Accuracy of dementia screening instruments in emergency medicine: A diagnostic meta-analysis

Background: Dementia is underrecognized in older adult emergency department (ED) patients, which threatens operational efficiency, diagnostic accuracy, and patient satisfaction. The Society for Academic Emergency Medicine geriatric ED guidelines advocate dementia screening using validated instruments.

Objectives: The objective was to perform a systematic review and meta-analysis of the diagnostic accuracy of sufficiently brief screening instruments for dementia in geriatric ED patients. A secondary objective was to define an evidence-based pretest probability of dementia based on published research and then estimate disease thresholds at which dementia screening is most appropriate. This systematic review was registered with PROSPERO (CRD42017074855).

Methods: PubMed, EMBASE, CINAHL, CENTRAL, DARE, and SCOPUS were searched. Studies in which ED patients ages 65 years or older for dementia were included if sufficient details to reconstruct 2 × 2 tables were reported. QUADAS-2 was used to assess study quality with meta-analysis reported if more than one study evaluated the same instrument against the same reference standard. Outcomes were sensitivity, specificity, and positive and negative likelihood ratios (LR+ and LR–). To identify test and treatment thresholds, we employed the Pauker-Kassirer method.

Results: A total of 1,616 publications were identified, of which 16 underwent full text-review; nine studies were included with a weighted average dementia prevalence of 31% (range, 12%–43%). Eight studies used the Mini Mental Status Examination (MMSE) as the reference standard and the other study used the MMSE in conjunction with a geriatrician's neurocognitive evaluation. Blinding to the index test and/or reference standard was inadequate in four studies. Eight instruments were evaluated in 2,423 patients across four countries in Europe and North America. The Abbreviated Mental Test (AMT-4) most accurately ruled in dementia (LR+ = 7.69 [95% confidence interval {CI} = 3.45–17.10]) while the Brief Alzheimer's Screen most accurately ruled out dementia (LR– = 0.10 [95% CI = 0.02–0.28]). Using estimates of diagnostic accuracy for AMT-4 from this meta-analysis as one trigger for more comprehensive geriatric vulnerability assessments, ED dementia screening benefits patients when the prescreening probability of dementia is between 14 and 36%. Conclusions: ED-based diagnostic research for dementia screening is limited to a few studies using an inadequate criterion standard with variable masking of interpreter's access to the index test and the criterion standard. Standardizing the geriatric ED cognitive assessment methods, measures, and nomenclature is necessary to reduce uncertainties about diagnostic accuracy, reliability, and relevance in this acute care setting. The AMT-4 is currently the most accurate ED screening instrument to increase the probability of dementia and the Brief Alzheimer's Screen is the most accurate to decrease the probability of dementia. Dementia screening as one marker of vulnerability to initiate comprehensive geriatric assessment is warranted based on test–treatment threshold calculations.