A digital nudge raised statin refills, and changing the channel did nothing

Design

Pragmatic, health system-embedded, prospective randomised open blinded endpoint trial with a two-stage SMART design

Population

20,604 adults with established or high-risk atherosclerotic cardiovascular disease and recent statin non-adherence; mean age 54.8, 40.9% female

Primary outcome

Statin refill within 14 days of each randomisation, with 28-day follow-up

Effect

14.4% vs 12.0% (adjusted risk difference 2.3 points, 95% CI 1.3-3.4); second-stage 13.0% vs 10.4%; switching channel 1.06 (0.94-1.17)

Statin non-adherence is one of the largest preventable causes of cardiovascular events, and health systems spend heavily on outreach without much randomised evidence that any of it works. ADHERE-ASCVD tested it inside Kaiser Permanente Northern California, randomising 20,604 adults with established atherosclerotic disease or high risk plus recent non-adherence to secure portal message, text message, non-secure email, or usual communication - and then re-randomising the non-responders at 14 days to repeat, switch channel, or usual care.

Initial outreach raised 14-day refills from 12.0% to 14.4%: adjusted risk difference 2.3 percentage points (95% CI 1.3-3.4), adjusted risk ratio 1.20 (1.10-1.30). Among non-responders, a second round raised refills again, 13.0% against 10.4% (difference 2.5 points, 1.3-3.7). Cumulative refill at 28 days was 24.9% against 21.2%, adjusted hazard ratio 1.21 (1.13-1.30). Switching channel was no better than repeating the same one (13.2% vs 12.5%, adjusted risk ratio 1.06, 0.94-1.17).

Two things follow. The first is that persistence beats sophistication: sending a second message works, and agonising over which channel does not. The second is the ceiling. Even with two rounds of outreach, three-quarters of these patients had not refilled by 28 days, so digital nudging is a cheap increment on a large problem, not a solution to it.

The design is worth noticing on its own account. This was a trial embedded in routine care, using data the system already had, testing something the system was already doing - and it produced a randomised answer at negligible cost. That is a template any health system with a functioning electronic record could copy, including large Indian hospital networks, and it is more transferable than the intervention itself.

• Send a second message to non-responders rather than redesigning the first one
• Do not invest in switching communication channels - it added nothing here
• Expect a small absolute gain: about 2 to 4 more refills per 100 patients contacted
• Pair outreach with a route to address the reason for stopping; a reminder does not fix a side effect
• Note that this was a system with an integrated record and pharmacy - the effect depends on that infrastructure

The statistics, in plain English

An adjusted risk ratio of 1.20 sounds substantial; the adjusted risk difference of 2.3 percentage points is the same result stated usefully, and means about two extra refills per hundred patients contacted. The comparison of switching versus repeating channel gave 1.06 with an interval from 0.94 to 1.17, which crosses 1.0 - so there is no evidence switching helps, and the trial was large enough to have found a meaningful difference. Refilling a prescription is a proxy for taking it, and this trial did not measure cholesterol, events, or whether tablets were swallowed.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.