Sign Up
Login
Try for Free
Login
The CIWA-Ar (Clinical Institute Withdrawal Assessment for Alcohol, revised) is a 10-item alcohol-withdrawal severity scale published by Sullivan and colleagues in 1989 and scored 0 to 67. Hospital, emergency, and detoxification teams use it for serial monitoring under symptom-triggered and fixed-schedule protocols. This page covers how to document CIWA-Ar assessments defensibly, with a fictional sample note.
Nurse-administered in most settings under protocol; combines observation with substantial patient self-report
Hospital and detox nursing teams, hospitalists, addiction medicine, ED and ICU clinicians, order-set committees, auditors
One flowsheet row per assessment plus a decision note · administration about 2 to 5 minutes
Nurse-administered withdrawal scale (10 items: nine scored 0 to 7, one 0 to 4, total 67)
Serial severity monitoring after alcohol withdrawal is diagnosed, symptom-triggered medication orders, escalation, protocol discontinuation
The 1989 appendix grants free use of the scale; every threshold, interval, and stop rule is a guideline or order-set convention
The Clinical Institute Withdrawal Assessment for Alcohol, revised (CIWA-Ar) is a 10-item, clinician-administered measure of current alcohol-withdrawal severity, published by Sullivan, Sykora, Schneiderman, Naranjo, and Sellers in the British Journal of Addiction in 1989. Nine items are scored 0 to 7 and one, orientation and clouding of sensorium, is scored 0 to 4, so the maximum total is 67. The design history explains what the number can and cannot say. The authors selected a ten-item subset that efficiently reproduced the older 15-item CIWA-A total and validated it against that predecessor scale, not against seizures, delirium, ICU transfer, or death. Seizure was considered as an item and rejected (two events among 137 source observations, with timing unsuited to repeated scoring), blood pressure was tested and excluded, and pulse was recorded on the form but never scored: vital signs live outside the total by design.
Two distinctions carry the documentation weight. First, licensing: the 1989 appendix states the scale "is not copyrighted and may be used freely," which is why it can sit in order sets and EHRs without a license, a permission the copyrighted CIWA-AD derivative and most adjacent instruments do not share. Second, the score is the instrument and everything else is a protocol. Severity bands, medication thresholds, reassessment intervals, and stop rules come from guidelines and local order sets, and they legitimately differ. The scale also assumes a patient who can participate: it measures apparent severity after alcohol withdrawal has been clinically identified, and it does not diagnose withdrawal (that lives in the substance use assessment), predict who will later seize or develop delirium, or function in delirious, intubated, or non-communicative patients. Its opioid neighbor, the COWS, is a different instrument for a different syndrome; the scores are never comparable or interchangeable.
Hospital medicine and detoxification units run serial CIWA-Ar scores under symptom-triggered or fixed-schedule order sets, and nurses administer and chart most assessments, which raises the documentation stakes: under symptom-triggered orders the recorded score directly triggers medication. Emergency departments use it while the withdrawal diagnosis and disposition take shape. ICUs are where it is most often misapplied: ASAM recommends against CIWA-Ar for alcohol-withdrawal delirium because the scale depends on patient report, pointing instead to CAM-ICU, RASS, and MINDS-type instruments, and an ICU cohort study found CIWA-Ar documented in 56% of alcohol-withdrawal patients against 94% for RASS. In Canada, Alberta Health Services publishes score-linked intervals and notification triggers, while the 2023 CRISM national guideline builds risk assessment around PAWSS and clinical parameters rather than any CIWA number. In Australia the instrument itself must be verified before charting: Western Australia operationalizes a CIWA-Ar chart, while Queensland leans on the locally developed Alcohol Withdrawal Scale, so an Australian withdrawal chart may not be a CIWA-Ar at all. Medication decisions land in the medication management note, deterioration escalates through a crisis note, and the withdrawal course is summarized in the discharge summary.
No statute prescribes a CIWA-Ar note format, and no law or accreditor sets a CIWA-Ar threshold or interval. What survives review is a record that makes the decision trace auditable: a valid assessment, the ordered criterion, the medication or the documented hold, the reassessment, and the response. Each element below carries the pitfall that most often undermines it.
A time-stamped total with its component values. Record the actual bedside-assessment time (label late entries; the EHR entry time is not the assessment time), the total over 67, and all ten component values, ideally in a flowsheet, under the name CIWA-Ar with the active order set identified. Pitfall: a bare total. In a 2025 handoff study, unassessable components made a valid total impossible to calculate in 34.2% of paired assessments, and inter-rater correlation was 0.68; the components and observations are what make the number defensible.
The two validity gates. Chart documented recent alcohol use (and how it is known) and that the patient was alert, communicative, and able to understand the questions, with interpreter details when language matters. Pitfall: skipping them. In a Mayo Clinic Proceedings review only 48% of patients on symptom-triggered therapy met both criteria, and a later sample found 20% with no documented recent alcohol use and 14% unable to communicate when the protocol was ordered.
Separate objective observations. Blood pressure, pulse, respiratory rate, oxygen saturation, temperature, and sedation or arousability are recorded alongside, never inside, the score: the developers tested vital signs and left them out of the total. Pitfall: a falling total standing in for safety. Oversedation and respiratory depression develop outside the score, and ASAM requires monitoring them separately.
Confounders and attribution. Say which component may be unreliable and why: pain, delirium from another cause, head injury, postictal states, baseline tremor, psychiatric symptoms, intoxication, or a language barrier. Pitfall: zero-filling unanswerable items or nudging a component toward a desired dose. A subtotal with silent zeros is not a valid CIWA-Ar; chart "not interpretable as a valid total" with the reason and escalate instead.
The named protocol and its exact criterion. Name the order set with its version or date and paraphrase the band actually met. Hospitals legitimately trigger at 8, 10, or 12, or use a non-CIWA pathway; ASAM offers bands only as examples and states that severity classification is ultimately clinician judgment. Pitfall: "CIWA above 8 requires benzodiazepines" charted as a universal rule. No US, Canadian, or Australian statute or regulation sets any CIWA-Ar threshold.
The medication or the documented hold. Drug, dose, route, and exact administration time tied to the criterion met; or the hold parameter, contraindication, sedation, diagnostic concern, refusal, or provider instruction, with who was notified, when, and the response. Pitfall: "Ativan given for CIWA," or a score-linked dose silently not given. Both break the trace between score and action.
Reassessment against the ordered interval. When the reassessment was due, when it actually happened, the repeat total or the alternative assessment, the response, and adverse effects. Intervals are protocol property: ASAM describes every 1 to 4 hours initially for moderate, severe, or treated withdrawal, and Alberta, Western Australia, and New South Wales each publish different score-linked ladders. Pitfall: "serial CIWA per protocol" with no clock times, or an interval remembered from a different hospital.
Escalation, discontinuation, and handoff. Notification triggers with times, persons, and orders; the exact local stop criterion (Western Australia, for example, requires four consecutive totals of 8 or less plus at least 48 hours since the last drink); any switch to CAM-ICU, RASS, or a MINDS-type instrument with the reason; and a handoff carrying the last valid total, trend, last dose and cumulative exposure, next due time, and unresolved confounders. Pitfall: quietly stopping assessments, or continuing totals after the patient can no longer participate.
CIWA-AR DOCUMENTATION BLOCK Date: [ ] Actual assessment time: [ ] (label late entries) Setting: [ ] Assessor: [ ] Instrument: CIWA-Ar Order set: [name + version/date] Validity gates: [recent alcohol use + how known] [alert, communicative, understands questions: y/n] [interpreter service + language, if used] Total: [ ]/67 Components: [all 10 values in the flowsheet] Separate observations: [BP / pulse / RR / SpO2 / temp / sedation-arousability] Confounders + attribution: [affected items, why, action] Criterion: [ordered band, paraphrased] Met: [y/n] Medication: [drug, dose, route, exact time] OR Hold / variance: [reason; provider notified + time + response] Reassessment: [due -> done; repeat total; response; adverse effects] Escalation: [trigger, person notified, time, orders received] Stop / transition: [local stop criterion met, or switched to: instrument + reason] Handoff: [last valid total + trend, last dose + cumulative, next due, notification triggers] Clinician signature / credentials: Date:
Free to use and share, no signup. The PDF includes a one-page cheat sheet with element-by-element pitfalls and a pre-sign checklist; the DOCX is the blank documentation block, ready to adapt. Neither reproduces the scale itself.
Scenario: uncomplicated moderate withdrawal on a symptom-triggered order set, from first assessment through step-down and handoff. The order set, bands, and doses are invented to illustrate documentation structure; they are not treatment instructions. All details are fictional.
Patient: M.H., 48 · Setting: Medical unit, symptom-triggered order set · Clinician: R. Osei, RN / L. Grant, MD · Note date: 08/13/2026
08:00, assessment: Last drink approximately 14 hours ago per patient, corroborated by spouse. Alert, oriented, conversant in English, answers internally consistent. CIWA-Ar 12/67: tremor 3, paroxysmal sweats 2, anxiety 3, nausea 1, headache 2, agitation 1; perceptual items and disorientation 0; all ten component values in the flowsheet. Recorded separately: BP 148/92, pulse 104, RR 16, SpO2 98% on room air, afebrile, no sedation. The unit's Alcohol Withdrawal Order Set (revision April 2026, fictional) specifies lorazepam 2 mg PO for a valid total of 10 to 15: criterion met, dose administered 08:05, provider aware. Reassessment due in one hour per the order set.
09:05, reassessment (60 minutes post-dose): Easily arousable, no respiratory depression, RR 15. CIWA-Ar 6/67 (tremor 2, sweats 1, anxiety 2, headache 1). Below the ordered medication bands; no dose indicated. Baseline essential tremor considered and excluded: no tremor history per patient and chart.
13:05: CIWA-Ar 4/67, patient comfortable, tolerating oral fluids. Interval extended to every four hours per the order set's step-down rule; vital signs and sedation checks continue on the unit schedule.
19:00, handoff: Scores 12, 6, 4 today; last lorazepam 2 mg PO at 08:05 (cumulative 2 mg); next assessment due 21:05; both validity gates intact all shift; no active confounders. Notify provider for a total of 20 or higher, seizure, new disorientation or hallucinations, unstable vital signs, or oversedation.
This sample is fictional and for educational purposes. It does not describe a real patient or record; the order set, bands, doses, and scores are invented to show documentation structure and are not clinical guidance.
Writing these after every session? BastionGPT drafts complete notes from bullets, dictation, or a transcript.
Generate a note from bulletsUnited States: under 42 CFR 482.24, hospital records must be complete, dated, timed, and authenticated, and standing orders and protocols must be adopted and periodically reviewed through medical, nursing, and pharmacy leadership (LAW), but nothing in it names a CIWA-Ar cutoff or interval. Federal nursing conditions require medication administration to follow practitioner orders and hospital policy (LAW), with scope details left to state nurse-practice acts. No accreditor-specific CIWA-Ar rule was verified: accreditation exposure runs through order-set currency, staff competency, administration per orders, reassessment records, and deterioration response (ACCREDITATION REQUIREMENT), so a specific threshold should never be labeled as accreditor-required without a citable current standard. No standalone CIWA-Ar CPT code or national payer threshold exists; the score supports severity and medical-necessity documentation but is not itself billable (PAYER POLICY). ASAM's 2020 guideline prefers symptom-triggered treatment when trained staff can reassess reliably and describes monitoring every 1 to 4 hours initially, relaxing to every 4 to 8 after sustained stability (CONVENTION). The oversight record is concrete: a 2025 VA Office of Inspector General inspection found inaccurate and delayed scoring and protocol-inconsistent medication administration in a fatal case, a 2024 national review found local-guidance gaps across 30 systems, and a settled federal suit alleged scores assigned without the underlying assessments (an allegation resolved by settlement, not an adjudicated finding).
Canada and Australia run the same pattern with different anchors. The 2023 CRISM guideline recommends PAWSS plus clinical parameters for severe-withdrawal risk and non-benzodiazepine options for lower-risk patients (CONVENTION); Alberta Health Services publishes a protocol using every-four-hour scoring for totals of 0 to 9, hourly scoring at 10 or higher until three consecutive totals are below 10, provider notification after two scores of 20 or higher, and reconsideration of the protocol after five days (Alberta CONVENTION, not national law); and British Columbia's revised nursing documentation standards, effective April 2026, require timely, chronological records with actual times and labeled late entries (PROFESSIONAL REGULATION). Australia has no single national workflow: New South Wales guidance, mandatory in NSW Health facilities, describes roughly four-to-six-hourly assessment below 10, two-to-four-hourly from 10 to 20, and hourly above 20; Western Australia's chart uses four-hourly for 0 to 8, two-hourly for 9 to 14, and hourly at 15 or higher, with discontinuation after four consecutive totals of 8 or less and at least 48 hours since the last drink; Queensland leans on the Alcohol Withdrawal Scale, whose validation was never formally published. A 2025 New South Wales inquest into a custody death (an AWS case, not CIWA-Ar) recommended that the withdrawal-monitoring form itself carry observation-frequency guidance and a field for an individualized plan, documentation-design evidence any jurisdiction can learn from.
The psychometrics counsel the same humility as the law. The 1989 validation reproduced the predecessor CIWA-A total rather than predicting outcomes; a 2025 handoff study found unassessable components blocking a valid total in 34.2% of paired assessments, with inter-rater correlation of 0.68; a 2019 analysis of 479 acutely ill and injured inpatients did not support the scale's reliability or validity in those populations; and an ICU cohort showed CIWA-Ar documented in only 56% of alcohol-withdrawal patients. A 1988 study associated totals above 15 with higher risk of severe withdrawal (relative risk 3.72), an association, not a validated treatment cutoff. On rights: the 1989 appendix grants free use of the scale itself; that permission does not extend to the journal article's typesetting, commercial repackagings, or derivatives, and the CIWA-AD adaptation is copyrighted. BastionGPT is not affiliated with the instrument authors or the journal. This page describes the instrument in original prose and reproduces no item text or scoring anchors.
The misuse literature is unusually quantified. In a Mayo Clinic Proceedings review, only 48% of patients on symptom-triggered therapy met both validity criteria (recent alcohol use and intact verbal communication); a later general-hospital sample found 20% with no documented recent alcohol use, 57% with one or zero documented withdrawal risk factors, 14% unable to communicate, and benzodiazepine-associated adverse events in 15%. A 2025 handoff study could not compute a valid total in 34.2% of paired assessments. And a 2025 VA Office of Inspector General inspection connected a death to exactly these records: inaccurate and delayed scoring, and medication administration inconsistent with the facility's own CIWA-Ar protocol. The BastionGPT Clinical Advisory Board sees the same errors most often in CIWA-Ar documentation reviews:
BastionGPT is specifically trained, tuned, and clinically tested on behavioral health progress notes and screening documentation.
See how clinicians use it day to day on the AI therapy notes page.
Many BastionGPT users report saving more than 90 minutes per day on documentation.
HIPAA-compliant with a signed BAA on every plan. Your data is never used to train models. BastionGPT drafts, you review and sign.
Ten items: nine scored 0 to 7 and one (orientation and clouding of sensorium) scored 0 to 4, for a maximum of 67. The total estimates the apparent severity of alcohol withdrawal at one moment, after the diagnosis has been made clinically. The bands attached to it are conventions that vary by source: ASAM's examples are below 10 mild, 10 to 18 moderate, and 19 or higher severe, offered with the explicit caveat that classification is ultimately clinician judgment. The score does not diagnose withdrawal, predict who will later seize or develop delirium, or contain any vital sign; it was validated by reproducing the older CIWA-A total, not by predicting outcomes. Cite whose bands you are using, or better, the order set actually in force.
None, by law: no US, Canadian, or Australian statute or regulation sets a CIWA-Ar medication threshold. The instrument's own appendix says only that patients scoring below 10 do not usually need additional withdrawal medication. Everything firmer is a protocol convention: one hospital's order set starts at 8, another at 10 or 12, and some units run non-CIWA pathways entirely. The randomized evidence (median 9 versus 68 hours of treatment and 100 versus 425 mg of chlordiazepoxide in the 1994 trial; 39% versus 100% of patients medicated in the 2002 trial) supports symptom-triggered dosing in communicative, selected patients, not automatic medication from any particular number. Chart the named order set, its version, and the band actually met.
The instrument sets no interval; the order or protocol does. ASAM describes every 1 to 4 hours initially for moderate, severe, or pharmacologically treated withdrawal, relaxing to every 4 to 8 hours after sustained stability. Alberta's protocol uses every four hours for totals of 0 to 9 and hourly at 10 or higher until three consecutive totals fall below 10; Western Australia uses four-hourly for 0 to 8, two-hourly for 9 to 14, and hourly at 15 or higher; New South Wales describes roughly four-to-six-hourly below 10, two-to-four-hourly from 10 to 20, and hourly above 20. Chart the interval actually ordered, the actual reassessment times, and the reason for any delay, never a ladder remembered from a different hospital.
Local policy decides, and the chart should say what happened either way. Sleep is not a substitute for assessing the ten domains, and unanswered subjective items should never be silently scored zero; some services (Western Australian nursing education, for example) instruct staff to wake the patient for a scheduled assessment, which is a local rule rather than a property of the instrument. Whatever the policy, document arousability, respiratory status and sedation, what was genuinely observable while the patient slept, and why a full assessment was or was not completed, with the provider instruction when one was given.
Treat the mismatch as clinical data and never adjust a component to fix it. Verify the history and timing, repeat the assessment when appropriate, name the components affected by confounders (pain, delirium from another cause, head injury, baseline tremor, psychiatric symptoms, intoxication, language barriers), and use collateral information or an interpreter. If a score-linked instruction conflicts with what you see, a drowsy patient the number says to medicate, follow the ordered hold parameters, notify the prescriber, and document who was notified, when, the response, and the revised plan. Chart observable inconsistency neutrally; "patient exaggerating" without objective support is the phrasing that fails review.
The scale needs two validity gates: documented recent alcohol use and a patient who can communicate and understand the questions. A Mayo Clinic Proceedings review found only 48% of patients on symptom-triggered therapy met both. It is unreliable or invalid in delirium, intubation, deep sedation, postictal states, significant cognitive impairment, and unmanaged language barriers, and ASAM specifically recommends against CIWA-Ar for alcohol-withdrawal delirium because it depends on patient report, pointing to CAM-ICU, RASS, the Delirium Detection Score, or MINDS-type instruments instead. Document the switch explicitly: the last valid total, why the scale became invalid, the replacement monitoring plan, and the provider order behind it, rather than a bare "CIWA inaccurate."
Against the local stop criterion, because no universal one exists. Western Australia requires at least 48 hours since the last drink plus four consecutive totals of 8 or less; Alberta scores hourly until three consecutive totals fall below 10 and asks whether the protocol is still needed after five days; ASAM allows reduced monitoring after sustained stability. Chart the exact criterion met, the last valid score and trend, cumulative sedative exposure, the order or protocol authority for stopping, what monitoring continues, and the handoff. A short symptom-triggered benzodiazepine course does not automatically require a prolonged taper; that is an individualized prescriber decision, documented as such.
The 1989 appendix states the scale "is not copyrighted and may be used freely," which is why it can be embedded in order sets and EHRs without a license, and it is the exception among withdrawal instruments. The permission covers the scale itself, not the journal article's typesetting or commentary, not commercial repackagings, and not derivatives: the CIWA-AD adaptation is copyrighted, and adjacent instruments (PAWSS, the Australian AWS, SEWS, COWS) carry their own rights and should be described, not reproduced. A locally modified form should be identified by organization and version rather than treated as psychometrically identical to the original.
Yes. Give it the assessment facts (times, totals and components, the validity gates, separate vitals and sedation, the named order set and band, doses and times, confounders) and it drafts the full trace: assessment blocks, criterion reasoning, hold documentation with notification, serial response, escalation, and the handoff, ready for your review. It can also check a finished note for thresholds charted as requirements, missing validity gates, zero-filled items, bare totals, and missing clock times. BastionGPT is HIPAA-compliant with a signed BAA on every plan, and your data is never used to train models.
The instrument facts and compliance claims on this page trace to these sources, last verified August 2026:
Educational content, not legal or billing advice. Sample notes are fictional. Follow your organization's policies and your board, payer, and jurisdiction requirements.