Most treatment outcomes are marketing in disguise. Atlas built the Arc so we — and the field — could finally see real signal: engagement, validated-instrument deltas, dropout, sustained recovery. Honest numbers, published as we have them.
The recovery industry trades in inflated success rates that paper over relapse, attrition, and the patients who never made it to discharge. Atlas does not participate in that. We publish what the Arc measures, including what looks bad.
The numbers below are an early read from the first deployments of the Atlas Method across our anchor entities. They will get better as our sample grows, and we will publish them quarterly — with the dropout rate next to the engagement rate, every time.
Validated 9-item depression screen. Tracked at intake, monthly, and discharge.
Intake · Monthly · DischargeValidated 7-item anxiety screen. Tracked at intake, monthly, and discharge.
Intake · Monthly · DischargeWHO-developed screen for alcohol risk; complemented by DAST-10 for drug use.
Intake · Quarterly5-item subjective wellbeing measure. Weekly inside the Arc.
Weekly · in-Arc20-item trauma symptom measure. Trauma-affected clients only, monthly.
Monthly · indicatedDay-by-day Arc completion — sections opened, journals submitted, breath practices completed. Measured passively as patients use the Arc.
Daily · automaticMedian daily completion across the first deployments at NIRC and Millennium.
Median improvement in depression score from intake to day 90. Clinically meaningful threshold: −5.
Median improvement in anxiety score from intake to day 90.
Patients who left care or stopped engaging the Arc before day 90. Industry average: ~30–50%.
N = 124 patients · methodology: median, not mean — to control for outliers; intent-to-treat (attrition counted).
Recovery is not a binary. Anyone publishing a single rate is hiding either the denominator or the timeframe.
A 5-star rating from someone in residential care under contract pressure is not a clinical outcome.
The industry’s leading numbers are mostly fiction. We don’t grade on that curve.
We don’t filter the cohort to the patients who finished. Dropouts count — including the ones we drove away.
Of course people feel better at discharge. The real number is the one we publish at 6 months, 12 months, and 24 months. We are building toward that.
A revised report — with the prior quarter’s cohort closed out at six and twelve months — will post here on the first of every quarter.
Clinicians and researchers may request raw, de-identified data via the contact form. The Arc methodology paper is in preparation.