CIWA-Ar on shift: when the scale misleads

It is three in the morning and nursing calls: the man in his fifties admitted for pneumonia is tremulous, sweaty and asking for his wife, who has already left. The order says "CIWA-Ar every four hours". The nurse applies the scale, adds up the items and reaches a high number. The course of action seems obvious. Except he is also more drowsy than he was in the afternoon, glucose has not been measured since admission and nobody knows for sure when his last drink was. The scale answered the question that was asked of it. The clinical question was another one.

NEXO is a daily medical diagnosis game: one clinical case a day, up to six progressive clues, and ICD-10 proximity feedback across five levels.

This text treats the CIWA-Ar as a reasoning instrument, not as a recipe. You will see what the scale measures, what it assumes, when the score misleads and how the decision between a fixed schedule and a symptom-guided schedule depends on questions that come before the number. Addiction here is the setting. The subject is how you decide on shift when a validated instrument seems to decide for you. Doses and local protocols remain with your institution's guideline and with your department's supervision.

Key points

• The CIWA-Ar was validated for people in alcohol withdrawal who can report symptoms; outside those conditions, the number loses its meaning.
• The score measures symptom intensity, not their cause. Sepsis, hypoglycemia, hepatic encephalopathy, head trauma and other withdrawals produce high item scores.
• The question that comes before the scale is diagnostic: has there been recent and relevant alcohol consumption, and does the picture fit the expected time window?
• Symptom-guided schedules reduced the amount of medication in clinical trials, but they only work when the scale is applicable and the team actually reassesses.
• Wernicke often presents incompletely and alongside withdrawal; waiting for the classic triad is a frequent reasoning error.
• Training the differential with progressive clues, in fictional cases, helps separate the number from the hypothesis.

What the CIWA-Ar measures and what it assumes

The Clinical Institute Withdrawal Assessment for Alcohol, revised (CIWA-Ar), was published by Sullivan and colleagues in 1989 as a shortened version of an earlier instrument. There are ten items: nausea and vomiting, tremor, paroxysmal sweats, anxiety, agitation, tactile disturbances, auditory disturbances, visual disturbances, headache or feeling of a heavy head, and orientation. The first nine range from zero to seven; orientation ranges from zero to four. The maximum total is 67.

The design tells a story that matters for reasoning. Half the items depend on the person's report: anxiety, tactile, auditory and visual changes, headache. The examiner asks "do you feel itching, pins and needles, bugs crawling on your skin?" and scores the answer. The other items are observed, but they are still nonspecific. Tremor, sweating and agitation appear in pain, fever, hypoglycemia, hyperthyroidism and anxiety of any origin.

That is why the scale carries three assumptions that rarely appear in writing on the order:

• The diagnosis of alcohol withdrawal has already been made. The CIWA-Ar quantifies the severity of a known syndrome. It was built for that, not to discover the cause of tremor and confusion.
• The person can communicate. Subjective items require understanding the question, a shared language and a level of consciousness sufficient to answer.
• The application is repeated and compared. The clinical value comes from the trajectory: rising, falling or stabilizing after an intervention. A single number says little.

When any of these assumptions fails, you still have a number, but it has stopped measuring what the name promises. This is the kind of trap clinical reasoning needs to see: a validated instrument, applied outside the population in which it was validated, produces a false sense of objectivity.

When the scale misleads: reasoning traps

In a general hospital study, Hecksel and colleagues reviewed admissions in which CIWA-Ar-guided therapy was prescribed and found a large share of people who did not meet the criteria for using the scale: some could not communicate, others had no history of recent alcohol use. In some of them, the medication given for a "high score" coincided with adverse effects. The problem was not the scale itself, but the decision to apply it.

It is worth looking at the traps one by one, because each one corresponds to a known bias.

Anchoring on the "alcoholic" label

When the history of alcohol use appears at admission, it tends to explain everything that follows. Fever becomes "withdrawal", confusion becomes "withdrawal", tachycardia becomes "withdrawal". Croskerry describes anchoring as the tendency to fix on the first information and adjust little in the face of new data. The scale score reinforces the anchor, because it turns into a number the hypothesis that was already in the team's head. The post on examples of heuristics in clinical reasoning details how this shortcut works in other scenarios.

Delirium from another cause

The "orientation" item scores disorientation, and the hallucination items score abnormal perceptions. A person with encephalitis, sepsis, hyponatremia or subdural hematoma can add up many points with no relation to alcohol. Schuckit recalls that delirium from withdrawal is a diagnosis that requires excluding other causes of acute confusion, and that the interval since the last dose helps to calibrate the suspicion.

Those who cannot answer

People who are intubated, or have aphasia, dementia, a language barrier or a lowered level of consciousness receive low or impossible scores on the subjective items. The total falls, the team interprets it as improvement, and real withdrawal goes untreated. The opposite also happens when the team scores "agitation" in someone who has pain or urinary retention.

Premature closure in the face of the number

A score looks like a conclusion. It is easier to prescribe from it than to reopen the question "what else could this be?". Graber and colleagues, analyzing diagnostic errors in internal medicine, identified premature closure as one of the most frequent cognitive factors. The diagnostic pause before closing the case is the simplest countermeasure: before acting on the number, ask whether the hypothesis explains all the findings.

Wernicke hidden behind withdrawal

Wernicke encephalopathy shares confusion and ataxia with severe withdrawal. If the team attributes the confusion to alcohol and waits for the complete triad of confusion, ataxia and ocular changes, it misses the window. The next section returns to this point.

Withdrawal or something else? The differential the score does not make

Think of the CIWA-Ar as you would think of a laboratory test with low specificity: it changes the probability of "withdrawal" little if you do not yet have a strong reason to suspect it. The strong reason comes from history and timing. Recent, regular and heavy alcohol use, with the last dose within a compatible interval, supports the hypothesis. Without that data, the same findings require a broad search.

A practical way to organize the differential is to ask, for each high finding on the scale, "what else produces this?" and "what would I need to see to rule it out?".

• Hypoglycemia: tremor, sweating and confusion. Capillary glucose gives an answer in a minute and must come before any interpretation of the scale.
• Infection and sepsis: fever, tachycardia, confusion. Aspiration pneumonia, meningitis and spontaneous bacterial peritonitis appear frequently in people with alcohol use disorder.
• Hepatic encephalopathy: fluctuating confusion, asterixis, reversed sleep pattern. The context of liver disease changes the differential and the risk of sedation.
• Head trauma: falls are common, the history may be incomplete, and a subdural hematoma grows over hours or days.
• Other substances: benzodiazepine withdrawal, stimulant intoxication, anticholinergic syndrome and serotonin syndrome mimic several items.
• Metabolic and endocrine disorders: hyponatremia, hypercalcemia, thyrotoxicosis.

Time is the most underestimated piece of data. Alcohol withdrawal usually begins in the first hours after the last drink, reaches its greatest intensity in the first days and, when it progresses to delirium, typically does so between the second and fourth day, as Schuckit summarizes. A confusional state that begins a week after admission, in someone who has not drunk since being admitted, fits poorly into that timeline. So does a state that begins in the first hour, before the alcohol has left the body. When the clock and the hypothesis do not match, the clock is usually right.

The history of use must be taken with the same care as any other history. Asking about quantity, frequency, time of the last dose, previous withdrawal episodes and use of other substances, without judgment, increases the chance of a useful answer. When the person cannot provide the information, family members, previous records and the team that brought them in help reconstruct the chronology. Writing "last dose unknown" in the chart is more honest, and more useful to the next team, than assuming a time.

This is the same exercise you do when you practice differential diagnosis with clues: each new piece of data should increase or decrease concrete hypotheses, not merely confirm the one already in the lead.

Wernicke deserves separate attention. Wernicke encephalopathy results from thiamine deficiency and appears in people with alcohol use disorder, malnutrition, prolonged vomiting and the postoperative period of bariatric surgery. Sechi and Serra review the condition and show that the complete triad is the exception. Caine and colleagues proposed operational criteria: two of four signs (dietary deficiency, oculomotor abnormality, cerebellar dysfunction and altered mental state or memory) are enough to raise the hypothesis in people with chronic alcohol use. The NICE CG100 guideline treats the prevention and management of Wernicke in people with harmful alcohol use as part of hospital care, and a Cochrane review on thiamine shows how limited the evidence on regimens still is. For reasoning, the lesson is simple: confusion in someone who drinks calls for actively looking for nystagmus, ophthalmoplegia and ataxia, instead of waiting for them to appear together.

In practice, a good habit is to write the differential next to the score. "High CIWA-Ar; hypotheses: withdrawal, consider infection and Wernicke; normal glucose; no focal sign" takes one line and changes how the next team interprets the same number. The note stops being a record of a scale and becomes a record of reasoning.

Fixed dose or symptom-guided: the decision as reasoning

The discussion between a fixed-schedule regimen and a symptom-guided regimen is usually presented as a service preference. It is, above all, a decision about information: how much you trust your ability to measure, to repeat the measurement and to react to it.

The trials that support the symptom-guided regimen were conducted in selected populations. Saitz and colleagues randomized patients from a specialized unit and observed a lower total amount of medication and a shorter treatment duration with the individualized approach, with no worsening of outcomes. Daeppen and colleagues found a similar result in an inpatient setting. The meta-analysis and guideline by Mayo-Smith consolidated the scale as a way to individualize treatment. The detail that changes the interpretation: these studies largely excluded people with severe medical illness, with difficulty communicating and with other evident causes of confusion.

The current ASAM guideline on management of alcohol withdrawal describes both approaches and reinforces that the symptom-guided regimen presupposes frequent assessment by a trained team. In other words, the choice depends on the conditions of the service and of the person, and not only on the trial evidence.

One way to think about the choice:

1. Is the scale applicable to this person? Communication preserved, diagnosis of withdrawal established, absence of another evident cause.
2. Can the team reassess within the interval the schedule requires? A symptom-guided schedule without reassessment becomes a schedule without treatment.
3. What is the risk of progressing to severe withdrawal? The systematic review by Wood and colleagues in JAMA showed that a prior history of severe withdrawal, such as seizure or delirium, is among the strongest predictors. Prediction tools such as PAWSS have been proposed to estimate this risk on admission.
4. What is the risk of sedation in this person? Liver disease, lung disease, advanced age and use of other substances shift the balance.

Notice that none of these questions is about the dose. All of them are about the quality of the information and about risk. It is the kind of reasoning described in the post about clinical decision-making: choosing a course of action means choosing which error you accept making and how you will notice if you are making it.

Table: score, reading and check question

The table below organizes the reading of the scale as step-by-step reasoning. Approximate ranges follow the original description and the cited guidelines; the specific action for each range depends on the local protocol.

Situation on the scale	Possible reading	Check question before acting
Low and stable score	Mild or absent withdrawal	Can the person answer the subjective items? Has recent use been confirmed?
Intermediate score rising	Withdrawal progressing	Is the interval since the last dose compatible? Have glucose and vital signs been checked?
High score with confusion	Severe withdrawal or delirium from another cause	Is there fever, focal sign, trauma, nystagmus, ataxia or liver disease?
Score falling with drowsiness	Improvement or excessive sedation	Did the drop come from the observed items or from an inability to answer?
High score without a clear history of alcohol	Scale applied outside the validated population	What other syndrome produces tremor, sweating and agitation in this context?
Scale impossible to apply	Communication or consciousness impaired	What instrument or clinical assessment replaces the scale in this case?

The last row is the most forgotten. When the scale does not apply, the answer is a repeated clinical examination and explicit hypotheses in the chart, not an "approximate" score. Recording why the scale was not used protects the next team from the same trap, as the post about organizing clinical reasoning in the chart shows.

Teaching and training the scale: from the shift to the simulated case

Students learn the CIWA-Ar as a form. Learning to add up is easy. The hard part is learning when to stop adding up and start asking. This skill can be trained, and it looks a lot like what is trained in any differential diagnosis case: recognizing when a piece of data confirms the hypothesis and when it merely does not contradict it.

Three practices help turn the scale into reasoning:

• State the hypothesis before the number. "I think it is withdrawal because the last dose was thirty hours ago and he has drunk daily for years" is very different from "CIWA 18".
• List at least two alternatives. For each one, say what data would rule it out. If you cannot name that data, the alternative is still open.
• Review the trajectory, not the last value. A curve that rises despite treatment calls for diagnostic reassessment, not just more medication.

Dual process theory helps explain why this fails at three in the morning: the fast system recognizes "tremulous drinker" and proposes the answer; the analytical system needs a trigger to engage. The scale can be that trigger, as long as you treat it as a question and not as an answer. The same reasoning appears in the post about priority and hypothesis in the residency, where the challenge is deciding what to look at first with little time.

In teaching, simulated cases work well because they allow the information to be controlled. A case that starts with "man with a history of daily alcohol use, tremulous and confused" and only reveals the glucose or the nystagmus at the third clue safely exposes the student's anchoring. The error happens in the fictional case, and the discussion comes right after.

A fictional case in three clues

Imagine the following sequence, put together for teaching. First clue: a woman in her early forties, brought in by family members, with reported daily alcohol use, tremulous and anxious. Almost every student thinks of withdrawal, and rightly so. Second clue: she has been vomiting for several days and barely eats. The withdrawal hypothesis still stands, but a question about nutritional status comes up. Third clue: on examination, horizontal nystagmus and a wide-based gait. Now the question is no longer "how much withdrawal" and has become "is this also Wernicke?".

The value of the exercise lies in the sequence, not in the final diagnosis. Each clue asks the student to say what changed in the list of hypotheses. Those who merely add up symptoms on the scale reach the third clue with a higher number. Those who reason reach it with a new hypothesis and a different priority. It is this difference that calibration of diagnostic confidence seeks to measure: how closely your certainty tracks the evidence you have.

How NEXO trains this reasoning

NEXO presents fictional cases with realistic clinical presentation, with up to six clues revealed in sequence. You propose a diagnosis, search for the ICD-10 code and receive proximity feedback at five levels, which shows whether your hypothesis fell in the right category, the right group or the right chapter. In a scenario like a shift on the ward, this means experiencing the moment when "withdrawal" seems to be the answer and the next clue points to another chapter.

The format speaks to the central idea of this text: the number, or the clue, only has value when it changes a hypothesis. You can use the daily case to practice the question "what else would explain this?" before closing, and review the explanation afterwards to see where the anchor crept in. The Collection page shows the range of diagnoses covered by the cases, and the terms of use are described at /economy.

If you are a physician and want to review cases in your specialty, the NEXO Curator program explains how to take part. Researchers interested in diagnostic reasoning find rules and contact details at NEXO Researcher. Medical content creators have their own program at NEXO Creators, and teaching institutions find the path at /closer. To understand the design of the game, read what NEXO is.

Frequently asked questions

Is CIWA-Ar used to diagnose alcohol withdrawal?

The scale measures symptom intensity in people who already have a diagnosis of withdrawal. The diagnosis comes from the history of use, the time since the last drink and the exclusion of other causes. Using the scale to "discover" withdrawal leads to the error described by Hecksel and colleagues.

What should be done when the person cannot answer the scale questions?

Record that the scale is not applicable and use repeated clinical assessment, with vital signs, neurological examination and sedation or agitation scales suited to the context. Discuss it with the team and follow the service protocol. The ASAM guideline addresses these situations.

Does a low score guarantee that the condition is mild?

A low score reflects the moment and the ability to report. People with a history of seizure or delirium during withdrawal are at higher risk of severe progression, according to Wood and colleagues. The trajectory and prior risk count as much as the current value.

Why think of Wernicke if the person is "only" in withdrawal?

Because the two conditions coexist and are easily confused. The Caine criteria show that two signs already raise the hypothesis. Looking for eye changes and ataxia is part of examining anyone confused with chronic alcohol use.

Is symptom-guided dosing always better?

The trials showed less medication with similar results in selected populations. Outside them, with compromised communication or irregular reassessment, the advantage disappears. The choice depends on the person and the team.

Does this replace my hospital's protocol?

The text is educational and deals with reasoning. Management, doses and protocols follow the local guideline and the supervision of your service. If you or someone close to you is having difficulty related to alcohol, seek the help you need from a health service.

Final checklist

• I confirmed recent alcohol use and the interval since the last dose before applying the scale.
• I measured glucose and vital signs and looked for fever, focal signs and trauma.
• I assessed whether the person can respond to the subjective items.
• I looked for nystagmus, oculomotor changes and ataxia.
• I wrote the main hypothesis and at least two alternatives in the chart.
• I agreed with the team on the reassessment interval and on what to do if the curve rises.

Play today's case: open NEXO, read the first clue and say out loud what would make you change your mind before closing the diagnosis.

References

1. Sullivan JT, Sykora K, Schneiderman J, Naranjo CA, Sellers EM. Assessment of alcohol withdrawal: the revised Clinical Institute Withdrawal Assessment for Alcohol scale (CIWA-Ar). Br J Addict. 1989. PMID 2597811.
2. Hecksel KA, Bostwick JM, Jaeger TM, Cha SS. Inappropriate use of symptom-triggered therapy for alcohol withdrawal in the general hospital. Mayo Clin Proc. 2008. PMID 18315992.
3. Saitz R, Mayo-Smith MF, Roberts MS, et al. Individualized treatment for alcohol withdrawal: a randomized double-blind controlled trial. JAMA. 1994. PMID 8046805.
4. Daeppen JB, Gache P, Landry U, et al. Symptom-triggered vs fixed-schedule doses of benzodiazepine for alcohol withdrawal: a randomized treatment trial. Arch Intern Med. 2002. PMID 12020181.
5. Mayo-Smith MF. Pharmacological management of alcohol withdrawal: a meta-analysis and evidence-based practice guideline. JAMA. 1997. PMID 9214531.
6. Wood E, Albarqouni L, Tkachuk S, et al. Will this hospitalized patient develop severe alcohol withdrawal syndrome? The Rational Clinical Examination systematic review. JAMA. 2018. PMID 30167704.
7. Maldonado JR, Sher Y, Ashouri JF, et al. The "Prediction of Alcohol Withdrawal Severity Scale" (PAWSS): systematic literature review and pilot study. Alcohol. 2014. PMID 24657098.
8. Schuckit MA. Recognition and management of withdrawal delirium (delirium tremens). N Engl J Med. 2014. PMID 25427113.
9. Sechi G, Serra A. Wernicke's encephalopathy: new clinical settings and recent advances in diagnosis and management. Lancet Neurol. 2007. PMID 17434099.
10. Caine D, Halliday GM, Kril JJ, Harper CG. Operational criteria for the classification of chronic alcoholics: identification of Wernicke's encephalopathy. J Neurol Neurosurg Psychiatry. 1997. PMID 9010400.
11. Day E, Bentham PW, Callaghan R, Kuruvilla T, George S. Thiamine for prevention and treatment of Wernicke-Korsakoff Syndrome in people who abuse alcohol. Cochrane Database Syst Rev. 2013. PMID 23818100.
12. Croskerry P. The importance of cognitive errors in diagnosis and strategies to minimize them. Acad Med. 2003. PMID 12915363.
13. Graber ML, Franklin N, Gordon R. Diagnostic error in internal medicine. Arch Intern Med. 2005. PMID 16009864.
14. American Society of Addiction Medicine. The ASAM Clinical Practice Guideline on Alcohol Withdrawal Management. asam.org.
15. National Institute for Health and Care Excellence. Alcohol-use disorders: diagnosis and management of physical complications (CG100). nice.org.uk/guidance/cg100.