Abstract

RATIONALE:

Delirium is often unrecognized in ICU patients and associated with poor outcome. Screening for ICU delirium is recommended by several medical organizations to improve early diagnosis and treatment. The Confusion Assessment Method for the ICU (CAM-ICU) has high sensitivity and specificity for delirium when administered by research nurses. However, test characteristics of the CAM-ICU as performed in routine practice are unclear.

OBJECTIVES:

To investigate the diagnostic value of the CAM-ICU in daily practice.

METHODS:

Teams of three delirium experts including psychiatrists, geriatricians, and neurologists visited 10 ICUs twice. Based on cognitive examination, inspection of medical files, and Diagnostic and Statistic Manual of Mental Disorders, 4th edition, Text Revision criteria for delirium, the expert teams classified patients as awake and not delirious, delirious, or comatose. This served as a gold standard to which the CAM-ICU as performed by the bedside ICU-nurses was compared. Assessors were unaware of each other's conclusions.

MEASUREMENTS AND MAIN RESULTS:

Fifteen delirium experts assessed 282 patients of whom 101 (36%) were comatose and excluded. In the remaining 181 (64%) patients, the CAM-ICU had a sensitivity of 47% (95% confidence interval [CI], 35%-58%); specificity of 98% (95% CI, 93%-100%); positive predictive value of 95% (95% CI, 80%-99%); and negative predictive value of 72% (95% CI, 64%-79%). The positive likelihood ratio was 24.7 (95% CI, 6.1-100) and the negative likelihood ratio was 0.5 (95% CI, 0.4-0.8).

CONCLUSIONS:

Specificity of the CAM-ICU as performed in routine practice seems to be high but sensitivity is low. This hampers early detection of delirium by the CAM-ICU.