HRMP) have been associated with significant systolic and diastolic blood pressure (BP) reduction; however, limited information is available on program components that drive clinical improvement. The objective of this study was to quantify mobile nudge effectiveness that deliver behavior change content on BP control for HRMP participants.

Methodology: A retrospective study was conducted on HRMP participants newly enrolled from January 1, 2021, to December 31, 2022, focusing on mobile nudge interventions categorized into content domains of BP checks, physical activity, coaching, medications, journey engagement, nutrition, sleep, stress, and webinars. The first two months (M1, M2) following initial use of the cellular-enabled BP monitor was analyzed separately to evaluate BP control (BP <130/80 mmHg) over the subsequent three months utilizing the Double Machine Learning (DML) method, which estimates heterogeneous treatment effectiveness of nudge categories with control of confounders (e.g. demographics, medication, enrollments, and engagement.). No adjustment for impact of M1 nudges were made in M2 modeling.

Results: The study included 65,795 participants, mean age 56 (SD 10.4), 49.6% female, and 75.9% with baseline BP ≥ 130/80. Effectiveness of each nudge consumption was quantified by the increase in probability of HTN participants achieving control after exposure to additional nudges during the intervention period. For M1, mean effectiveness was 3.6% for sleep, 1.9% for BP checks, 1.8% for medication, 1.6% for journey engagement, and 1.5% for physical activity. For M2, mean effectiveness was 4.7% for BP checks, 3.4% for sleep, 3.5% for journey engagement, 2.8% for medication, and 2.3% for webinar.

Discussion: Nudges focused on BP checks and sleep early in HRMP participation had the greatest increase in the likelihood of BP control within 3 months. Casual inference methods allow the determination of appropriate content and timing to maximize clinical outcome improvement. Utilizing these methods will allow for design of more effective behavior change outreach for short-term and long-term clinical impact.