News & Events




Known as 3D-CAM, the test could be an important tool to help clinicians and nurses more quickly identify – and treat -- this common and costly complication of hospitalization

BOSTON -- Delirium is a state of confusion that develops suddenly, often following an acute  medical illness, a surgical procedure or a hospitalization. Although delirium is estimated to complicate hospital stays for over 2.5 million elderly individuals in the U.S. each year, this common condition often goes undetected. The end result can be serious complications with sometimes devastating consequences for vulnerable hospitalized elders.

Now, a study led by investigators at Beth Israel Deaconess Medical Center (BIDMC)  reports that 3D-CAM, a three-minute diagnostic assessment based on the widely used Confusion Assessment Method algorithm, identified delirium with greater-than-90-percent specificity and sensitivity, and of particular note, was almost equally as accurate in identifying the condition in patients with dementia.

Published in the October 21 issue of the Annals of Internal Medicine, the new findings suggest that this easy-to-administer tool could significantly improve detection of this common and morbid condition in vulnerable older hospital patients.

“Prompt recognition of delirium is the first step to timely evaluation and treatment, preventing complications and keeping older patients safe while in the hospital,” says lead author Edward Marcantonio, MD, SM, Section Chief for Research in the Division of General Medicine and Primary Care at BIDMC and Professor of Medicine at Harvard Medical School. “As growing numbers of older adults are being hospitalized, it’s critically important that doctors, nurses and other hospital care providers be able to recognize delirium. We wanted to develop a brief and simple method to make this easier to accomplish.”

“Unfortunately,  delirium, which affects 33 percent of older medical patients and between 15 and 50 percent of older surgical patients, remains distressingly under-recognized, with average detection rates of only 12 to 35 percent in most clinical settings,” adds Marcantonio. “Moreover, he adds, “the identified cases of delirium tend to be agitated patients who are disruptive to patient care, whereas hypoactive [quiet, lethargic] patients remain unrecognized. Studies have shown that hypoactive patients with delirium have either similar or somewhat worse outcomes than agitated patients. The 3D-CAM showed excellent sensitivity in our study, despite the fact that 88 percent of the patients with delirium had the more-difficult-to-diagnose hypoactive delirium.”    

The CAM algorithm was developed in 1990 by the study’s senior author Sharon K. Inouye, MD, MPH, Director of the Aging Brain Center in the Institute for Aging Research at Hebrew Senior Life and HMS Professor of Medicine in the Division of Gerontology at BIDMC. To date, The CAM has been used in over 4000 original studies and has been translated into over 14 languages. The CAM diagnostic algorithm requires that the assessor determine the presence or absence of four key features of delirium: 1) acute change and fluctuating course; 2) inattention; 3) disorganized thinking; and 4) altered level of consciousness. To be diagnosed with delirium, a patient must have features 1 and 2 and either 3 or 4.

“We have found that there are many different cognitive tests that the person rating the CAM can use to assess for these four features, and we’ve shown that the quality of the assessment makes a big difference in the accuracy of identification of delirium,” explains Inouye. “The 3D-CAM is a major advance, since it provides a brief, easy-to-administer approach that operationalizes the CAM algorithm in three minutes, and provides highly accurate results compared to a gold standard clinical assessment.”

To develop the 3D-CAM assessment tool, the investigators reduced an original list of 160 questions and observations down to 20 items. To do this, each item was evaluated using a modern measurement approach called Item Response Theory, which is also used to create educational tests such as the SAT. Only the most informative items for delirium diagnosis were selected for inclusion in the final 3D-CAM assessment.  Examples include patient questions about symptoms (“Have you been feeling confused?”), cognitive testing of attention and orientation, and structured observations (“Did the patient fall asleep during the interview?”)

After selecting the 20 best items and assembling the 3D-CAM assessment, the authors conducted a prospective validation study by enrolling a  total of 201 hospital patients over age 75 from BIDMC’s General Medicine Service.  

The authors first conducted a “gold standard” clinical assessment for delirium and dementia, in which an experienced clinician conducted a full patient evaluation including a cognitive exam, a review of the patient’s medical records and conversations with the patient’s nurse and family caregiver. This assessment took between one and one and a half hours and resulted in data similar to a doctor’s initial evaluation.

An expert panel then reviewed all of this data and made a judgment as to the presence or absence of delirium and dementia. The “gold standard” assessment, determined that 21 percent of the participants had delirium, 88 percent of which was hypoactive or “quiet.” They also found that 28 percent of patients had dementia prior to being admitted to the hospital. In some cases, patients had both delirium and dementia. Research assistants subsequently administered the 3D-CAM assessment without knowledge of the gold-standard results.

“First, we timed the test, and found that, on average, it took only three minutes to administer,” says Marcantonio. The researchers then compared the results of the 3D-CAM with the gold standard assessment and found that the 3D-CAM correctly identified 95 percent of the patients with delirium (95 percent sensitivity) while correctly identifying 94 percent of patients without delirium (94 percent specificity). When a second research assistant went back and administered the 3D-CAM without knowledge of the first test results, the answer was the same 95 percent of the time (95 percent reproducibility.) Importantly, the 3D-CAM performed nearly as well in patients with dementia, which is a particularly challenging group in which to diagnose delirium.

 “We are extremely happy with the final results,” says Marcantonio. “Given its brevity, ease of use, and excellent accuracy and reproducibility, the 3D-CAM could be an important component of a program to improve recognition and management of delirium in older adults.”  Inouye adds, “Hospitals throughout the world are increasingly recognizing the importance of delirium as a major preventable adverse event.  The 3D-CAM holds great promise as an important advance for delirium care specifically and for acute care for elders more generally.”

In addition to Marcantonio and Inouye, study coauthors include BIDMC investigators Long H. Ngo, PhD, Margaret O’Connor, PhD and Eran Metzger, MD; Richard N. Jones, SCD of the Warren Alpert School of Medicine at Brown University; and Paul K. Crane, MD, MPH, of the University of Washington School of Medicine.

This study was funded by the following grants from the National Institute of Aging: R01AG030618; K24AG035075; P01AG031720; K07AG041835. The 3D-CAM instrument and instructions are available at: