Acute alcohol withdrawal — ASAM 2020 + CIWA-Ar symptom-triggered protocol. New dossier authored 2026-05-13 in B.8 re-dispatch. CIWA-Ar bands: mild <10 / moderate 10–19 / severe ≥20. Step 0 universal: thiamine 500 mg IV q8h × 3 d BEFORE glucose; folate, MVI, K, Mg, Phos repletion. Step 1 mild (outpatient candidate): chlordiazepoxide taper or gabapentin (ASAM 2020 option). Step 2 moderate: CIWA-triggered diazepam (or lorazepam if cirrhosis); symptom-triggered uses less total BZD. Step 3 severe (ICU): front-load BZD; phenobarbital adjunct (Rosenson 2013). Step 4 seizure/DT: lorazepam IV stat 4 mg; phenobarbital load if status; do NOT use phenytoin alone for alcohol-withdrawal seizures (different mechanism). Step 5 AUD MAT bridge at discharge: naltrexone (po or XR-IM), acamprosate (renal-dosed), disulfiram (motivated patient + supervised admin), gabapentin (anxiety/insomnia residual). BZD selection in cirrhosis: lorazepam or oxazepam (non-CYP glucuronidation); AVOID long-acting diazepam/chlordiazepoxide. Wernicke triad (ophthalmoplegia + ataxia + confusion) → 500 mg IV thiamine × 5 d; progresses to Korsakoff if untreated. PRODUCTION blockers: (1) calc.ciwa_ar / calc.pawss registry binding pending, (2) RxCUIs need RxNav validation, (3) PMIDs need PubMed verification, (4) registry import not added (forbidden), (5) test file pending. HUMAN REVIEW: PAWSS-specific PMID and 2024 ASAM update verification deferred; clinical thresholds match ASAM 2020 and standard ICU practice. Deepened 2026-05-14 (shard-5-obped-id depth-pass-1): added co-located _research-bundles/psych.alcohol_withdrawal.core.v1.md + companion _briefs/psych.alcohol_withdrawal.core.v1.depth.md. Added outpatient AUD MAT continuation setting playbook (naltrexone / acamprosate first-line per ASAM 2020 hierarchy + disulfiram second-line + off-label gabapentin / topiramate; Brief Intervention + Motivational Interviewing; AA / SMART Recovery referral; follow-up at 1/2/4 weeks; LFT + CBC at 2-4 weeks; PHQ-9 + C-SSRS surveillance at EVERY visit because SI risk peaks 2-12 weeks post-withdrawal during post-acute withdrawal anhedonia / protracted GABA dysregulation; lethal-means counseling for any SI history; family / support involvement). Added severity triggers: delirium_tremens_present (life-threatening; complements existing delirium_tremens row by being behaviourally explicit + PAWSS ≥ 4 admission predictor + adjunct phenobarbital OR dexmedetomidine choice rationale), withdrawal_seizure_first_episode (severe; first-episode uncomplicated GTC 12-48 h post-cessation with 24-h observation requirement + status routing edge to neuro.status-epilepticus.core.v1), kindling_history_or_dt_history (severe; prior DT or seizure history = strongest predictor of current-admission complication LR+ ≈ 5-10; admit even with mild-moderate CIWA-Ar; aggressive front-loading), and wernicke_features_present (severe; codifies the urgent thiamine-before-glucose rule even for incomplete-triad presentations — complements life-threatening wernicke_triad row). Appended verified PMIDs 9244334 (Mayo-Smith JAMA 1997 benzodiazepine guideline + seizure / DT pharmacology) and 22834916 (Rosenson 2013 phenobarbital ED RCT — lowered ICU admission 8% vs 25%), bringing dossier.evidence.pmids from 2 to 4 verified anchors (Maldonado PAWSS, Cook 1998 Wernicke, Saitz JAMA 1994, COMBINE 2006, Hack 2006, Mason 2014, Hendey 2011, Sachdeva 2015, Awissi 2013 remain NEEDS_SOURCE_REVIEW pending Stage-A verification — not added per verification rule). Phenotype matrix (CIWA-Ar band × DT-history × age × comorbid hepatic / cardiac / seizure-history / polysubstance / pregnancy / psychiatric) is encoded indirectly via severity_triggers (ciwa_ge_20, withdrawal_seizure, delirium_tremens, delirium_tremens_present, withdrawal_seizure_first_episode, kindling_history_or_dt_history, bzd_refractory_withdrawal, autonomic_storm, wernicke_triad, wernicke_features_present) and via sibling_differentiation routing (psych.opioid_use_disorder.core.v1, psych.suicidality.ed.core.v1, gi.cirrhosis.core.v1) and via per-setting playbook drug logic (BZD selection in cirrhosis = lorazepam/oxazepam; avoid disulfiram in pregnancy/CAD; thiamine universal). First-class TS field for phenotype matrix is schema-blocked — see id.sepsis.core.v1 brief Schema-blocked queue. Bayesian linkage (CIWA-Ar band LRs: ≤8 LR− ≈ 0.3-0.5; 9-15 LR+ ≈ 1.5-2; 16-19 LR+ ≈ 3-4; ≥20 LR+ ≈ 8-15 for DT progression; PAWSS ≥ 4 LR+ ≈ 6 for complicated withdrawal anticipated per Maldonado 2014; prior DT/seizure history LR+ ≈ 5-10 — kindling phenomenon; T_admit = CIWA-Ar ≥ 15 OR comorbid medical / age > 65 / DT history / autonomic instability OR PAWSS ≥ 4; T_discharge = CIWA-Ar ≤ 8 sustained × 24 h + ambulating + tolerating PO + mental status at baseline + electrolytes stable + Brief Intervention + AUD MAT initiated + outpatient appointment confirmed; cross-dossier routing to psych.suicidality.ed.core.v1 for SI screen during withdrawal — high comorbidity, ~2-3× completed-suicide risk; gi.cirrhosis.core.v1 if hepatic decompensation; nutrient-replacement workup if Wernicke features; psych.opioid_use_disorder.core.v1 if polysubstance opioid use; neuro.status-epilepticus.core.v1 if status develops) is documented in the co-located _research-bundles/psych.alcohol_withdrawal.core.v1.md. ROS/DDx LR seed data audited by npm run audit:ros-ddx-coverage (cross-cutting; not touched by this shard). Prehospital recognition (EMS IM/IV BZD for witnessed alcohol-withdrawal seizure en route — RAMPART Silbergleit NEJM 2012 supports IM midazolam 10 mg for prehospital status; alcohol-withdrawal seizures follow same pathway) is encoded indirectly via the ED-setting playbook escalation entry; a first-class "prehospital" DossierSetting value is schema-blocked, and the transitions[] array (admit/escalation/de-escalation pattern from sepsis dossier) was not authored in this pass — deferred. Kindling phenomenon — each subsequent alcohol withdrawal is more severe than the prior episode, biologically driven by repeated GABA-A receptor down-regulation + NMDA receptor up-regulation cycling. Prior DT or withdrawal-seizure history is the single strongest predictor of complicated current-admission withdrawal (LR+ ≈ 5-10). Motivates the kindling_history_or_dt_history severity trigger (admit even with mild-moderate CIWA-Ar), the aggressive front-loading IV BZD ± phenobarbital adjunct in patients with kindling history, the outpatient setting playbook emphasis on SI surveillance at 2-12 weeks (post-acute withdrawal anhedonia / protracted GABA dysregulation period), and the AUD MAT initiation before discharge (every subsequent withdrawal compounds the kindling risk).