32 * SINGLE SCREENING QUESTIONS FOR COGNITIVE IMPAIRMENT IN OLDER PEOPLE; A SYSTEMATIC REVIEW

Abstract

Scope: Single item screening questions may be a practical first step in screening for cognitive problems in older people. A form of single item cognitive assessment is proposed in the NHS England and Wales dementia strategy. Our objective was to provide a review and synthesis of the evidence regarding the performance of single item tests for detection of cognitive impairment. Search methods: We performed a systematic review using Cochrane diagnostic test accuracy procedures. Two independent researchers searched for relevant papers across multiple, cross disciplinary electronic databases using previously validated search strings. Our index test of interest was any single screening question for cognitive impairment, including single items derived from factor analysis of multi-item tools. Our reference standards included clinical diagnosis (dementia, delirium and mild cognitive impairment) and accepted cut-offs on multi-domain cognitive assessments. We assessed risk of bias and external validity using the QUADAS-2 tool. Results: From 884 initial titles 11 studies were identified as eligible for inclusion in this review. Four studies were graded low risk of bias and recruited a representative patient sample. Clinical heterogeneity precluded formal meta-analysis. Five papers described a single screening question for cognitive impairment, each used a differing single question. Sensitivity ranged from 96% to 26%; specificity 45% to 100% depending on setting and question used. Six papers described component analysis of multiple item tools, in general broadly worded single questions around “decline in memory function”; “changes in ability to think and reason” or “learning new things” performed well. Conclusions: Informant based single item screening questions show promise for detection of cognitive impairment. However, there was substantial heterogeneity in format / application of single item screening tools and several papers had methodological or generalizability issues. Without further high quality test accuracy studies, we would not recommend large scale screening using a single item approach.