Delirium screening in an acute geriatric ward by nurses using 4AT: results from a quality improvement project

Key summary points

AbstractSection Aim

We aimed to describe the prevalence and duration of delirium, and to evaluate the test accuracy of the delirium screening tool 4AT performed by nurses in an acute geriatric ward.

AbstractSection Findings

The prevalence of delirium in old, frail medical patients was 36%, two out of three patients recovered from their delirium within 4 days after onset. The sensitivity of 4AT performed by nurses was lower than in previous studies where 4AT has been undertaken by experienced geriatricians or delirium researchers.

AbstractSection Message

A thorough training of the staff might be required before the implementation of delirium screening with 4AT in a real-life clinical setting.

Abstract

Methods

We conducted a retrospective chart-based analysis based on a quality improvement project in an acute geriatric ward. Delirium screening with the 4AT was performed by nurses within 24 h after hospital admission. Delirium according to DSM-5 criteria was diagnosed retrospectively based on review of patient records.

Results

In total, 59 patients (mean age 86 years) were included. The prevalence of delirium was 36%. Two out of three patients recovered from their delirium within four days after onset. The sensitivity and specificity of the single-point assessment with 4AT performed by nurses were 50.0% [95% confidence interval (CI) 27.2–72.8] and 86.2% (95% CI 68.3–96.1).

Conclusion

Patients admitted to the acute geriatric ward had a high prevalence of delirium. The sensitivity of the 4AT performed by nurses for delirium was lower than in previous studies where it was undertaken by experienced geriatricians or delirium researchers.

Background

Delirium is an acute condition characterized by disturbed awareness, cognition and attention [1]. The symptoms develop over hours to days and tend to fluctuate. Delirium is caused by one or more medical conditions or drugs, or withdrawal from drugs. The prevalence of delirium in hospitalized older adults varies dependent on setting and patient characteristics [2]. Advanced age, comorbid conditions, cognitive impairment, functional disabilities and impaired vision and hearing are among the most important predisposing factors for delirium [3]. Thus, geriatric patients have a high risk of delirium when acutely admitted to hospital. Nevertheless, many hospital wards lack procedures for detection of delirium and delirium often remains undetected [4]. With the aim to improve delirium detection and care in an acute geriatric ward, we conducted a quality improvement project that included implementation of delirium screening with the 4 ‘A’sTest (4AT; http://www.the4AT.com) performed by nurses. In this paper we report experiences from the project, including prevalence of delirium and the diagnostic accuracy of the 4AT.

Methods

This was a retrospective chart analysis based on a quality improvement project. We included consecutive patients admitted to an eight-bed acute geriatric ward at Bærum hospital, a middle-sized hospital in Norway from April 1st to May 31st 2018.

Data collection

4AT

Nurses were instructed to perform delirium screening with the 4AT within 24 h after admission. The 4AT assesses awareness, cognition, attention and acute change or fluctuation, and has been validated in in-hospital geriatric patients [5,6,7]. Before the project started, the nurses were given a lecture with information about the tool and completed a training period of 2 weeks. All tests performed during the project period were recorded in a database.

Reference diagnosis

Delirium was diagnosed retrospectively by the main author (MM) according to The Diagnostic and Statistical Manual of Mental Disorders (DSM-5) and independently from the screening with 4AT [1]. A chart-based method based on trigger words and phrases was used to extract evidence for each of the diagnostic criteria from the patient records [8, 9]. Results from the delirium screening with 4AT performed by the nurses were not used for the reference diagnosis.

Delirium duration

The duration of delirium was assessed retrospectively. The onset of delirium was defined as the time when the first symptom of delirium was described in the patient records by nurses or doctors. In patients with symptoms of delirium at admission, the onset of delirium was defined as the first observed symptoms reported by relatives or ambulant health care workers. The end of delirium was defined as documentation of permanent recovery from delirium symptoms.

Co-morbid conditions

Pre-existing diagnoses of dementia and other co-morbid conditions were sought through examination of previous discharge letters and medical notes from previous and the actual hospital stay.

Frailty

We assessed frailty retrospectively using the Clinical Frailty Scale (CFS) [10]. An informant questionnaire is used as routine in this ward and information about pre-existing decline in cognitive and physical function was available for most patients.

Ethical considerations

This was a quality improvement project without any intervention with potential to harm the patients. The 4AT is an established clinical screening tool for delirium that consists of methods that are commonly used in the clinical evaluation of acutely ill old patients. Thus, the implementation of the 4AT represents a systematization of these methods with addition of an attention test. We did not ask the patients to consent to the use of data collected in the project. The project and the publication of results from the project were approved by relevant authorities; the local officer for data protection and the Head of the Department of internal medicine.

Statistical methods

We calculated the prevalence of delirium based on delirium reference diagnosis. We used Student’s t test for means of continuous variables and Pearson’s Chi-square test of independence for categorical variables to compare characteristics of patients with and without delirium. We assessed the performance of the single-point screening with 4AT by nurses (first 4AT test performed within 24 h after hospital admission) compared with the reference diagnosis by calculating sensitivity, specificity, positive and negative likelihood ratios, and positive and negative predictive values with 95% confidence intervals using MEDCALC statistical software (http://www.medcalc.org). Other statistical analyses were conducted using SPSS version 25.0 (IBM, Armonk, NY, USA).

Results

In total, 59 patients (32 women (54%)) were admitted during the 2-month project period. The median age was 87 years (range 68–99 years), 13 patients (22%) had dementia, 36 (61%) were frail (CFS-score ≥ 5) and 16 (27%) had sustained a fall within the last 24 h before admission.

A total of 21 patients (36%) had delirium during their hospital stay. Table 1 shows characteristics of patients with and without delirium. Higher age, frailty, dementia and previous cerebrovascular disease were associated with delirium. Two out of three patients recovered from their delirium within 4 days after onset.

Table 1 Characteristics of patients (n = 59) with and without delirium admitted to an acute geriatric ward during 8 weeks

4AT was performed by nurses within 24 h after admission in 49 patients (83%). Four patients were screened later than 24 h after admission. Patients that were not screened had no symptoms of delirium (two patients), were admitted during a weekend with reduced staff capacity (three patients) or had severe dementia (one patient). Table 2 shows the test accuracy of 4AT for delirium compared with the reference diagnosis.

Table 2 Performance of the single-point screening with 4AT by nurses (first 4AT test performed within 24 h after hospital admission) compared with reference diagnosis in patients admitted to an acute geriatric ward during 2 months (n = 49)

Discussion

This project revealed a higher prevalence of delirium than previous studies in old medical patients [3, 5, 6, 11]. Delirium prevalence might vary due to the patient population, and the high prevalence in our project might be explained by a very high mean age and a high prevalence of frailty in the patient population. In line with previous studies, we found that higher age, frailty, dementia, and previous cerebrovascular disease were associated with delirium [12, 13]. The sensitivity of 4AT for delirium was lower in our project (50%) compared to previous studies. Hendry et al. studied the accuracy of several delirium screening tests against an expert evaluation among 500 patients in a geriatric medical assessment unit [6]. In this study, the testing was performed by a single highly specialized researcher and the sensitivity of 4AT for delirium was 87%. Bellelli et al. evaluated the accuracy of the 4AT against an expert evaluation based on the DSM-4-criteria among 234 patients in an acute geriatric ward and a department of rehabilitation and found a sensitivity of 90% for delirium [5]. The authors suggested that the 4AT does not require special training, but the delirium screening in this study was performed by senior geriatricians. The experience from our project suggests a lower sensitivity of the 4AT when performed as single-point screening performed by nurses in a real-life setting, even after a short training period. Many patients were falsely scored with zero points or not scored at the 4AT-item assessing “acute change or fluctuation”. This is important, as this item might be crucial for distinguishing between delirium and dementia. A possible explanation might be that complete collateral information about the habitual cognitive function was unavailable due to the timing of the assessment shortly after admission. In a previous study, 4AT could not be completed in 13.3% due to lack of collateral information [7].

We found that the majority of patients with delirium recovered within a few days. We are not aware of any studies that have investigated duration of delirium in patients acutely admitted to acute geriatric wards and suggest that factors influencing delirium duration should be assessed in prospective studies.

Limitations

The duration of the project was only two months and the number of patients included was low, therefore, the estimated accuracy of the 4AT must be interpreted with caution. Delirium was diagnosed retrospectively using a chart-based method and incomplete or imprecise documentation of delirium symptoms might have influenced the estimations of delirium prevalence and duration. Patients with symptoms caused by for instance dementia, psychiatric disorders, anxiety or pain might have been misclassified as delirium and have caused too high a prevalence of delirium. On the other hand, some patients with delirium might have been misclassified as without delirium due to lack of documentation in the patient records. However, we performed the patient record review based on previous studies that have demonstrated that delirium can be diagnosed retrospectively with good accuracy in hospitalized patients [8, 9]. Compared to an interview-based method, a chart-based method might be more likely to diagnose delirium if the symptoms occur at night and in individuals with hyperactive delirium, but less sensitive to hypoactive symptoms and milder delirium.

Conclusions

Patients admitted to the acute geriatric ward were old and frail and had a high prevalence of delirium. The sensitivity of the 4AT was lower than in previous studies where 4AT was undertaken by experienced geriatricians or delirium researchers. The experiences from this quality improvement projects suggest that a thorough training of the staff is required before implementation of delirium screening with 4AT in a real-life clinical setting.