Found Health × COYL. Interim data.
A pre-registered randomized trial of behavioral-interrupt protocol vs. standard care across GLP-1-active adults. This page is the public ledger — pre-registered before recruitment, updated as interim data lands.
Last updated . Pre-registered 2026-04-15. Study ID COYL-FH-001.
01 · Pre-registered design
What we said we'd measure. Before recruitment opened.
Population
GLP-1-active adults, 18–65, English-speaking, US residents
Arms
Intervention
COYL behavioral-interrupt protocol via iOS + web. Danger-window-aware notifications, same-night recovery flow, weekly archetype digest.
Control
Standard Found Health care + non-personalized weekly behavioral-health email (sham contact-time control).
Primary endpoint
Slip-recovery rate at 30 days — % of logged autopilot-loop slips that complete the same-night stabilization protocol within 24 hours.
Secondary endpoints
- D7 retention (% active in days 0–7)
- D30 retention (% active in days 14–30)
- Streak integrity at 30 days (% of streaks > 1 with grace-window survival vs. legacy reset behavior)
- Excuse-category frequency change (baseline → week 4)
Exclusion
Active eating disorder per DSM-5; current substance-use crisis; <30 days from any major mental-health event. Routed to crisis services per the /safety policy.
IRB status
IRB-exempt (minimal-risk behavioral)
02 · Results
What we've measured. Or haven't yet.
Every endpoint below is pre-registered above. PENDING rows convert to INTERIM as data locks. INTERIM converts to FINAL on study completion. No post-hoc additions on this page.
Primary endpoint
Slip-recovery rate at 30 days — % of logged autopilot-loop slips that complete the same-night stabilization protocol within 24 hours.
Interim readout pre-registered for the 60-day data lock. Awaiting cohort completion of the recruitment window.
Recruiting
Secondary endpoint
D7 retention
D7 cohort active. Numbers post once the first wave passes day 14 in study.
—
Secondary endpoint
D30 retention
D30 cohort active. Numbers post on the 60-day interim window.
—
Secondary endpoint
Streak integrity at 30 days
Numerator: streaks that survived ≥1 missed day via grace window. Denominator: all streaks > 1 in the study cohort.
—
Secondary endpoint
Excuse-category frequency change
Baseline-to-week-4 delta per ExcuseCategory enum (8 categories). Reported as cohort distribution shift, not individual.
—
03 · The honesty frame
What we publish, what we don't, and what we do with a null result.
Positive interim — publish immediately. Methods, data, p-values, all of it. We trade speed for transparency every time.
Null interim— we keep the trial running and publish the null as a separate methodology proof point. Most behavioral RCTs come back null at p<0.05 on first read. That's expected. A pre-registered null with clean methodology is still rigor — and we don't hide it.
Underpowered— we extend the timeline and renegotiate cohort size with the Found Health PI. Documented here as “cohort enrollment ongoing” with the new target N.
Trial halt — published with reason. The page does not silently disappear.
04 · Partner with us on the next study
Clinics, telehealth providers, payers.
We're actively scoping a second arm. GLP-1 discontinuation relapse-prevention is the highest-leverage next study. If you have a patient population and an IRB pathway, we have the protocol and the instrumentation.