

18 Furthermore, patient-reported SMC did not predict future onset of dementia, whereas informant-reported SMC were associated with an increased risk of future dementia, a finding similar to a prior longitudinal study. One study of patients with mild and very mild Alzheimer’s type dementia ( AD) found informant complaints of memory loss (using a single question assessment) more reliably correlated with dementia than patient complaints ( sensitivity of informant complaints 98.1%, specificity 86.1%).

In contrast to patient-reported SMC, a small body of evidence suggests informant-reported SMC may better distinguish demented from non-demented individuals. 17 Of note, investigators did find greater discrepancies between patient interview results and chart notes among patients with more severe dementia. A multi-site primary care study of patients with largely mild dementia found patient-reported SMC also did not distinguish well between demented and non-demented participants.

16 However, these rates were likely a reflection of the method of SMC ascertainment, which relied on retrospective chart review. One community-based study found a very low sensitivity of 15.2%, but a relatively high specificity of 93.5%. We found several more recent studies that were not included in the Mitchell review. 37 both examined large community-based samples, but used different methods for eliciting SMC: 1 study used a 9-item questionnaire, 36 while the other study simply asked whether participants had had memory loss during the past year. Of the 3 studies that used a widely-accepted gold standard to establish a dementia diagnosis, 36–38 one was mainly a study of different subtypes of mild cognitive impairment (MCI) and the number of patients with dementia was very low. 41–43 Because of significant heterogeneity among studies, the pooled results cannot be used with any degree of confidence. 40 Three of the studies used only brief assessment methods to validate the diagnosis of dementia. 36–38 One of the studies used a less widely-accepted cognitive evaluation, 39 and another study used an informant interview to establish a diagnosis. Three studies established a dementia diagnosis using a widely-accepted gold standard. A number of the included studies had substantial flaws which reduce the strength of conclusions that can be drawn from this body of literature. 15 All studies were community-based and ranged in size from 156 to 3,220 participants. One recently published systematic review found 8 cross-sectional studies which compared elicited patient-reported SMC in demented and non-demented persons. Informant-reported memory complaints may better distinguish demented from non-demented individuals. In populations with low prevalence of dementia, the absence of SMC may have some utility in excluding a diagnosis of dementia given its relatively high negative predictive value in these settings. Patient-reported SMC were poorly sensitive for detecting dementia and only moderately specific. We found patient-reported SMC did not reliably distinguish demented from non-demented individuals. We examined cross-sectional studies comparing rates of SMC between persons with dementia and healthy elderly controls.Ĭonclusions from this body of evidence should be tempered by methodologic flaws in some of the studies, as well as variability across studies in populations included and methods used for memory complaint elicitation. 34, 35 The ability, however, of subjective memory complaints to discriminate effectively between healthy elderly adults and those with dementia is uncertain. Epidemiologic studies suggest subjective memory complaints (SMC) – in most cases elicited with single- or multi-item questionnaires, rather than spontaneous – are common in community-living elderly adults.
