Does This Patient Have Dementia?

Systematic Review Source 

This is a rational clinical examination abstract, a regular feature of the Annals' Evidence-Based Emergency Medicine (EBEM) series. Each features an abstract of a rational clinical examination review from the Journal of the American Medical Association and a commentary by an emergency physician knowledgeable in the subject area.

The source for this rational clinical examination review abstract is: Holsinger T, Deveau J, Boustani M, et al. The rational clinical examination: does this patient have dementia? JAMA. 2007;297:2391-2404. The Annals' EBEM editors assisted in the preparation of the abstract of this rational clinical examination review, as well as selection of the Evidence-Based Medicine Teaching Points.

Objective 

To review the literature concerning the practicality and accuracy of brief cognitive screening instruments in primary care.

Data Sources 

The authors report a MEDLINE search (including data from AIDSLINE, BioethicsLine, and Health STAR) from 1994 through April 2006, using search terms “exp Alzheimer's disease” and “exp dementia” with Clinical Queries diagnostic search strategy. Also, psycINFO was searched using “sensitivity” and “specificity” as additional key words.1

Study Selection 

The authors included studies enrolling subjects older than 60 years and which used an acceptable criterion standard to diagnose dementia. Exclusion criteria included non–English language articles, inpatient or nursing home isolated populations, memory disorder clinic populations without an adequately characterized outside control group, or populations with less than 6 years' median education. Ancillary test studies incorporating laboratory or imaging testing modalities were not evaluated in this review.

Data Extraction and Analysis 

Eligible studies were assessed for quality by 2 independent raters according to the sample size, participant selection, and the use of a credible criterion standard applied blindly and independently. High-quality studies included sample sizes of more than 100 participants, with uniform application of the criterion standard interpreted independently from the screening instrument being tested. Studies rated as good used a random or consecutive sampling of community-dwelling or primary care-setting patients with the independent, blinded application of the intact instrument being assessed. Poor studies had fatal flaws such as the criterion standard applied only to individuals with a positive screening test result. Data were abstracted about population demographics, screening instrument details, and the possibility of selection bias or verification bias. Data required to construct 2×2 tables and to compute sensitivity, specificity, and likelihood ratios were abstracted. Because of variation in design, medians with data ranges were reported rather than pooled results with confidence intervals (CIs).

Main Results 

Twenty-five screening instruments were identified in 29 studies, including 18 completed by patients, 3 by informants, and 4 using information from both patients and informants. Methodological problems were common, and only 3 studies were rated as good compared with 26 fair studies. Verification bias was identified in several studies that obtained only the criterion standard for individuals who screened positive for cognitive dysfunction. Other studies recruited patients only from a memory clinic (spectrum bias), failed to use intact screening instruments, or did not blind investigators to the results of the criterion standard.

The authors separated the screening instruments into 3 categories: brief instruments, comprehensive instruments, and special situation instruments. The Mini-Mental State Examination2 has been studied more than any other screening tool but generally requires 7 to 10 minutes to administer, has significant performance flaws in both poorly educated and highly educated populations,3, 4 and is insensitive to mild impairment,5 and its reproduction is restricted by copyright issues.2, 6, 7 The median positive likelihood ratio for the Mini-Mental State Examination is 6.3 (range 3.4 to 47.0), whereas the median negative likelihood ratio is 0.19 (0.06 to 0.37), with cut points for an abnormal result ranging in studies from 23 to 26.

Among the brief screening instruments, the best diagnostic performances were noted in the 7-minute screen, Memory Impairment Screen, Brief Alzheimer Screen, Mini-Cog, and Six-Item Screener (Table). Unfortunately, the Brief Alzheimer Screen, Mini-Cog, and Six-Item Screener each had 2 methodological flaws: first, the study screening test was not administered as reported, and second, blinding to the criterion standard was not ensured. One special situation instrument, the Cognitive Assessment Screening Test, targeted patients waiting for evaluation who are able to independently use a paper and pencil, demonstrating a positive likelihood ratio of 17.0 (95% confidence interval [CI] 4.2 to 66.0) and negative likelihood ratio of 0.13 (95% CI 0.02 to 0.81). This instrument was also the only screening tool including questions about functional impairment.

Table. Brief dementia screening tools' diagnostic test performance

	Screening Test	Positive LR (95% CI)	Negative LR (95% CI)
	7-Minute Screen	47(3–730)	0.09(0.01–0.59)
	Memory Impairment Screen	33(15–72)	0.08(0.02–0.30)
	Brief Alzheimer Screen	25(17–35)	0.02(0.01–0.04)
	Mini-Cog	13.0(9.9–17.0)	0.25(0.17–0.37)
	6-Item Screener	7.3(5.1–10.0)	0.15(0.04–0.14)

LR, Likelihood ratio.

Conclusions 

An aging population will increase the volume of patients with dementia encountered in primary care settings. Clinicians should select one population-appropriate primary screening tool and consider others for specific situations. For example, if one has very little time available, the Clock Drawing Test may be the most useful screening tool, whereas the Hopkins Verbal Learning Test may be superior in mildly impaired or highly educated patients. The Mini-Mental State Examination has been evaluated most extensively, but current copyright restrictions limit its use, and diagnostic inaccuracy is a problem in relationship to educational levels. High-functioning, educated populations can be tested with instruments demonstrating less ceiling effect, but so far these tools are more time consuming.

Commentary: Clinical Implication 

Aging adults represent an increasing percentage of emergency department (ED) patients, yet emergency physicians report insufficient training in their evaluation and management8 and ED-based studies have confirmed the poor recognition of cognitive impairment among older adults.9, 10, 11 Recognition of cognitive dysfunction in emergency medicine may improve patient safety by enhancing clinician awareness of limitations in patient histories, enhancing detection of new cognitive impairments, and suggesting the need for reinforcement of follow-up and treatment plans with nonimpaired caregivers and family.

The screening tests presented in this rational clinical examination offer a variety of tools of moderate quality to identify dementia. The comprehensive instruments generally require more than 10 minutes to administer and include personnel and equipment not readily available in the ED; however, the brief instruments demonstrate impressive test performance and most can be completed in fewer than 10 minutes. Unfortunately, ED patients have rarely been the target population for studies of such screening tests. One single-center, ED-based study prospectively compared the test characteristics of the Six-Item Screener and Mini-Cog in 149 older patients and found the Six-Item Screener to have relatively high potential utility, with a positive likelihood ratio of 6.6 (95% CI 3.5 to 13) and a negative likelihood ratio of 0.06 (95% CI 0.01 to 0.44).12 If the results are validated, a promising option would be self-administered dementia screening tools (such as the Cognitive Assessment Screening Test) that could potentially be completed during ED waiting periods.

Effective interventions exist to modify the disease burden of Alzheimer's disease and other dementing illnesses when detected early, making screening and identification a potentially high-yield proposition.13 Despite the challenge that screening modalities present to crowded EDs, limiting dementia disease-related morbidity may reduce ED crowding and thereby improve care for all populations.

Take-Home Message 

Brief dementia screening tools with excellent diagnostic test characteristics exist, but most have yet to be validated in ED settings. For the ED, the Six-Item Screener was superior to the Mini-Cog in a single trial. More comprehensive, time-consuming instruments have been described, but their diagnostic test characteristics are not superior to the briefer, ED-friendly tools.

EBEM Teaching Point