243 Accuracy of Emergency Department Delirium Screening: A Diagnostic Meta-Analysis

Study Objectives To identify and summarize the pooled diagnostic test characteristics for dementia screening instruments in the emergency department (ED). Methods This was a systematic review and meta-analysis adherent to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis of Diagnostic Test Accuracy guidelines of prospective observational ED studies comparing appropriately brief dementia screening instruments against an acceptable criterion standard with sufficient detail to reconstruct 2x2 contingency tables. Two investigators independently assessed risk of bias using the Quality Assessment Tool for Diagnostic Accuracy Studies (QUADAS-2). When ≥1 study evaluated the same dementia screening instrument with the same criterion standard, pooled estimates of diagnostic accuracy were computed using a random-effects model and Meta-DiSc Version 1.4. Results From 1604 unique citations, 9 met inclusion criteria. QUADAS-2 assessment indicated moderate quality studies. The criterion standard for dementia was solely the Mini Mental Status Exam (MMSE) in 7 studies. The criterion standard for one of the other studies was a combination of the MMSE and a delirium screen, while the remaining study used a geriatrician’s assessment of Diagnostic Statistical Manual of Mental Disorders criteria as the gold standard. The weighted mean prevalence of dementia was 31% and ranged from 12% to 43% across studies. Eight instruments were described of which the Abbreviated Mental Test-4 (AMT-4) demonstrated the highest positive likelihood ratio [pooled LR+ 7.7, 95% CI 3.5-17.1] and the Brief Alzheimer’s Screen (BAS) demonstrated the lowest negative likelihood ratio [LR- = 0.10, 95% CI 0.02-0.28]. The Six Item Screener test time was reported as under 1 minute compared with 1.5 minutes for the Mini-Cog and 4.7 minutes for the AMT-4. Conclusions Existing research is limited by inadequate criterion standards juxtaposed with the reality that more widely acceptable comparators are impractical in ED settings. Acknowledging these limitations, the AMT-4 most accurately rules in dementia, while the BAS most accurately rules out dementia.