Menu

For healthcare professionals

CAM-ICU Delirium Algorithm

The Confusion Assessment Method for the Intensive Care Unit (CAM-ICU) screens adult ICU patients for delirium using four features. Delirium is rated present when features 1 and 2 are present plus feature 3 or feature 4. Ely and colleagues published its validation in 2001. This tool applies the published algorithm to findings you have already assessed; it is written for clinicians and is not a diagnosis.

Last reviewed: By the uncovermyrisk.com editorial team Educational only, not medical advice.

Sudden new confusion or drowsiness outside a monitored setting: get urgent medical care, and call 911 if the person cannot be woken or has signs of a stroke.General safety advice (not part of the published algorithm).

Findings

Assess each feature with the CAM-ICU worksheet and training materials from its developers, then record the findings here. Nothing you enter leaves this page.

Does the patient respond to voice (RASS -3 or above)?
Feature 1: acute change from baseline mental status, or a fluctuating course over the past 24 hours
Feature 2: inattention
Feature 3: disorganized thinking
Feature 4: altered level of consciousness (any level other than alert)
Result appears here

What you will see

CAM-ICU positive, negative, or unable to assess at RASS -4 or -5.

Educational only, not medical advice. Not a diagnosis. Decisions belong to the treating team.

Source

Ely EW and colleagues, JAMA 2001. The four features and the rule are taken from two open-access (CC BY) papers, Tsuruta R and Oda Y, Journal of Intensive Care 2016, and Serafim RB and colleagues, Annals of Intensive Care 2012; the RASS limit from Pisani MA and colleagues, Critical Care 2006 (CC BY). The tests check all 16 combinations of the four features and the RASS -4 or -5 case. This page applies the rule only; it does not reproduce the CAM-ICU worksheet or its bedside tasks.

Validated in

Adult ICU patients. A meta-analysis of nine studies covered 969 patients (Gusmao-Flores 2012). Pisani 2006 reports sensitivity of 95% to 100% and specificity of 89% to 93% in three validation cohorts against expert delirium assessment.

Limits

The result is only as good as the bedside assessment of each feature, which needs training. Patients at RASS -4 or -5 cannot be assessed. Delirium fluctuates, so repeat assessments as your unit requires. In the meta-analysis, pooled sensitivity was 80%, so some delirium is missed (Gusmao-Flores 2012). The CAM-ICU is a screening and monitoring tool, not a full diagnostic interview.

The algorithm

The CAM-ICU looks at four features: acute change or fluctuating course of mental status, inattention, disorganized thinking, and altered level of consciousness (Serafim 2012). It is positive when features 1 and 2 are present along with either feature 3 or feature 4 (Tsuruta 2016) (Serafim 2012). Feature 4 counts any level of consciousness other than alert (Gusmao-Flores 2012). The method was published by Ely and colleagues (Ely 2001).

Sedation is rated first with the Richmond Agitation-Sedation Scale. The CAM-ICU is not performed when the patient is not arousable, at RASS -4 (stupor) or -5 (coma) (Pisani 2006).

Worked example: a patient at RASS -1 has had a change in mental status since admission (feature 1), cannot hold attention (feature 2), and is drowsy, so the level of consciousness is not alert (feature 4). Features 1 and 2 plus feature 4 make the CAM-ICU positive.

What the studies found

Across nine studies with 969 patients, pooled sensitivity was 80.0% and pooled specificity 95.9% (Gusmao-Flores 2012). Because specificity was high, the authors suggest that a positive CAM-ICU need not be confirmed against DSM-IV criteria. Earlier validation cohorts against expert assessment reported sensitivity of 95% to 100% and specificity of 89% to 93% (Pisani 2006).

Related tools

To rate sedation before the CAM-ICU, see the Richmond Agitation-Sedation Scale. Outside the ICU, the 4AT delirium test is a short bedside screen. Other bedside scores are listed under For healthcare professionals.

Sources

  1. Tsuruta R, Oda Y. A clinical perspective of sepsis-associated delirium. J Intensive Care. 2016;4:18. doi:10.1186/s40560-016-0145-4. PMID 27011789. Open access (CC BY). Text (the four CAM-ICU features; positive with features 1 and 2 plus feature 3 or 4). pmc.ncbi.nlm.nih.gov/articles/PMC4804610/
  2. Serafim RB, Dutra MF, Saddy F, et al. Delirium in postoperative nonventilated intensive care patients: risk factors and outcomes. Ann Intensive Care. 2012;2:51. doi:10.1186/2110-5820-2-51. PMID 23272945. Open access (CC BY). Methods (the four features; delirium present with features 1 and 2 plus feature 3 or 4). pmc.ncbi.nlm.nih.gov/articles/PMC3544687/
  3. Pisani MA, Araujo KL, Van Ness PH, Zhang Y, Ely EW, Inouye SK. A research algorithm to improve detection of delirium in the intensive care unit. Crit Care. 2006;10(4):R121. doi:10.1186/cc5027. PMID 16919169. Open access (CC BY). Background (sensitivity 95% to 100%, specificity 89% to 93% in three cohorts) and Methods (not performed at RASS -4 or -5). pmc.ncbi.nlm.nih.gov/articles/PMC1750978/
  4. Gusmao-Flores D, Salluh JI, Chalhub RA, Quarantini LC. The confusion assessment method for the intensive care unit (CAM-ICU) and intensive care delirium screening checklist (ICDSC) for the diagnosis of delirium: a systematic review and meta-analysis of clinical studies. Crit Care. 2012;16(4):R115. doi:10.1186/cc11407. PMID 22759376. Open access (CC BY). Results (9 studies, 969 patients; pooled sensitivity 80.0%, specificity 95.9%). pmc.ncbi.nlm.nih.gov/articles/PMC3580690/

Written and maintained by the uncovermyrisk.com editorial team. We do not have a medical reviewer; every claim links to its primary source above. Last reviewed: 2026-11-02. Spotted an error? See About.