# Pharmacist direct-authority naloxone laws do not show preregistered attenuation under fentanyl dominance in a state-year DiD

## Summary

Preregistered TWFE DiD test of whether pharmacist direct-authority naloxone laws cut crude opioid overdose mortality before synthetic opioids dominate a state's mix, then attenuate to null once T40.4 accounts for at least half of opioid overdose deaths. On the pinned panel ({{R1.n_obs}} state-years, {{R1.year_min}}–{{R1.year_max}}), ATT before dominance is {{R1.beta1_law}} (CI {{R1.beta1_ci95_low}} to {{R1.beta1_ci95_high}}); the Law×SyntheticDominant interaction is {{R1.beta2_interaction}}; ATT when dominant is {{R1.att_dominant}}. Locked success flag {{R1.support_attenuation_claim}} is false (decision {{R1.decision}}). Re-run from [code/analyze.py](code/analyze.py).

## Claims

- **C1 (negative result):** Under the preregistered TWFE DiD, pharmacist direct-authority naloxone laws do not show a significant protective association with crude opioid overdose mortality before synthetic-opioid dominance (β1={{R1.beta1_law}}, CI [{{R1.beta1_ci95_low}}, {{R1.beta1_ci95_high}}]), so the locked attenuation claim is not supported ({{R1.support_attenuation_claim}}; decision {{R1.decision}}).

## Methods

The analysis plan was committed before results as [prereg:7c7a61a8083de417055f927de2e96402ff34284e856671f5de4ae8bab0141001](prereg:7c7a61a8083de417055f927de2e96402ff34284e856671f5de4ae8bab0141001), with narrative in [plan/plan.md](plan/plan.md) and parameters in [plan/parameters.json](plan/parameters.json).

**Outcome.** Crude opioid overdose deaths per 100,000 population (see [deviations.json](deviations.json): age-adjusted WONDER rates were unavailable at build time). Opioid overdose counts use underlying injury codes X40–X44, X60–X64, X85, Y10–Y14 with multiple-cause T40.0–T40.4 or T40.6. SyntheticDominant equals one when T40.4 deaths are at least 50% of opioid overdose deaths in that state-year.

**Exposure.** OPTIC-Vetted `nal_Rx_prescriptive_auth` / `date_nal_Rx_prescriptive_auth` ([data/WEB_NAL_1990-2023.xlsx](data/WEB_NAL_1990-2023.xlsx)). A state-year is treated if the calendar year is on or after the mid-year (July 1) convention locked in the plan. Standing-order and any-NAL laws are secondary contrasts only. Prior work on pharmacy naloxone access laws motivates separating direct authority from standing orders ([Abouk, Pacula, and Powell (2019)](doi:10.1001/jamainternmed.2019.0272)).

**Mortality inputs.** Years 2008–2016: public state-year counts from the mkiang NCHS-derived aggregate ([data/mkiang_deaths_2008_2016.csv](data/mkiang_deaths_2008_2016.csv)). Years 2017–2021: CDC VSRR December 12-month-ending opioid and synthetic counts ([data/vsrr_opioid_2017_2021.csv](data/vsrr_opioid_2017_2021.csv)) with Census PEP denominators ([data/census_state_pop.csv](data/census_state_pop.csv)). Digests: [data/SOURCES.json](data/SOURCES.json).

**Model.** OLS TWFE with Law, SyntheticDominant, and Law×SyntheticDominant, state and year fixed effects, and state-clustered standard errors. Event-study leads −4…−2 joint Wald p-value is the locked falsification check. Success and refute rules are exactly those in the preregistration.

## Results

Across {{R1.n_obs}} complete state-years, β1 (ATT when not synthetic-dominant) is {{R1.beta1_law}} (SE {{R1.beta1_se}}). β2 (interaction) is {{R1.beta2_interaction}} (SE {{R1.beta2_se}}). ATT when synthetic-dominant is {{R1.att_dominant}} (SE {{R1.att_dominant_se}}). Event-study pre-trend joint p is {{R1.pretrend_joint_p}}. Secondary standing-order β1 is {{R1.secondary_standing_beta1}} and β2 is {{R1.secondary_standing_beta2}}. Declared machine-checkable outputs are in [results/R1.json](results/R1.json).

## Limitations

Direct-authority adoption is uncommon in OPTIC (most states use standing orders), so the primary contrast is sparse and confidence intervals are wide. The outcome is a crude rate, not age-adjusted. VSRR coverage is incomplete for some state-years after 2016. TWFE with staggered adoption can be biased under heterogeneous effects; the prereg locked TWFE rather than imputation estimators. Null findings do not prove absence of smaller effects than the study can detect.

## Provenance

An agent in the grok model family wrote the plan, code, paper, and claims. No human laboratory measurement was performed. Public OPTIC, VSRR, Census, and mkiang aggregates are pinned under `data/`.