Kent, WA
Harm Reduction Ban
24-Month Simulation

Affected PopulationsThis model covers persons who use drugs (PWUD) across two routes: injection (~1,600 default) and smoking (~4,800 default). KUOW/KC Public Health (2023) documents smoking as the majority route qualitatively; the 70% proportion and 3–5× ratio are modeled estimates, not published statistics. The PWSD count excludes recreational users exposed to fentanyl-contaminated non-opioid supply (cocaine, MDMA, counterfeit pills) — making these estimates conservative.

Test Strip Removal — OD ImpactFentanyl test strips reduce risk via behavior change (~40% of users who get a positive result change behavior — Goldman 2019; Krieger 2021). The 32% OD risk reduction is a modeled approximation from this behavior-change literature — not a single published per-event OD reduction rate. Ramp: 4 months.

Shared Pipe Infection MultipliersThe sharedPipeMult values (Constrained: 1.4×, Moderate: 2.1×, Severe: 3.2×) represent the increase in respiratory/oral infection rate when clean smoking supplies are unavailable and equipment sharing rises. These are modeled scenario gradients — no single peer-reviewed study provides per-scenario infection rate multipliers for fentanyl-era smoking supply removal. Basis: general infection risk literature documents 2–4× higher infection rates with shared equipment (Thibodeau et al. 2021, harm reduction nursing literature). The values used here are conservative within that range. Treat as directional estimates, not precise measurements.

Route Escalation8–15% of smokers transition to injection within 18 months when supplies removed — derived inference from Vancouver cohort literature (Ti 2011; Small 2013), not a direct experimental finding. These new injectors are added to totalInjectionPop for HIV/HCV calculations.

Endocarditis — Proxy RationaleEndocarditis risk among people who inject drugs is driven by the same behaviors that transmit HCV: frequency of injection, equipment sharing, and injection technique. The model uses the HCV excess multiplier as a proxy for endocarditis risk increase — a defensible approximation because both outcomes respond to the same behavioral drivers. This is a stated modeling convenience; no published study directly measures endocarditis rates as a function of SSP closure. Excess formula: totalInjectionPop × 0.5% × (hcvMult × adj − 1) × 2.

HIV / HCV — Excess Cases OnlyKPI and cost figures show ban-attributable excess cases — cases above the baseline rate that would occur regardless. Formula: population × baseRate × (multiplier − 1) × 2 years. This is methodologically conservative and correctly isolates ban causation. Baseline HIV: 0.3%/yr (optimistic, well-functioning SSP context; CDC range without SSP: 1–3%/yr).

Equity and Population Gaps — DisclosedThe model does not stratify by race/ethnicity: KC data shows OD mortality ~30% higher for Black residents and ~5.5× higher for Native American/Alaska Native residents — a ban's disproportionate impact on these communities is not captured. Unhoused PWUD cannot effectively displace to adjacent cities — the displacement metric applies only to those with housing/mobility. Xylazine ("tranq") co-contamination is growing in KC; wounds require wound care beyond OD response, and naloxone does not reverse xylazine — this model does not model xylazine-specific harms.

Community-Wide Impacts — Families, Children, BusinessesEMS dispatch ($2,400/call); ME costs ($3,800/death); secondary HIV to non-using partners (~8% of excess cases); HCV spillover to general population (~3% of excess cases via healthcare/household routes — Alter et al., CDC HCV surveillance); child welfare/CPS ($18K/case, applied to 0.495 children per OD death per Jones et al. 2024 × a modeled 40% CPS involvement rate — see source 19; the "25% of decedents" figure from prior versions has been withdrawn as unverifiable); foster care placement (modeled 40% removal rate, $35K/yr × 2yr avg — AFCARS 2023); non-fatal OD child exposure (~15% of non-fatal events with child present, $4,200/CPS investigation); small business disruption ($8,500/death, modeled estimate — no single published source, disclosed as approximate); workforce loss ($48K = ~1yr individual median income, conservative). Mortality valuation is a deliberate direct-cost floor — see source 20 for why the $1.16M "Florence 2021 VSL" figure used in prior versions was withdrawn (wrong paper, wrong method). The HCV spillover line reflects the key point that infectious disease does not stay contained within drug-using populations — it reaches families, healthcare workers, and the general community.

The neighbouring-city contradiction (read with the companion model)These two simulations were built separately and, read together, they contradict each other in a way that is itself the finding. The Tukwila camping ban model exports displaced people to Burien, SeaTac, Renton, Kent, and Seattle. The Kent harm reduction ban model exports displaced people to Renton, Auburn, SeaTac — and Tukwila is in the same corridor. Each model assumes the neighbour absorbs. Post–Grants Pass, south King County jurisdictions are moving on ordinances in parallel, which means the assumption fails in every direction at once: if Tukwila, Kent, Burien and Renton all enforce, nobody absorbs. The output is not exit, it is circulation — the same people moved repeatedly around a five-city loop, each jurisdiction booking a local reduction, none producing a housing outcome, and the regional total unchanged or worse. A displacement slider therefore measures cost transfer, not cost avoidance, and the regional total is invariant to it. Any city citing either model to argue that people will "go somewhere else" is citing precisely the mechanism that makes the regional number worse — and is relying on a neighbour making the opposite choice.

Do not add these two models togetherThey share a denominator overlap and summing them double-counts real people. King County recorded 1,338 overdose deaths in 2023; the King County Medical Examiner separately recorded 415 deaths of people experiencing homelessness that year — and a large share of those 415 are overdose deaths already inside the 1,338. The Kent model is denominated in overdose deaths among people who use drugs; the Tukwila model is denominated in mortality among unsheltered people. A person who is both — and in south King County many are — sits in both denominators. Each model is internally valid and each is defensible on its own. Neither this nor the companion figure may be added into a single regional total without de-duplicating that overlap first. This is disclosed here rather than left for an opponent to discover.

Cost AccountingHIV = lifetime ART liability ($420K, conservative; current range $430–550K inflation-adjusted). HCV = generic DAA course ($24K; Medicaid negotiated rates may be lower). OD: fatal costs use direct-costs-only basis ($31.5K total: $27K direct + $4.5K hosp), not VSL. Non-fatal OD: $11K/hospitalization (R Street Institute). Endocarditis: $85K/case (post-2020 literature midpoint). All primarily Medicaid- and tax-funded.