Automated Screening May Improve the Detection of Delirium in Older Adults

Delirium affects an estimated 10% to 30% of older adults in the emergency department (ED), yet its often-subtle presentation can make the common condition difficult to recognize. A new Cleveland Clinic study suggests that using the electronic medical record (EMR) to identify patients at increased risk for delirium could help close that gap.

Published in JACEP Open, the study evaluated a shift from symptom-based delirium screening to an automated, risk-based approach across 17 EDs. By using EMR-based risk stratification to target at-risk patients for delirium assessments, researchers hoped to increase screening rates and identify cases that might otherwise be missed.

“Delirium can cause enormous distress for both patients and families and have major consequences for health outcomes, but it can be especially challenging to recognize in patients who present without obvious changes in mental status,” says lead author and Cleveland Clinic geriatrician Saket Saxena, MD. “A targeted approach can help bring those patients to the attention of the care team earlier, when there may be greater opportunity to identify and address the underlying causes.”

Moving beyond symptoms

Before Dr. Saxena’s team implemented the new EMR-based process, geriatric patients in the participating EDs were considered for delirium screening when a triage nurse or caregiver identified concerning mental status changes. Those patients then underwent the 4AT, a brief, standardized delirium evaluation. A score of 4 or higher was considered positive for possible delirium.

Researchers were concerned, however, about an important limitation of the standard approach: It required caregivers to recognize altered mental status before the formal screening occurred.

“Symptom-based screening tools ask clinicians to recognize the problem before they have completed the assessment designed to identify it,” explains coauthor Stephen Meldon, MD, an emergency physician at Cleveland Clinic. “That creates an unfortunate opportunity for subtle and atypical presentations to fall through the cracks.”

To address that gap, researchers developed automated EMR criteria based on established delirium risk factors. Patients were considered high risk if they were aged 80 or older with at least one risk factor, or aged 65 to 79 with at least two. Risk factors included falls, a history of dementia, use of 10 or more medications and five or more ED visits during the previous year. When the EMR identified a high-risk patient, it prompted the triage nurse to complete the 4AT.

Increased screening

The study compared the ED encounters of older adults during a baseline period from Jan. 1 through July 6, 2023, with encounters during the intervention period from July 7, 2023, through Jan. 6, 2024. More than 217,000 visits were ultimately included in the analysis.

During the baseline period, 3,267 encounters resulted in a documented concern for altered mental status. Of the 2,241 patients who underwent 4AT screening, 84.8% had a positive result.

With EMR-based risk stratification, 39,065 encounters were identified as involving a high-risk geriatric patient, and 14,125 received the 4AT. Although the proportion of positive screens was lower — 14.3%, compared with 84.8% in the baseline group — the targeted approach resulted in a six-fold increase in the number of completed screenings.

The difference in positive-screen distribution also provided an important insight. The symptom-based approach identified a greater proportion of patients with very high 4AT scores, whereas the risk-stratified approach identified more patients with lower positive scores of 4 to 6.

“This suggests that automated targeting may help detect subtler presentations that could be missed when screening is triggered only by obvious mental status changes,” Dr. Meldon explains.

Merging technology and clinical expertise

Importantly, the researchers do not suggest that EMR-based screening should replace clinical judgment. Instead, they propose combining risk stratification with additional patient screenings when clinicians have concerns.

Notably, the intervention did not appear to substantially increase operational burden. In both study periods, the median time between ED arrival and delirium screening was approximately 10 minutes. Median ED length of stay increased by four minutes during the intervention period, a difference that the authors characterized as having no meaningful impact on ED operations.

“Technology is most useful when it supports — not replaces — clinical expertise,” Dr. Saxena says. “The EMR can identify patients who meet objective risk criteria, while clinicians retain the ability to recognize additional patients who warrant screening.”

He notes that the study also highlights an opportunity to identify possible cognitive impairment beyond delirium. During the intervention period, 15.6% of screened patients had 4AT scores between 1 and 3, indicating possible cognitive impairment, compared with 7% during the baseline period. Although these scores do not meet the threshold for possible delirium, they may identify patients who warrant further cognitive evaluation and measures for delirium prevention.

Although additional research is needed, the authors say their findings demonstrate that automated risk stratification can be implemented at scale and may complement clinical assessments by expanding delirium screening beyond patients with obvious changes in mental status.

“For healthcare teams managing older adults in the ED, the goal is not simply to perform more screenings,” Dr. Meldon adds. “It is to create a reliable process that helps clinicians recognize delirium — including the cases that are easiest to overlook — and respond as early as possible.”