How we measure the SEVARA cohort.

Subscribers in the guarantee cohort have outcome measurements at baseline (month 0), month 3, month 6, month 12, month 18, month 24, and month 30. Weight is self-reported (via the app, with photo proof monthly). Body composition is measured via DEXA at one of our partner-network scan sites quarterly. Lab panels are pulled from Quest Diagnostics via patient-portal authorization.

Cohort average percent of weight loss maintained at each measurement point. Distribution of retention bands (lost 5-10%, 10-15%, 15%+). Eligibility-state distribution (pending, warning, paused, qualified, forfeit, refunded). Comparison to SURMOUNT-4 baseline curves where relevant.

Names, emails, phone numbers, dates of birth, exact body weights at a specific date, any combination of fields that could re-identify a subscriber, individual lab results, individual DEXA scans, or any row-level data of any kind.

Pre-launch and Year-1, the dashboard runs in aggregate-only mode under our internal data-handling protocol. Formal IRB submission for the Year-2 real-world evidence publication (the first proper RWE comparing bundled vs. unbundled GLP-1 outcomes) is owned by the medical director and kicks off within their first 90 days. Standard institutional review timelines apply (2-3 months minimum post-submission).

The guarantee applies to subscribers who: (1) entered at the Pre-GLP-1 (Stage D) or Onboarding (Stage B) journey stage; (2) maintained a compliance score ≥ 75 across the 30-month window (with up to two grace months for life events); (3) reached the month-12 weight-loss milestone of ≥ 15% of baseline body weight; (4) maintained at least 90% of that loss through month 30. Subscribers entering at maintenance (Stage A) have a separate guarantee: maintain ≥ 90% of their starting weight loss through month 30 from their bundle start date. Full terms shown at signup, never buried.

Outcome data corrections happen via compensating events on an append-only ledger. The historical state at any past date is always reconstructible. If we discover a methodology error that materially changes a published number, the correction is published in the same place with an explanation. We don’t silently update the page.