GiveBetter x USA

Shatterproof

Addiction treatment navigation, quality improvement, prevention and advocacy

Research time: ~16 min on GPT-6 Astra Light + ~1 min on GPT-6.1 Sol
  • Research — reviewed programs, finances and impact evidence.
  • Source audit — checked claims, assumptions and calculations.
  • Modeling — estimated costs, QALYs and uncertainty.

Updated: 2026-10-07

Summary

What do they do? Shatterproof works to improve addiction prevention, treatment access and quality through navigation, provider education, stigma campaigns and policy. Treatment Atlas is an information service, not a treatment provider; reaching a website does not mean receiving effective care. Our weak-confidence best estimate is about $4.5M per DALY on full reported resource costs; the estimate is highly sensitive to unmeasured treatment conversion.

Why we’re interested in this organization:

  • Navigation toward effective opioid treatment has a plausible, clinically consequential mechanism.

  • Public monitoring defines engagement metrics, making it possible to distinguish activity from verified treatment.

  • Current reports and audited accounts reveal the broader portfolio and unusually large donated-media component.

Our main reservations:

  • No comparative evidence located establishes how many additional treatment-years Atlas causes.

  • A favorable scenario is about $19K per DALY, but weak conversion produces negligible benefit; zero and harm remain possible.

  • Public contracts, available financial assets and an unspecified next-dollar workplan make marginal capacity uncertain.

What do you get for your dollar? $4.5M per DALY in USA. DALY is our one-year comparison unit. QALY-based health gains and income-welfare equivalents retain their original assumptions; they are not measured clinical DALYs. Retained clinical navigation reference; treatment payments and household resources assessed unknown.

Users, sessions, profile views, additional treatment starts and retained treatment-years. Health counted only after the explicit treatment conversion and clinical bridge.

1. What do they do?

Shatterproof combines treatment navigation, quality improvement, prevention and advocacy. Its 2025 report describes 1.2 million Atlas users, 1.6 million sessions and 6,700 provider learners. These are different denominators, not treatment starts. The report also describes stigma campaigns and policy activity; those branches are not assigned automatic health credit.

2. Monitoring and information sharing

The strongest inspectable monitoring is a Florida quarterly report: 29,613 users and 1,961 referrals, with referral explicitly defined as a facility-profile view. It does not verify patient contact, enrollment, medication, retention or survival. National reporting says 63% of facilities contacted ranked in the top fifth for science-based care; this is not evidence of the counterfactual provider choice. A useful beta evaluation would link consented user referrals to treatment records and compare against otherwise similar users of existing referral routes.

3. Qualitative assessment

Provider resources encourage established treatment principles. The clinical bridge is plausible: Sordo et al. found lower mortality during methadone or buprenorphine treatment in 19 observational cohorts, while induction and the period after stopping carried extra risk. Historical cohort differences are not randomized Atlas effects and may not transport to current fentanyl exposure or its user mix. The model uses a smaller net mortality benefit and short retained treatment duration. Prevention restrictions may shift substance use rather than eliminate it; stigma and policy effects may help or harm and remain explicitly unpriced.

Health-first resource assessment leaves actual signed household effects unknown, not absent; free navigation does not prove free treatment or a 90%-health share.

4. What do you get for your dollar?

Household resources are assessed unknown after a focused health-first check. The original conditional clinical reference remains about $4.5M per DALY: 11.80 USA QALYs at $52,809,073 of full annual recipient resources, including recognized donated media and services. The $21,522,901 cash-like numerator remains a separate diagnostic, not the main cost. This is not a complete combined valuation, a zero-income finding, or an empirical 90%-health share.

The current Treatment Atlas FAQ confirms that the public navigation tool is free and is not a treatment provider. Insurance and self-payment filters and facility-profile details do not guarantee coverage; users are advised to confirm with their insurer. Matching to covered, appropriate care may displace actual private payments or search time, but free Atlas does not mean free clinical treatment. Someone who otherwise goes without care does not automatically save a retail treatment bill; initiating care can instead increase copays, travel, treatment time and opportunity costs.

Current gift-attributable unique patients and resource-sharing households, state-contract allocation, actual alternative directories or clinicians, insurer assistance, private-payment changes and full net burdens remain unidentified. Full incurred negative costs cannot be discounted by clinical treatment response or positive overlap; health and autonomy overlap applies only to gains. Taxes, benefits, finite delay and duration, and household baseline resources require joint netting. Public-payer, provider, donated-media and patient resources need separate incidence without double charging costs already represented elsewhere. Payer savings, retail medical bills and ordinary earnings of survivors are not household consumption gains. All eight original clinical and cost worlds remain unchanged; broader resources remain unquantified, not zero.

The central model yields about 119 additional treatment-years before funding substitution, 60 after it, and 11.80 USA QALYs per annual activity cohort. At full recipient resource costs this is about $4.5M per DALY; excluding recognized donated resources gives about $1.8M. Neither is a provider treatment price. Clinical care paid by insurers, public programs and patients is additional to recipient costs and is not a social cost-effectiveness numerator. The favorable scenario requires much stronger conversion and retention; its roughly $185,000 price is not an evidence-based confidence limit. All other portfolio health is set net zero centrally, not asserted to be zero or used to claim a rigorous lower bound.

Pairing 2024 resource costs with 2025 reach is a stationary-scale approximation, not an observed same-year cost-per-output ratio. The alternative denominator excludes recognized donated resources from accrual expense; it is not cash flow. Additional treatment starts are net of alternative care in e; t is retained exposure conditional on those additional starts, not a second subtraction of the same start.

Model, assumptions and sensitivity

Full annual recipient resources, all programs and support; recognized in-kind and gross direct event costs included. Separate cash-like sensitivity. Clinical provider and wider social costs excluded.

Domestic treatment-navigation health; g=.99 subjective resident share. No overseas or territorial benefits counted. State footprint is not multiplied by national population.

PY=U*f*o*e*t; Q_all=b*(PY*(d*q+w)-h); Q_USA=g*Q_all; price=10*C/Q_USA for positive Q. One annual activity cohort, short treatment exposure, discounted remaining-life QALYs in q. No repeated lifetime cohort credit.

C
52809073 USD annual full recipient resource cost (observed). Audited gross 2024 expense, restoring direct donor benefits; includes recognized donated media and services. The 2024 costs and 2025 reach are paired as a stationary-scale approximation, not a same-year measured unit cost. [shat-audit24]
U
1200000 annual Atlas users (observed). 2025 report says more than 1.2 million people; use rounded floor, not unique patients or starts. [shat-impact25]
f
0.06622091648937967 profile engagements per user (judgment). Florida quarter 1,961 profile views / 29,613 users = .06622. Transport to national annual population is subjective; repeat views and national mix could differ. [shat-florida]
o
0.2 OUD-relevant fraction of profile engagements (judgment). 20% reserves most engagements for alcohol/other disorders, caregivers or irrelevant contacts; not measured Atlas case mix.
e
0.03 net additional effective treatment starts per OUD engagement (judgment). 3% after non-enrollment, inappropriate capacity, alternative locators and treatment that would happen anyway. A weak prior, not a measured Atlas effect; test zero. [shat-florida] [shat-resources]
t
0.25 additional retained treatment years per start (judgment). Quarter-year retained treatment exposure conditional on an additional start already adjusted for alternative care in e; no second subtraction of the same otherwise-received start and no lifelong adherence assumed. [shat-moud]
b
0.5 marginal funding additionality (judgment). 50% allows replacement of other donations, reserves and public contracts. This is separate from patients finding care anyway in e. [shat-audit24]
d
0.01 net deaths prevented per additional treatment-year (judgment). 1% between historical buprenorphine and methadone cohort rate gaps; discounts confounding and modern transport, includes induction/discontinuation risks rather than using peak untreated risk. [shat-moud]
q
15.00 discounted QALYs per premature death prevented (judgment). 15, broadly about 30 remaining years at .77 utility and 3% discount; not patient-level survival evidence. Test 10–20.
w
0.05 nonfatal QALYs per additional treatment-year (judgment). Small .05 utility improvement while retained, separately from mortality. Could be zero/negative; no long-term recovery dividend.
g
0.99 50-state/DC resident share of modeled health (judgment). 99% for domestic navigation, allowing nonresident use. Not inferred from headquarters or a claim all online users are residents.
h
0 other net health harm per annual activity cohort before funding additionality (judgment). Net zero centrally for unpriced prevention, stigma, policy and adverse effects; not established zero or a proven lower bound.

Weak-confidence best estimate: Cost: $52.8M; USA health QALYs: 11.80; all-population health QALYs: 11.92. {"C":52809073,"U":1200000,"f":0.06622091648937967,"o":0.2,"e":0.03,"t":0.25,"b":0.5,"d":0.01,"q":15,"w":0.05,"g":0.99,"h":0}; additional treatment-years before funding additionality=119.1976496808834.

Much stronger treatment conversion and retention: Cost: $52.8M; USA health QALYs: 2851.20; all-population health QALYs: 2880.00. {"C":52809073,"U":1200000,"f":0.15,"o":0.4,"e":0.1,"t":1,"b":0.8,"d":0.02,"q":20,"w":0.1,"g":0.99,"h":0}; additional treatment-years before funding additionality=7200.

Almost no additional effective treatment: Cost: $52.8M; USA health QALYs: 0.00; all-population health QALYs: 0.00. {"C":52809073,"U":1200000,"f":0.03,"o":0.1,"e":0.001,"t":0.1,"b":0.1,"d":0.002,"q":10,"w":0.01,"g":0.99,"h":0}; additional treatment-years before funding additionality=0.36000000000000004.

Excludes recognized donated resources from numerator: Cost: $21.5M; USA health QALYs: 11.80; all-population health QALYs: 11.92. {"C":21522901,"U":1200000,"f":0.06622091648937967,"o":0.2,"e":0.03,"t":0.25,"b":0.5,"d":0.01,"q":15,"w":0.05,"g":0.99,"h":0}; additional treatment-years before funding additionality=119.1976496808834.

No net treatment change: Cost: $52.8M; USA health QALYs: 0.00; all-population health QALYs: 0.00. {"C":52809073,"U":1200000,"f":0.06622091648937967,"o":0.2,"e":0,"t":0.25,"b":0.5,"d":0.01,"q":15,"w":0.05,"g":0.99,"h":0}; additional treatment-years before funding additionality=0.

No treatment benefit plus 10.00 QALYs other harm: Cost: $52.8M; USA health QALYs: -4.95; all-population health QALYs: -5.00. {"C":52809073,"U":1200000,"f":0.06622091648937967,"o":0.2,"e":0,"t":0.25,"b":0.5,"d":0.01,"q":15,"w":0.05,"g":0.99,"h":10}; additional treatment-years before funding additionality=0.

25.00 QALYs other net harm: Cost: $52.8M; USA health QALYs: -0.57; all-population health QALYs: -0.58. {"C":52809073,"U":1200000,"f":0.06622091648937967,"o":0.2,"e":0.03,"t":0.25,"b":0.5,"d":0.01,"q":15,"w":0.05,"g":0.99,"h":25}; additional treatment-years before funding additionality=119.1976496808834.

Only 10% additional capacity: Cost: $52.8M; USA health QALYs: 2.36; all-population health QALYs: 2.38. {"C":52809073,"U":1200000,"f":0.06622091648937967,"o":0.2,"e":0.03,"t":0.25,"b":0.1,"d":0.01,"q":15,"w":0.05,"g":0.99,"h":0}; additional treatment-years before funding additionality=119.1976496808834.

Combined resources unknown; clinical reference retained: Cost: $52.8M; USA health QALYs: 11.80; all-population health QALYs: 11.92. Original full-resource and cash-like diagnostics retained; no invented household cash.

Counterfactual: e is additional treatment net of alternative directories, clinicians, existing care and unavailable slots; b separately adjusts the effect of incremental support on recipient activity. Zero conversion or no funding additionality gives zero health absent other effects.

Attribution: Atlas receives only the net change in effective treatment, not all health among visitors. National transport of Florida engagement ratio and treatment conversion are explicit subjective judgments. Other provider/payer inputs are necessary complements, not free social resources.

Weak-confidence judgment model, not a measured causal return or probability interval. Scenarios span strong conversion, negligible yield, zero and negative health. No source identifies the marginal conversion parameter; root should treat price as provisional.

Sensitivity

  • Central effective treatment yield before funding substitution is about 0.0000993 treatment-years per user. This is a judgment, not an observed conversion rate.
  • At other central inputs, e must be 1.3425 to beat $100K per DALY, or 13.4254 to beat $100,000: impossible as a probability without stronger case mix, retention or other beneficial pathways.
  • At full resource costs, thresholds are about 2,667 / 26,671 additional treatment-years after funding substitution for $100K / $10K per DALY, using 0.20 QALY per year and g=.99.
  • Net other harm above 23.84 QALYs before funding additionality erases the central benefit; harm is not multiplied by the treatment-conversion probability.
  • The cash-like denominator lowers price by 59.2%, but donated advertising fair value is not necessarily its opportunity cost. A social model must add care costs and account for media displacement.
  • If geographic share is .95 rather than .99, price rises 4.2%; treatment conversion and funding additionality dominate.

Unresolved inputs

  • National linked enrollment, medication mix, retained treatment duration and comparator outcomes, with repeat users/caregivers separated.
  • Current activity-to-budget relationship and specific unfunded expansion; marginal public/private substitution and state contract restrictions.
  • Causal treatment/health effects of provider education, stigma campaigns and policy branches, including substitution and potential harms.
  • Current patient-level survival and quality-of-life inputs, clinical payer costs, media opportunity costs and resident attribution.
  • Current marginal gift and state-contract allocation; unique treatment patients and resource-sharing households; actual alternative private payments, insurer coverage and treatment costs; incremental search/travel/time/work burdens; finite duration, baseline resources and disjoint public/provider incidence.

5. Funding and previous grants

The 2024 audit reports $52,747,632 expense; restoring $61,441 direct event benefits gives $52,809,073. Recognized donated media/services total $31,286,172, leaving a $21,522,901 cash-like accrual numerator—not cash-flow expenditure. Available financial assets were $14,073,387; some contracts reimburse eligible costs. A donation can substitute for reserves or public/private support, so the model halves expected marginal activity. The 2023 audit gives gross comparable 2023/2022 costs of $25,439,567/$16,796,216; controlled affiliates had no activity. Three-year mean $31,681,619 is distorted by donated-media valuation and is not a current cash budget. No current quantified unfunded capacity plan was located.

Annual expenses

Organization-level spending, including programs, administration and fundraising. The research list averages three comparable, consecutive full fiscal years when available.

  • FY 2022: $16.8M; Shatterproof; controlled affiliates inactive in comparative years, 12-month period, Accrual GAAP, full recognized resources including donated media/services and gross direct event costs. Source
  • FY 2023: $25.4M; Shatterproof; controlled affiliates inactive in comparative years, 12-month period, Accrual GAAP, full recognized resources including donated media/services and gross direct event costs. Source
  • FY 2024: $52.8M; Shatterproof; controlled affiliates inactive in comparative years, 12-month period, Accrual GAAP, full recognized resources including donated media/services and gross direct event costs. Source

6. Sources

  1. 2024 audited financial statements, pp. 4–5 and notes 2, 7. Shatterproof / CliftonLarsonAllen. Published: 2025-09-12; retrieved: 2026-09-14.
  2. 2023 audited financial statements with 2022 comparative information. Shatterproof / independent auditor. Published: 2024-09-27; retrieved: 2026-09-14.
  3. 2025 Impact Report. Shatterproof. Published: not stated; retrieved: 2026-09-14.
  4. Current financial reports index. Shatterproof. Published: not stated; retrieved: 2026-09-14.
  5. Atlas quarterly monitoring, October–December 2025. Florida Department of Children and Families / Shatterproof. Published: 2026-01-27; retrieved: 2026-09-14.
  6. Mortality risk during and after opioid substitution treatment: cohort meta-analysis. Sordo and colleagues / BMJ. Published: 2017-04-26; retrieved: 2026-09-14.
  7. Treatment Atlas provider resources and quality improvement. Shatterproof. Published: not stated; retrieved: 2026-09-14.
  8. Current Treatment Atlas FAQ: public charges, financing and payment options. Shatterproof / Treatment Atlas. Published: not stated; retrieved: 2026-10-07.