Brief Report
Performance of the modified Richmond Agitation Sedation Scale in identifying delirium  in older ED patients

https://doi.org/10.1016/j.ajem.2017.05.025Get rights and content

Abstract

Background

Delirium in older emergency department (ED) patients is associated with severe negative patient outcomes and its detection is challenging for ED clinicians. ED clinicians need easy tools for delirium detection. We aimed to test the performance criteria of the modified Richmond Agitation Sedation Scale (mRASS) in identifying delirium in older ED patients.

Methods

The mRASS was applied to a sample of consecutive ED patients aged 65 or older by specially trained nurses during an 11-day period in November 2015. Reference standard delirium diagnosis was based on Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR) criteria, and was established by geriatricians. Performance criteria were computed. Analyses were repeated in the subsamples of patients with and without dementia.

Results

Of 285 patients, 20 (7.0%) had delirium and 41 (14.4%) had dementia. The sensitivity of an mRASS other than 0 to detect delirium was 0.70 (95% confidence interval, CI, 0.48; 0.85), specificity 0.93 (95% CI 0.90; 0.96), positive likelihood ratio 10.31 (95% CI 6.06; 17.51), negative likelihood ratio 0.32 (95% CI 0.16; 0.63). In the sub-sample of patients with dementia, sensitivity was 0.55 (95% CI 0.28; 0.79), specificity 0.83 (95% CI 0.66; 0.93), positive likelihood ratio 3.27 (95% CI 1.25; 8.59), negative likelihood ratio 0.55 (95% CI 0.28; 1.06).

Conclusion

The sensitivity of the mRASS to detect delirium in older ED patients was low, especially in patients with dementia. Therefore its usefulness as a stand-alone screening tool is limited.

Introduction

Delirium is frequent in older emergency department (ED) patients and associated with adverse outcomes such as mortality, prolonged hospitalizations, as well as persistent cognitive and functional decline [1], [2]. However, in the ED, detection of delirium is often difficult, especially if delirium is superimposed on dementia [3].

Several two-step bedside tests have been developed to aid ED clinicians in systematically screening for and assessing delirium. These are, in essence, adaptions of the original Confusion Assessment Method [4], [5], [6], [7], [8]. As inattention is one of the cardinal symptoms of delirium, a first step is screening for inattention. The application of such tools is feasible, but is dependent on patients' ability to cooperate and communicate [6]. Therefore, a simple alternative is observing patients' level of consciousness because an altered level of consciousness is another main feature of delirium. Level of consciousness can readily be evaluated with the Richmond Agitation Sedation Scale (RASS) [9]. In a trial investigating several screening tools in the ED [5], the RASS performed well: a RASS score other than zero revealed a sensitivity of 0.840 (95% CI 0.738–0.942), specificity of 0.876 (95% CI 0.842–0.911), positive likelihood ratio of 6.8 (95% CI 5.0–9.2), and negative likelihood ratio of 0.2 (95% CI 0.1–0.3) [10]. However, analyses in that study were conducted in a convenience sample without stratification into patients with and without dementia. We therefore aimed to investigate the performance of the RASS in identifying delirium in a consecutive sample of older ED patients both with and without dementia.

Section snippets

Methods

This was a pre-planned subanalysis of data collected in a larger prospective trial in which we tested the performance of the modified Confusion Assessment Method for the ED (mCAM-ED) [6] (ClinicalTrials.gov NCT02782143). The study was approved by the local ethics committee (EKNZ-2015-123). We included a consecutive sample of ED patients aged 65 and older presenting to our institution's ED during an 11 day period in November 2015. Approximately 50,000 adult patients from all medical specialties

Results

During the study period, 416 patients aged 65 and older presented to our ED (Fig. 1) of which 285 were included in the analysis. The median age was 79.9 years (interquartile range 72.4; 86.7) and 58.9% were female. Most patients were assigned Emergency Severity Index 3 (56.7%) and 2 (26.8%). Cross tabulation of the mRASS by delirium is displayed in Table 1.

The performance of the mRASS to diagnose delirium is shown in Table 2. An mRASS other than 0 had a sensitivity of 0.70 (95% CI 0.48; 0.85)

Main discussion

The RASS is different from other delirium screening tools in that it can be administered by simply observing patients. This makes it attractive for use in a busy ED setting [7], [8]. However, there are only few other studies on the performance of the RASS to identify delirium [10], [12], [13], [14], of which only two were performed in the ED setting [10], [13]. In an explanatory study on hip fracture patients the RASS had a sensitivity of 0.80, and specificity 0.79 for delirium [14]. In another

Acknowledgements

We thank the research assistants who performed the mRASS and the attending geriatricians serving as a reference standard and Duncan Shabb, MD, for helpful discussions.

References (15)

There are more references available in the full text version of this article.

Cited by (17)

  • How can we identify patients with delirium in the emergency department?: A review of available screening and diagnostic tools

    2017, American Journal of Emergency Medicine
    Citation Excerpt :

    Grossmann et al. also performed a relevant subclass analysis, allocating subjects to either a group with or without dementia [11]. The aforementioned parameters in patients with dementia were 0.55, 0.83, 0.55 and 0.83, respectively, leading the authors to conclude that the mRASS was not sensitive enough to identify delirium in the ED setting especially in patients with dementia [11]. However, the same parameters were 0.89, 0.94, 0.38 and 1.00, respectively, in patients without dementia [11], a finding which seems sufficient to integrate mRASS into routine ED practice for patients with no previous history of dementia.

View all citing articles on Scopus
View full text