How accurate is the CAM-ICU tool for the diagnosis of delirium in adult patients admitted to critical care units?

Key messages

Our review suggests that the CAM-ICU is a reasonable test to look for delirium in critical care settings. The test may be especially good for ruling out delirium. If the CAM-ICU suggests no delirium, it is unlikely the person has delirium.

What was the aim of the review?

To evaluate the accuracy of the CAM-ICU tool for detecting delirium in patients in critical care units.

Why is recognizing delirium important?

When people are unwell, they can experience changes in their thinking, level of consciousness, or ability to sustain attention. This is called delirium. Delirium becomes more likely as people become more unwell. Delirium is very common when people are so unwell that they need enhanced monitoring and treatment, for example in an intensive care unit. Areas of the hospital providing this enhanced care are called ‘critical care’.

Compared to people who don’t experience delirium, people with delirium are more likely to require a longer stay in hospital and are more likely to die. Delirium is associated with distress for both the patient and their family.

Prompt recognition of when a person has delirium would allow for treatment. However, assessing for the presence of delirium is difficult. It is even more complicated in a critical care setting, as the patient may have equipment to help them breathe, which means they can’t easily speak with the assessor.

What is the CAM-ICU?

The Confusion Assessment Method for the Intensive Care Unit (CAM-ICU) is a test to look for delirium. It is based on a series of questions and commands that are designed to detect the features of delirium.

Certain aspects of the CAM-ICU test make it suitable for busy critical care settings. The CAM-ICU test can be completed even if the patient being tested can’t speak. The CAM-ICU is quick to perform and requires minimal training, thus it can be used by any member of the healthcare team.

What was studied in the review?

We reviewed studies that assessed the accuracy of the CAM-ICU when used in a critical care setting. We included those studies that compared the results of the CAM-ICU test, performed by a CAM-ICU trained professional, with a more detailed ‘gold standard’ diagnostic assessment.

What are the main results of the review?

We included 25 studies with a total of 2817 participants. CAM-ICU correctly identified patients with delirium about 78% of the time. The test was better at identifying those without delirium, giving correct results about 95% of the time.

If the CAM-ICU was applied to a group of 100 patients in critical care, where 40% (40) of them have delirium, it is estimated that the CAM-ICU tool would correctly detect around 31 patients with delirium and miss nine cases. The CAM-ICU would suggest that three patients have delirium when they do not.

How reliable are the results of the studies in the review?

Despite the low risk of bias and the high numbers of participants in the included studies, we are uncertain about the results due to differences among studies in the way the test was applied and in the interpretation of the ‘gold standard’ comparison.

To whom do the results of this review apply?

We only looked at papers that studied the CAM-ICU test when used in a critical care setting. CAM-ICU is used in other healthcare settings, but that was not the focus of this review.

How up‐to‐date is this review?

Our search of the literature included papers published up to 8 July 2022.

Authors' conclusions: 

The CAM-ICU tool may have a role in the early identification of delirium, in adult patients hospitalized in intensive care units, including those on mechanical ventilation, when non-specialized, properly trained clinical personnel apply the CAM-ICU. The test is most useful for exclusion of delirium. The test may miss a proportion of patients with incident delirium, therefore in situations where detection of all delirium cases is desirable, it may be best to repeat the test or combine CAM-ICU with another assessment. Future studies should compare different screening tests proposed for bedside assessment of delirium, as this approach will reveal which tool yields superior accuracy. In addition, future studies should consider and report the flow and timing of the tests and clearly report key characteristics related to patient selection. Finally, future research should focus on the impact of CAM-ICU screening on patient outcomes.

Read the full abstract...

Delirium is an underdiagnosed clinical syndrome typified by an acute alteration of mental state. It is an important problem in critical care and intensive care units (ICU) due to its high prevalence and its association with adverse outcomes. Delirium is a very distressing condition for patients, with a huge impact on their well-being. Diagnosis of delirium in the critical care setting is challenging. This is especially true for patients who are mechanically ventilated and are therefore unable to engage in a verbal interview.

The Confusion Assessment Method for the Intensive Care Unit (CAM-ICU) is a tool specifically designed to assess for delirium in the context of ICU patients, including those on mechanical ventilation. CAM-ICU can be administered by non-specialists to give a dichotomous delirium present/absent result.


To determine the diagnostic accuracy of the CAM-ICU for the diagnosis of delirium in adult patients in critical care units.

Search strategy: 

We searched MEDLINE (Ovid SP, 1946 to 8 July 2022), Embase (Ovid SP, 1982 to 8 July 2022), Web of Science Core Collection (ISI Web of Knowledge, 1945 to 8 July 2022), PsycINFO (Ovid SP, 1806 to 8 July 2022), and LILACS (BIREME, 1982 to 8 July 2022). We checked the reference lists of included studies and other resources for additional potentially relevant studies. We also searched the Health Technology Assessment database, the Cochrane Library, Aggressive Research Intelligence Facility database, WHO ICTRP,, and websites of scientific associations to access any annual meetings and abstracts of conference proceedings in the field.

Selection criteria: 

We included diagnostic studies enrolling adult ICU patients assessed using the CAM-ICU tool, regardless of language or publication status and reporting sufficient data on delirium diagnosis for the construction of 2 x 2 tables. Eligible studies evaluated the diagnostic performance of the CAM-ICU versus a clinical reference standard based on any iteration of the Diagnostic and Statistical Manual of Mental Disorders (DSM) criteria applied by a clinical expert.

Data collection and analysis: 

Two review authors independently selected and collated study data. We assessed the methodological quality of studies using the QUADAS‐2 tool. We used two univariate fixed-effect or random-effects models to determine summary estimates of sensitivity and specificity. We performed sensitivity analyses that excluded studies considered to be at high risk of bias and high concerns in applicability, due mainly to the target population included (e.g. patients with traumatic brain injury). We also investigated potential sources of heterogeneity, assessing the effect of reference standard diagnosis and proportion of patients ventilated.

Main results: 

We included 25 studies (2817 participants). The mean age of participants ranged from 48 to 69 years; 15 of the studies included critical care units admitting mixed populations (e.g. medical, trauma, surgery patients). The percentage of patients receiving mechanical ventilation ranged from 11.8% to 100%. The prevalence of delirium in the studies included ranged from 12.5% to 83.9%. Presence of delirium was determined by the application of DSM-IV criteria in 13 out of 25 included studies. We assessed 13 studies as at low risk of bias and low applicability concerns for all QUADAS-2 domains. The most common issue of concern was flow and timing of the tests, followed by patient selection.

Overall, we estimated a pooled sensitivity of 0.78 (95% confidence interval (CI) 0.72 to 0.83) and a pooled specificity of 0.95 (95% CI 0.92 to 0.97).

Sensitivity analysis restricted to studies at low risk of bias and without any applicability concerns (n = 13 studies) gave similar summary accuracy indices (sensitivity 0.80 (95% CI 0.72 to 0.86), specificity 0.95 (95% CI 0.93 to 0.97)).

Subgroup analyses based on diagnostic assessment found summary estimates of sensitivity and specificity for studies using DSM-IV of 0.79 (95% CI 0.72 to 0.85) and 0.94 (95% CI 0.90 to 0.96). For studies that used DSM-5 criteria, summary estimates of sensitivity and specificity were 0.75 (95% CI 0.67 to 0.82) and 0.98 (95% CI 0.95 to 0.99). DSM criteria had no significant effect on sensitivity (P = 0.421), but the specificity for detection of delirium was higher when DSM-5 criteria were used (P = 0.024). The relative specificity comparing DSM-5 versus DSM-IV criteria was 1.05 (95% CI 1.02 to 1.08).

Summary estimates of sensitivity and specificity for studies recruiting < 100% of patients with mechanical ventilation were 0.81 (95% CI 0.75 to 0.85) and 0.95 (95% CI 0.91 to 0.98). For studies that exclusively recruited patients with mechanical ventilation, summary estimates of sensitivity and specificity were 0.91 (95% CI 0.76 to 0.97) and 0.98 (95% CI 0.92 to 0.99). Although there was a suggestion of differential performance of CAM-ICU in ventilated patients, the differences were not significant in sensitivity (P = 0.316) or in specificity (P = 0.493).