Adherence to Dyslipidemia Therapy: Harnessing Evidence From Randomized Evaluations in AtheroSclerotic CardioVascular Disease

Atherosclerotic cardiovascular disease (ASCVD) is the most common cause of death worldwide. Statins can reduce ASCVD events by 25%, yet nearly half of high-risk adults remain untreated, and adherence is poor. Since statin adherence impacts quality metrics like Medicare Stars ratings, even small gains in refill rates can boost health plan performance. However, no data driven approaches exist to optimize statin adherence.

The ADHERE-ASCVD study is a pragmatic, Prospective Randomized Open-label Blinded End-point (PROBE) randomized clinical trial designed to evaluate and optimize patient-facing messaging strategies to improve statin refill and adherence. The project leverages a two-stage Sequential Multiple Assignment Randomized Trial (SMART) design to evaluate both initial outreach strategies and adaptive follow-up approaches among patients with suboptimal statin adherence.

In the first stage, eligible participants will be randomized to one of four outreach strategies: secure portal message, SMS (text message), non-secure email, or usual care (no proactive outreach). The primary response window is 14 days following message delivery, during which statin refill will be ascertained using pharmacy dispensing records. Participants who refill within this window will be classified as responders and will not receive further study intervention.

Participants who do not refill within 14 days will be classified as non-responders and re-randomized in a second stage to receive one of three follow-up strategies: (1) continued outreach using the same communication modality ("stay"), (2) outreach using an alternative modality ("switch"), or (3) usual care with no additional outreach. For the initial implementation phase, message content will remain consistent across stages within each modality, and the intervention focuses on evaluating modality and sequencing rather than content optimization.

The primary estimands correspond to (1) first-stage effects comparing outreach modalities, (2) conditional second-stage effects among non-responders comparing stay, switch, and usual care strategies, and (3) dynamic treatment regime (DTR) effects comparing embedded adaptive sequences of interventions. DTR effects will be estimated using inverse-probability-weighted estimators that account for sequential randomization.

In addition to estimating average treatment effects, the study will develop predictive models to characterize treatment effect heterogeneity across patient subgroups. The modeling framework focuses on estimating differences in response across messaging strategies conditional on baseline patient characteristics. Model evaluation will emphasize calibration of predicted treatment contrasts and the identification of clinically meaningful differences between strategies that can inform targeted outreach.

This study is designed to generate actionable evidence within a learning health system by identifying optimal messaging strategies and sequences for improving statin refill and adherence, and enabling future deployment of targeted, data-driven outreach approaches.

Inclusion criteria

• Age ≥18 years
• Active Kaiser Permanente Northern California (KPNC) membership
• Diagnosis of atherosclerotic cardiovascular disease (ASCVD) or diabetes and inclusion in the Preventing Heart Attacks and Strokes Everyday (PHASE) Registry
• No statin prescription fill within the prior 3 months
• Statin adherence <80% in the prior 12 months
• Active kp.org account within the past 12 months
• Valid email address and text-capable phone number on file

Exclusion criteria

• Documented allergic reaction or contraindication to statin therapy
• Receiving hospice care
• Medicare beneficiaries
• Other exclusions applied per operational standard practice