The clinical significance of subjective memory complaints in the diagnosis of mild cognitive impairment and dementia: a meta‐analysis

Abstract

Subjective memory complaints (SMC) are frequently reported by individuals with objective evidence of cognitive decline although the exact rate of complaints and their diagnostic value is uncertain. A meta-analysis was conducted for all studies examining SMC and either concurrent dementia or mild cognitive impairment (MCI). Eight studies reported the rate of SMC in dementia, seven studies reported the rate of SMC in MCI and of these four compared the rate of SMC in dementia and MCI head-to-head. SMC were present in 42.8% of those with dementia and 38.2% of those with MCI. Across all levels of cognitive impairments 39.8% of people had SMC compared with 17.4% in healthy elderly controls (Relative Risk 2.3). In head-to-head studies there was a significantly higher rate of SMC in dementia vs MCI (48.4% vs 35.1%). Examining the diagnostic value of SMC in dementia, the meta-analytic pooled sensitivity was 43.0% and specificity was 85.8%. For MCI, meta-analytic pooled sensitivity was 37.4% and specificity was 86.9%. In community studies with a low prevalence the positive and negative predictive values were 18.5% and 93.7% for dementia and 31.4% and 86.9% for MCI. The clinical utility index which calculates the value of a diagnostic method suggested 'poor' value for ruling in a diagnosis of dementia but 'good' value for ruling out a diagnosis. When assessed by simple questions, SMC appear to be present in the minority of those with mild cognitive impairment and dementia. In cross-sectional community settings, even when people agree that they have SMC there is only a 20% or 30% chance that dementia or MCI are present, respectively. Despite this, the absence of SMC may be a reasonable method of excluding dementia and MCI and could be incorporated into short screening programs for dementia and MCI but replication is required in clinical settings.