Personalized behavioural nudges improve statin adherence

Using artificial intelligence-powered software to deliver personalized behavioural nudges helps improve patient adherence to statins, a new trial has found.

Researchers enrolled and randomly allocated 182 adult patients to receive either behavioural nudges or a no-nudge control. The content, frequency, delivery modality, timing, and feedback metric of nudges were personalized using machine learning algorithms. To avoid fatigue, nudges were limited to once per week, where possible. The primary trial outcome was patient adherence to statin prescriptions at 12 months after baseline, measured as the proportion of days covered (PDC).

Over the study duration, each patient in the intervention arm received an average of 56 nudges, corresponding to 1.07 nudges per week. Such alerts were delivered mostly as text messages (78.4 percent), followed distantly by emails (13.4 percent) and interactive voice responses (8.2 percent).

At 12 months, mean statin PDC was significantly higher among patients who received nudges vs controls (0.742±0.318 vs 0.639±0.358; p=0.042). PDC in the nudge arm was also higher at 3, 6, and 9 months but not significantly so relative to controls.

When researchers dichotomized patients as adherent (PDC ≥80 percent) or nonadherent (PDC <80 percent), they found that there was a significantly greater proportion of adherent patients at 12 months in the nudge arm (66.3 percent vs 50.5 percent; p=0.036).

Nudges were also safe, yielding a 12-month major adverse cardiovascular event rate of 6.7 percent, as opposed to 10.8 percent in controls (p=0.44).

“In a proof-of-concept trial, precision behavioural nudges driven by artificial intelligence increased statin adherence a clinically important amount and held the gains over a 12-month period in general cardiology patients,” the researchers said. “This persuasive approach may improve health by guiding patient choices in the home, workplace, and other locales where a patient lives their life.”