GiveBetter x SF

Micah's Hugs

Risk-network naloxone access within education and recovery support

Research time: ~6 min on GPT-6 Astra Light
  • Research — reviewed programs, finances and impact evidence.

Published: 11 September 2026.

Donate

Summary

What do they do? Micah's Hugs educates students, distributes naloxone and test strips, maintains or supports stations, and assists recovery housing and scholarships. Its own stations are described in Sonoma/Marin; its map also lists unrelated providers statewide. More

Why this approach interests us

  • A stocked, accessible station or prepared peer may change timely rescue for someone otherwise missed. Actual low whole expense makes even a small additional high-risk network potentially consequential.

Our main reservations

  • The favorable10% world supplies 98.9% of signed expected benefit. Without it, renormalized price is $34,636,396 per 10 . Neither annual channel mix nor unique-risk coverage is known.

What do you get for your dollar?

The research list uses the central scenario: $4,751,636 per 10 Bay QALYs. Signed prior-weighted estimate $410,820 per 10 Bay QALYs, or 0.243416 Bay QALYs per$10,000. Central $4,751,636 per 10 . These are exploratory assumptions, not measured impact. Inspect the model and assumptions.

whole expense
$66,076 — Calendar 2025; all education, recovery housing and scholarship costs retained.
activity, not outcome
15,000+ doses — Annual doses are not unique patients or rescues; the map includes other providers.
favorable tail
98.9% — Share of signed expected benefit from the 10% favorable world; central price is $4.75M per 10 Bay QALYs.
scope
Partial health — Whole organization cost; unquantified recovery and education benefits are not assumed zero.

1. What do they do?

Micah's Hugs educates students, distributes naloxone and test strips, maintains or supports stations, and assists recovery housing and scholarships. Its own stations are described in Sonoma/Marin; its map also lists unrelated providers statewide.

Convert dose units into a risk network

Separate station and school/general carriers, infer only a small relevant at-risk network, then deduplicate shared people.

Compare finite survival

Apply a small net fatal-risk difference for one support year to the selected at-risk people—not to every student or kit holder. Both arms then face common mortality.

Ask what cash actually changes

Scale only within a bounded$10,000illustration, with explicit funding-null and public/volunteer substitution.

Scope of this review. All organizational expense retained while only incremental naloxone-related survival/morbidity is modeled. Test strips, general education, recovery housing, scholarships and family support remain unquantified, not zero. Donated inventory, volunteer time, EMS and partner resources are incompletely valued; this is not complete social resource return.

2. Monitoring and information sharing

AnnualSonoma distribution plus general school audiences

Original2025IRS990-EZ and current official education page. 15,000+annual doses;1,167 trainingparticipants at anOctober2025 school event;38,000cumulative doses on webpage.

Our assessment. Different denominators cannot be added. No annual channel split, own-site count or unique high-risk recipient count. Each student receiving a box does not imply personal overdose risk.

Micah hosts/adopters and third-party map listings

Official station and sponsorship descriptions. Host/adopters share refill duties; statewide map includes other suppliers.

Our assessment. Do not count all pins as Micah. Cash must add useful availability beyond other peers, pharmacies, public supplies andEMS; no priced extra tranche.

19 communities,2002–2009

Massachusetts observational interrupted time series. Naloxone/education implementation associated with lower mortality; adjustedratios.73/.54 for low/high implementation.

Our assessment. Not randomized or current Bay dose-to-death effect. Used only as mechanism support; no rate-ratio transfer.

67 communities in fourstates during fentanyl/pandemic period

HEALing Communities cluster-randomized multicomponent intervention. Primary opioid mortality difference not significant; aRR.91,95%CI.76–1.09.

Our assessment. Distribution growth is not sufficient proof of incremental mortality reduction. Supports clinical-null scenarios without denying naloxone reversal pharmacology.

Selected overdose survivors, not school attendees

Historical US cohorts after nonfatal overdose. Larochelle4.7 and Olfson7.783all-cause deaths per 10 0 person-years in first-year observations.

Our assessment. Rough scale only. Do not impose these risks on general recipients or claim10–15 year rates validated. Hazardgain, cohort selection and long-run utility remain judgments.

Synthetic deduplicated risk networks

Analyst unique-person survival model. Central131.25 person-equivalents, .001net annual hazard reduction forone year, then common .07 mortality through10 years.

Our assessment. No repeated lifetime rescue credit. Numerically valid trajectory is not empirical calibration; school-only, common-survival andone-year sensitivities materially worsen price.

3. Qualitative assessment

A stocked, accessible station or prepared peer may change timely rescue for someone otherwise missed. Actual low whole expense makes even a small additional high-risk network potentially consequential.

Key reservations

  • The weighted headline is98.9%driven by the favorable world; central is$4.75M/10.
  • Annual channel mix and own-site count unknown; school recipients cannot inherit overdose-survivor risks.
  • No observed clinical effect, reported rescue ledger or verified prospective cohort.
  • Cash response and volunteer/public inventory scale unknown; sponsorship is not a priced offer.
  • Whole-cost partial-health scope omits potentially real recovery/education benefits.
  • No full economic resource valuation; external donated complements may increase cost.

Benefits not included in our estimate

  • General school education or test strips automatically preventing overdose
  • Recovery-housing health, scholarships and family support—not assigned zero, simply unquantified
  • Third-party statewide map outcomes
  • All doses counted as successful rescues
  • Unlimited survival or repeatedly restarting follow-up

4. What do you get for your dollar?

$410,820 per 10 Bay QALYs

Six correlated judgment worlds were locked before first output. Average signed QALYs, not prices. A favorable scenario is not the central case or a verified funding offer.

A better life is our comparison unit of 10 additional quality-adjusted life years (QALYs), potentially spread across people. These are uncertain estimates, not measured returns or verified donation offers.

How we calculate the estimate

Finite risk-network health: gift / whole expense × dose units / units per carrier × [station share × station network reach + (1−station share) × school/general network reach] × network dedup × cash response × [finite survival Q per person − morbidity harm] × Bay share
0.24341586015886482 Bay QALYs/$10,000

Model inputs and assumptions
Whole annual expense
66076 (range: 61176+4900;2025). All programs retained, not815 supplies budget; historical accounting envelope, not all economic resources. Accounting/activity anchor; not impact.
Annual dose units
15000 (range: Reported15000+). Lower edge of original annual claim; cumulative38000excluded. Accounting/activity anchor; not impact.
Dose units per distinct carrier
4 (range: 2–6). Judgment for repeated possession/refills/multiple-dose stock; not observed package size. Analyst judgment / uncertain.
Station versus school/general channel
.25 (range: .1–.5). Unknown annual split. Residual explicitly school/general; map pins not owned stations. Analyst judgment / uncertain.
At-risk network members per station carrier
.25 (range: .1–.5). Judgment, not all station users high risk; before shared-network dedup. Analyst judgment / uncertain.
At-risk network members per school/general carrier
.01 (range: .002–.05). Peer protection may concern someone else. No7% mortality among students. Analyst judgment / uncertain.
Distinct-risk-network fraction
.5 (range: .3–.65). Judgment eliminating repeated people reached by multiple carriers/channels. Analyst judgment / uncertain.
Net annual fatal hazard reduction
.001 (range: −.0002 to.003 plus0). One activeyear; already net of alternative rescue, availability/use and efficacy. Not estimated from doses or retrospective rescues. Analyst judgment / uncertain.
Selected at-risk cohort all-cause hazard
.07/year (range: .04–.10). Judgment comparable to HEPPAC; historical overdose cohorts are only rough scale anchors, not sampled Micah population. Analyst judgment / uncertain.
Common post-support hazard
.07/year (range: .04–.10). Same for both arms afteryear 1; no sustained funding assumed; constant extrapolation uncertain. Analyst judgment / uncertain.
Finite follow-up
10 years (range: 5–15;diagnostic1). Total horizon includes supportyear, not tenyears of program effectiveness. Analyst judgment / uncertain.
Health utility while alive
.65 (range: .55–.75). Judgment, not value of person or measured local outcome. Analyst judgment / uncertain.
Ordinary gift capacity response
.3 (range: 0–.6). Cash-created/sustained access beyond funded baseline; not another clinical efficacy penalty. Analyst judgment / uncertain.
Bay benefit fraction
.98 (range: .95–1). Own Sonoma/Marin footprint, with uncertain downstream geography; statewide third-party map excluded. Analyst judgment / uncertain.
Independent morbidity Q per new risk person
.0002 (range: 0–.002). Small signed allowance for adverse consequences; no measured incidence. Fatal harm separately represented by negativehazardGain. Analyst judgment / uncertain.
Annual health discount
.03 (range: Fixed). Annual compounding; one finite trajectory per distinct risk person. Analyst judgment / uncertain.
Bounded gift diagnostic
10000 (range: 0–10000). Not assurance that15% ofannualexpense can be deployed; no linear100koffer. Analyst judgment / uncertain.

Bounded$10,000illustration—not a current capacity commitment

  • funding_null: No positive finite price per 10 ; weight0.2. BayQ0; annual synthetic unique-risk cohort131.25; Q/person0.003803744023502164.
  • clinical_null: No positive finite price per 10 ; weight0.2. BayQ0; annual synthetic unique-risk cohort131.25; Q/person0.
  • harm: No positive finite price per 10 ; weight0.1. Bay Q-0.016119810933111284; annual synthetic unique-risk cohort131.25; Q/person-0.0007603048324360516.
  • cautious: $15,068,899,318 per 10 ; weight0.2. BayQ0.0000066361847595674125; annual synthetic unique-risk cohort8.85; Q/person0.0003521549264553287.
  • central: $4,751,636 per 10 ; weight0.2. BayQ0.021045382969140044; annual synthetic unique-risk cohort131.25; Q/person0.003803744023502164.
  • favorable: $41,525 per 10 ; weight0.1. BayQ2.40817437421396; annual synthetic unique-risk cohort1340.625; Q/person0.019882132705586527.
  • half_coverage: $821,639 per 10 . Diagnostic changes only its named parameter family; complete inputs/results in sensitivity artifact.
  • one_year: $9,273,000 per 10 . Diagnostic changes only its named parameter family; complete inputs/results in sensitivity artifact.
  • common_central_survival: $712,042 per 10 . Diagnostic changes only its named parameter family; complete inputs/results in sensitivity artifact.
  • f2f_survival: $1,691,883 per 10 . Diagnostic changes only its named parameter family; complete inputs/results in sensitivity artifact.
  • school_only: $2,264,686 per 10 . Diagnostic changes only its named parameter family; complete inputs/results in sensitivity artifact.
  • half_cash: $821,639 per 10 . Diagnostic changes only its named parameter family; complete inputs/results in sensitivity artifact.
  • double_resources: $821,639 per 10 . Diagnostic changes only its named parameter family; complete inputs/results in sensitivity artifact.

Uncertainty. For t≤1, S_without=exp(−baselineHazard×t), S_with=exp(−(baselineHazard−hazardGain)×t). Thereafter each arm carries its year 1survival forward with the same postHazard. Integrate their discounted difference to the finite horizon and multiply utility. No repeat rescue restarts survival. Channel/reach/dedup priors are not measured; all-cause .07 applies only to hypothesized at-risk network members, never all students. The central131.25 person-equivalents is model arithmetic, not a count. Continuous .03 F2F discount differs slightly; comparative sensitivity uses this model's annual3% consistently.

5. Funding and previous grants

HOLD giving recommendation. No priced marginal capacity or validated dose-to-network clinical conversion; favorable expected price alone is not donor readiness.

Original2025 expense$61,176 plus$4,900netted event costs =$66,076. ScheduleO identifies$13,136 recoveryhousing,$7,300 scholarships,$815harm-reduction supplies and donated officer effort; supplies alone are not total cost. Cash/savings$94,114,CD/shareaccount$34,690 and$3,490fundsheld are not an inferred funding gap. Public naloxone is available, but Micah's exact inventory source is unverified. Host/adopt roles show refill labor is essential. Advertised sponsorship tiers buy recognition, not a verified station/annual capacity bundle.

This review does not establish a verified marginal funding offer or a complete history of grants.

Donate

Opens the organization’s giving page. A general donation may not fund the specific activity modeled here; confirm allocation with the recipient.

6. Sources

  1. Original calendar2025 Form990-EZ. Micah's Hugs / IRS. Primary. Published: Submitted2026-07-14; retrieved: 2026-09-11.
  2. Original2025 ScheduleO. Micah's Hugs / IRS. Primary. Published: Submitted2026-07-14; retrieved: 2026-09-11.
  3. Education and student distribution. Micah's Hugs. Primary. Published: Undated; describes2025-10-17 event; retrieved: 2026-09-11.
  4. Narcan locations. Micah's Hugs. Primary. Published: Undated; retrieved: 2026-09-11.
  5. Host, adopt and sponsor stations. Micah's Hugs. Primary. Published: Undated; retrieved: 2026-09-11.
  6. Official donation page. Micah's Hugs. Primary. Published: Undated; retrieved: 2026-09-11.
  7. Naloxone Distribution Project. California DHCS. Primary. Published: Current program; prior phase source; retrieved: 2026-09-11.
  8. Post-overdose mortality cohort. Larochelle et al.. Primary. Published: 2018; retrieved: 2026-09-11.
  9. Causes of death after nonfatal overdose. Olfson et al.. Primary. Published: 2018; retrieved: 2026-09-11.
  10. Community overdose education and naloxone. Walley et al.. Primary. Published: 2013; retrieved: 2026-09-11.
  11. Randomized community intervention results. NIH / HEALing Communities investigators. Primary. Published: 2024-06-16; retrieved: 2026-09-11.

Annual expenses: years and sources

Average annual expenses (three consecutive fiscal years): Not available. Organization size is separate from the modeled cost-effectiveness of a donation.

Micahs Hugs

Form 990/990-EZ reported whole-entity expenses; includes program, administration and fundraising costs, but excludes any costs netted against revenue.

Historical extracted filing window; newer original filings may be available and have not yet been reconciled.