GiveBetter x SF

HIV Education and Prevention Project of Alameda County (HEPPAC)

Harm reduction, overdose prevention, treatment navigation, drug checking, and wraparound services

Research time: ~39 min on GPT-6 Astra Lite + GPT-5.6 Sol
  • Source audit — checked claims, assumptions and calculations.
  • Research — reviewed programs, finances and impact evidence.
  • Modeling — estimated costs, QALYs and uncertainty.

Published: 10 September 2026.

Donate

Summary

What do they do? HEPPAC is an Oakland nonprofit serving Alameda and Contra Costa Counties. Its portfolio includes overdose-prevention education and naloxone, syringe access, medication-treatment navigation, HIV/HCV services, drug checking, mobile outreach, drop-in basic needs, wound care, condoms, acupuncture/herbalist services, and B-DUHB interventions. Its donation page also identifies fiscal sponsorship for Punks With Lunch and J&MP Innovations. More

Why this approach interests us

  • Naloxone rescue, medication treatment, and sterile-syringe access have credible severe-endpoint mechanisms. HEPPAC operates an integrated East Bay harm-reduction portfolio and its official donation page says gifts support community services.

Our main reservations

  • HEPPAC does not publicly report current unique participants, annual reversals, retained medication treatment, infections averted, checking-linked behavior or cross-program overlap. A same-system navigation study is observational and noncurrent, and a newer randomized peer-navigation trial was inconclusive.

What do you get for your dollar?

The corrected unique-naloxone-cohort model estimates $8,133,469 per 10 Bay QALYs for an illustrative $100,000 whole-organization gift. All-beneficiary cost is $7,963,460 per 10 QALYs; these are different denominators. No verified marginal offer. Recurrence correction alone with old severe survival assumptions gives $12,849,714 per 10 Bay QALYs versus the preserved old $11,112,457; separately revised survival priors improve the combined estimate. Inspect the model, assumptions and scenarios.

bay whole-gift estimate
$8,133,469 — Per10 Bay QALYs;20% no-benefit prior, not calibrated.
central bay gift
$15,341,409 — Central scenario, not a verified quote.
whole expense
$4,953,186 — FY2025 full legal-organization cost retained.
bay / sf share
97.91% / 4.20% — Weighted prior geography; SF nested in Bay, not residence observations.
funding room
Unverified — No priced next-gift capacity.

1. What do they do?

HEPPAC is an Oakland nonprofit serving Alameda and Contra Costa Counties. Its portfolio includes overdose-prevention education and naloxone, syringe access, medication-treatment navigation, HIV/HCV services, drug checking, mobile outreach, drop-in basic needs, wound care, condoms, acupuncture/herbalist services, and B-DUHB interventions. Its donation page also identifies fiscal sponsorship for Punks With Lunch and J&MP Innovations.

Prevent and reverse overdose

OPEND distributes naloxone with brief training. Refill access is unlimited, so doses, people and reversal events require separate deduplication.

Link people to medication treatment

Substance-use navigators may improve 30-day engagement, but HEPPAC publishes no current annual starts, medication mix or retained treatment-years.

Reduce infection and supply risk

Syringe access has credible HIV-prevention evidence; FTIR and confirmatory drug checking may change behavior, but HEPPAC-specific infections and overdoses averted are not measured publicly.

Provide integrated support

Screening/linkage, mobile outreach, condoms, drop-in services, wound care, acupuncture/herbalist services and B-DUHB may add value. They remain in cost and receive no separate QALYs here.

Scope of this review. The model uses the entire FY2025 expense of $4,953,186. Events, items and people are never summed as recipients. Naloxone, navigation and checking share mortality endpoints and receive an explicit overlap sensitivity; person-level overlap remains unresolved. Fiscal-sponsorship/pass-through activity inside expense is also unresolved.

2. Monitoring and information sharing

1,328 discharged patients with alcohol, opioid, cocaine or methamphetamine diagnoses at three Alameda public emergency departments, September 1, 2021–January 31, 2022

Observational implementation study with propensity matching. 119 received navigation; 30-day treatment engagement was 50.4% versus 15.9%; matched OR 2.1 (95% CI 1.2–3.5).

Our assessment. Same-system mechanism evidence only. It is a five-month, noncurrent, all-SUD study and supplies no annual HEPPAC outcome volume, retained MOUD or mortality.

247 NYC emergency-department patients after opioid overdose

Randomized peer-navigation trial. Composite adverse-event RR 1.02 (0.72–1.45); mortality HR 0.55 (0.22–1.30); 90-day MOUD OR 1.64 (0.68–3.94).

Our assessment. Inconclusive rather than proof of no effect; mortality was underpowered and the peer model differs from Alameda medication-first navigation. It prevents treating observational linkage as certain mortality benefit.

Lay naloxone studies

Systematic review. Reported survival after lay naloxone was 93–98% across uncontrolled studies.

Our assessment. Supports rescue efficacy but does not reveal which HEPPAC reversals would otherwise be fatal or rescued elsewhere.

People who inject drugs using syringe-service programs

US evidence synthesis. Evidence was sufficient for HIV prevention; higher-quality studies in one meta-analysis reported RR 0.42 (0.22–0.81).

Our assessment. Local incidence, coverage, unique users and overlap with other providers are absent, so the model varies infections averted directly.

People using drug-checking services

Systematic review. Evidence suggests changes in intentions and sometimes behavior; mortality effects are not established.

Our assessment. HEPPAC drug-checking receives small central credit and a clearly labeled favorable stress assumption.

Whole HEPPAC legal organization

IRS Form 990 accounting data. FY2025 revenue $5.162M, expense $4.953M, surplus $209K and net assets $2.363M.

Our assessment. Full expense is the whole-organization denominator; fiscal-sponsorship/pass-through composition and marginal funding remain unresolved.

Synthetic HEPPAC selected high-risk cohort; unique people and person-time unobserved

Analyst finite-cohort model with historical survivor mortality context. One active year; finite5/10/20-year follow-up; supported common hazards.10/.07/.04; utility.55/.65/.75. All are explicit priors.

Our assessment. Historical first-year mortality4.7–7.8/100 person-years supports scale only, not prospective cohort selection or constant long-run hazards. Recurrence-only correction worsens old price; survival-prior revision separately improves it.

3. Qualitative assessment

Naloxone rescue, medication treatment, and sterile-syringe access have credible severe-endpoint mechanisms. HEPPAC operates an integrated East Bay harm-reduction portfolio and its official donation page says gifts support community services.

Key reservations

  • Current unique participants, kits or doses, reversal events, treatment starts and retention, checking-linked behavior, infections and cross-program linkage are not publicly reported.
  • Unlimited naloxone refills make item, person and event deduplication essential; reported reversals are not deaths averted.
  • Navigation, naloxone and checking may benefit the same person and share mortality; an overlap sensitivity is not person-level deduplication.
  • MOUD QALYs per retained-effective engagement are bounded sensitivities, not estimates derived quantitatively from the Sordo meta-analysis.
  • Naloxone now uses unique-cohort finite survival; drug checking retains the old finite-event sensitivity. Neither pathway has observed current local prognosis.
  • Gift deployability, realization, scenario weights, geography and gross multipliers are analyst priors.
  • The $700K county document is a staff-requested amendment to maintain staffing, not evidence in this packet of execution, expiry-created need or private-gift additionality.
  • Bay and SF shares are service-geography-informed assumptions; beneficiary residence is unknown.
  • The gross-resource multiplier is illustrative only, neither measured nor a demonstrated lower bound.
  • Fiscal sponsorship/pass-through activity may be included in expense, but this packet cannot reconcile it.

Benefits not included in our estimate

  • Drop-in showers, laundry, food, hygiene supplies and resource navigation
  • HIV/HCV screening, counseling, referral and treatment linkage distinct from syringe prevention
  • Community and mobile outreach, condoms, education, advocacy and task-force effects
  • Abscess wound care and other community-based medical services
  • Acupuncture and herbalist services
  • B-DUHB interventions
  • Benefits attributable to fiscally sponsored or pass-through programs not reconciled to HEPPAC expense
  • Criminal-justice, housing, dignity, stigma and emergency-service outcomes

4. What do you get for your dollar?

$8,133,469 per10 Bay QALYs; $7,963,460 all-beneficiary.

The executable first models annual organization-attributed QALYs under explicit pathway, causal-additionality and overlap judgments. It then scales a hypothetical $100,000 unrestricted gift pro rata to annual expense, applies funding deployability and realization discounts, and weights downside, central, favorable and no-benefit cases. The separate gross-resource figures are illustrative multipliers, not measured estimates or lower bounds. V2 replaces only within-naloxone event-linear survival with one finite trajectory per synthetic unique person. Other pathway values and coarse cross-pathway overlap remain unchanged and unvalidated.

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

BAY WHOLE-GIFT EXPECTATION: 10 × $100,000 / sum(world weight × whole-gift QALYs × Bay share)
0.12294876734953836 Bay QALYs per$100,000; $8,133,469 per10 Bay QALYs

Model inputs and assumptions
Annual treatment starts (prior; before retention)
25 (range: 10 / 25 / 50). No current HEPPAC outcome volume exists. Anderson et al. observed 119 SUN recipients within a five-month, all-SUD, noncurrent cohort; it is transfer context and is never annualized into HEPPAC outcomes. Model additionally multiplies retainedEffective .50/.60/.70; these starts are not already retained engagements. Very low; judgment-only.
QALY per retained-effective engagement
0.50 (range: 0.25 / 0.50 / 1.00). Sordo et al. supports the direction of benefit while in treatment, not these magnitudes. Drug mix, retained years, mortality risk difference, morbidity and local transfer are missing. Very low; bounded sensitivity.
Annual rescue-opportunity equivalents (prior)
50 (range: 20 / 50 / 137). Twenty and 50 are scenario values. The 137 figure is an old snapshot with no clean annual window and appears only in the favorable stress case. None is observed current annual throughput or prospective person-time. Very low for current HEPPAC.
Otherwise-fatal probability and HEPPAC attribution
10% × 75% (range: 5% × 60% / 10% × 75% / 15% × 85%). Lay naloxone survival is high, but the untreated counterfactual and alternative rescue are not observed. Low.
Legacy finite QALY term for drug checking only
2.291 (range: 1.342 / 2.291 / 3.640). Discounted finite sums use 5/10/20-year horizons, annual survival 65%/75%/82%, utility 0.55/0.65/0.75 and 3% discounting. Retained only in unchanged drug-checking pathway, NOT the revised naloxone term. Low; judgmental.
Infections averted and QALYs
1 infection × 65% attribution × 5 QALYs (range: 0.25 × 50% × 3 / 1 × 65% × 5 / 3 × 80% × 8). Syringe-service evidence supports HIV prevention, but HEPPAC publishes no unique exposure, baseline incidence or local infections-averted analysis. Very low locally.
Fatal-overdose equivalents from drug checking
0.25 (range: 0.05 / 0.25 / 1). Behavior evidence is emerging and mortality is not established. If the proposed 50 samples/month occurred, these imply one equivalent per 12,000/2,400/600 samples. Very low.
Shared mortality overlap adjustment
70% retained (range: 80% / 70% / 60% retained). Naloxone, navigation and checking may benefit the same person and share survival endpoints. This sensitivity does not solve deduplication. Very low; no person linkage.
Pathway outcome additionality
50% (range: 25% / 50% / 75%). Separates service activity from outcomes that would not otherwise occur; retained in both whole-organization and gift calculations. Low.
Ordinary-gift funding/capacity deployability
45% (range: 20% / 45% / 70%). Moderate reserves and substantial contributions allow some marginal use, but HEPPAC has not made a dated next-$100,000 commitment and public funding may be displaced. Low.
Gift-to-outcome realization
85% (range: 70% / 85% / 95%). Discounts implementation delay and failure between receipt, service expansion and health outcome. Low.
Unconditional scenario prior weights
20% no benefit / 24% downside / 48% central / 8% favorable (range: fixed for illustration). A material null reflects missing current outcomes and inconclusive navigation evidence. The favorable case is low-weight because it combines several unobserved favorable assumptions. Subjective; not empirically calibrated.
Illustrative gross-resource multiplier
1.15× (range: 1.10× / 1.15× / 1.30×). Flags free state naloxone, partners and clinical referrals without claiming quantities, fair values or a lower bound. Very low; sensitivity only.
Bay Area impact share
95% (range: 85% / 95% / 99%). Documented program sites are in Alameda and Contra Costa Counties; beneficiary residence is not measured. Low; service-geography-informed prior.
San Francisco impact share, nested in Bay
2% (range: 0% / 2% / 5%). No SF site was found, but that does not exclude SF residents. Zero applies only to the downside assumption, not as a measured fact. Very low; assumed.
Opportunities per synthetic selected person-year
2 (range: 3 / 2 / 1.5). Unique people=20/50/137 opportunity equivalents divided by recurrence. The same factor is an assumed one-year Poisson rate, not inferred from reported successful survivors. Very low; subjective, not calibrated.
Common supported/post-support mortality hazard
.07/year (range: .10 / .07 / .04). Historical first-year overdose-survivor all-cause rates provide scale only. This is remaining mortality with the selected rescue component; counterfactual adds only the missing-rescue component. No source identifies a constant long-run Bay hazard. Very low; subjective, not calibrated.
Active support and follow-up
1 year active;10 years follow-up (range: 1 active;5 / 10 / 20 follow-up). One unique trajectory; after support both arms have the same hazard. Annual discount3%; utility.55/.65/.75, delay0. No repeated lifetime restart. Very low; subjective, not calibrated.
Baseline and extra rescue in selected component
0 baseline;1 increment (range: Fixed conditional component). Not all overdoses or absence of EMS. Old attribution.60/.75/.85 and outcomeAdditionality.25/.50/.75 remain outside the trajectory; budget deployability/realization are separate. Very low; subjective, not calibrated.
Retained-effective share of starts
.60 (range: .50 / .60 / .70). Unchanged prior; multiply starts before QALY per retained-effective engagement. Very low; subjective, not calibrated.

Illustrative $100,000 unrestricted gift

  • downside: $765,117,832 per10 Bay QALYs. Weight0.24; Bay giftQALYs0.0013069882294843534. Full cohort assumptions in API.
  • central: $15,341,409 per10 Bay QALYs. Weight0.48; Bay giftQALYs0.06518305922057498. Full cohort assumptions in API.
  • favorable-stress: $875,779 per10 Bay QALYs. Weight0.08; Bay giftQALYs1.1418402718573266. Full cohort assumptions in API.
  • No benefit: No finite positive price. 20% prior; all cost retained.
  • Old event-linear diagnostic: $11,112,457 per10 Bay QALYs. Preserved old version, not current estimate.
  • Recurrence correction; old severe mortality: $12,849,714 per10 Bay QALYs. Separates structural correction from revised survival assumptions.
  • 10× recurrence: $12,068,836 per10 Bay QALYs. Fixed opportunity totals; smaller synthetic cohort.
  • Two-year horizon: $16,547,195 per10 Bay QALYs. Finite horizon sensitivity.
  • Independent morbidity harm: $8,197,141 per10 Bay QALYs. .01Q harm per credited person; not a calibrated harm-world probability.
  • Naloxone null: $20,456,517 per10 Bay QALYs. Other pathways retained.
  • Adverse rescue increment: No finite positive price. Signed negative net total; not zero cost.

Uncertainty. Scenario values and weights are analyst priors, not confidence intervals, observed HEPPAC outcomes or empirically calibrated probabilities. The favorable stress case contributes most expected QALYs despite only 8% weight, so the headline is right-tail-sensitive. Zero deployability or zero outcomes returns zero QALYs and no finite cost ratio. The five-month Anderson study is never annualized. New supported survival assumptions separately increase health credit; they are not evidence that deduplication improves effectiveness. The exact old baseline and structural-only sensitivity must remain inspectable. Cross-program overlap is still not identified.

5. Funding and previous grants

Obtain a dated next-$100,000 budget, successor-contract status, current unduplicated people and events, treatment starts and retention, infection outcomes, drug-checking behavior, residence shares, and person-level cross-program linkage before recommending a gift on health cost-effectiveness.

RFMF is plausible but unverified. FY2025 net assets of $2.363M equal about 5.7 months of expense and the organization ran a $209K surplus. The November 14, 2025 county staff recommendation requested a $700K increase through August 31, 2026 primarily to maintain existing navigator staffing; without an executed amendment or later budget, it is a dated public-funding baseline rather than a verified gap. HEPPAC's donation page gives examples and says donations support services, but it does not price health-outcome capacity or commit the next $100,000. Verified marginal donor and gross costs remain unknown.

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. Donate. HEPPAC. Official organization donation page. Published: Undated/current; retrieved: 2026-09-10.
  2. HEPPAC home and service portfolio. HEPPAC. Official organization program page. Published: Undated/current; retrieved: 2026-09-10.
  3. OPEND: Overdose Prevention and Education. HEPPAC. Official organization program page. Published: Undated/current; retrieved: 2026-09-10.
  4. HEPPAC nonprofit filings, EIN 94-3205535. Internal Revenue Service, mirrored by ProPublica Nonprofit Explorer. IRS Form 990 data. Published: FY2025 filing; retrieved: 2026-09-10.
  5. HEPPAC FY2025 program-service descriptions. IRS filing extraction via Philanthropy.org. IRS Form 990 extraction. Published: FY2025 filing; retrieved: 2026-09-10.
  6. Effectiveness of Substance Use Navigation for Emergency Department Patients With Substance Use Disorders. Annals of Emergency Medicine / PubMed. Peer-reviewed observational implementation study. Published: 2022-11-17 online; 2023 print; retrieved: 2026-09-10.
  7. Peer Recovery Coach Intervention for Emergency Department Patients With Opioid Overdose. JAMA Network Open. Peer-reviewed randomized trial. Published: 2026-02-06; retrieved: 2026-09-10.
  8. Mortality risk during and after opioid substitution treatment: systematic review and meta-analysis of cohort studies. The BMJ. Peer-reviewed systematic review and meta-analysis. Published: 2017-04-26; retrieved: 2026-09-10.
  9. Effectiveness of naloxone in preventing fatalities from opioid overdose. BMC Public Health. Peer-reviewed systematic review. Published: 2025; retrieved: 2026-09-10.
  10. Evidence synthesis for syringe-service programs. US evidence review / NCBI Bookshelf. Government evidence review. Published: Current evidence synthesis; retrieved: 2026-09-10.
  11. Overdose Prevention Education and Naloxone outcome snapshot. Alameda County. Official county presentation. Published: Undated historical presentation; retrieved: 2026-09-10.
  12. Staff recommendation and proposed HEPPAC amendment, HCSA_397004. Alameda County Health Care Services Agency. Official county staff recommendation and proposed budget. Published: 2025-11-14 staff recommendation; 2025-12-16 agenda; retrieved: 2026-09-10.
  13. Drug checking services for people who use drugs: a systematic review. International Journal of Drug Policy / PubMed Central. Peer-reviewed systematic review. Published: 2022; retrieved: 2026-09-10.
  14. Naloxone Distribution Project. California Department of Health Care Services. Official state program page. Published: Undated/current; retrieved: 2026-09-10.
  15. Medication for Opioid Use Disorder After Nonfatal Opioid Overdose and Association With Mortality. Annals of Internal Medicine. Primary historical cohort; reused evidence. Published: 2018; retrieved: 2026-09-11.
  16. Causes of Death After Nonfatal Opioid Overdose. JAMA Psychiatry. Primary historical cohort. Published: 2018; retrieved: 2026-09-11.

Annual expenses: years and sources

Average annual expenses (three consecutive fiscal years): $3,524,429. Organization size is separate from the modeled cost-effectiveness of a donation.

Hiv Education And Prevention Project Of Alameda County

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

Original returns checked for legal identity, full-year period and total expenses. Latest three linked original returns.