Preclinical Alzheimer disease (AD) has been associated with subtle deficits in memory, attention, and spatial navigation (Allison et al., 2019; Aschenbrenner et al., 2015; Hedden et al., 2013). There is a need for a widely distributable screening measure for detecting preclinical AD. The goal of this study was to examine whether self- and informant-reported change in the relevant cognitive domains, measured by the Everyday Cognition Scale (ECog; Farias et al., 2008), could represent robust clinical tools sensitive to preclinical AD.

Clinically normal adults aged 56-93 (n=371) and their informants (n=366) completed memory, divided attention, and visuospatial abilities (which assesses spatial navigation) subsections of the ECog. Reliability and validity of these subsections were examined using Cronbach’s alpha and confirmatory factor analyses (CFA). The hypothesized CFA assumed a three-factor structure with each subsection representing a separate latent construct. Receiver operating characteristics (ROC) and area under the curve (AUC) analyses were used to determine the diagnostic accuracy of the ECog subsections in detecting preclinical AD, either defined by cerebrospinal fluid (CSF) ptau181/Aß42 ratio >0.0198 or hippocampal volume in the bottom tertial of the sample. Hierarchical linear regression was used to examine whether ECog subsections predicted continuous AD biomarker burden when controlling for depressive symptomatology, which has been previously associated with subjective cognition (Zlatar et al., 2018). Lastly, we compared the diagnostic accuracy of ECog subsections and neuropsychological composites assessing the same or similar cognitive domains (memory, executive function, and visuospatial ability) in identifying preclinical AD.

All self- and informant-reported subsections demonstrated appropriate reliability (a range=.71-.89). The three-factor CFA models were an adequate fit to the data and were significantly better than one-factor models (self-report x2(3)=129.511, p<.001; informant-report X2(3)=145.347, p<.001), suggesting that the subsections measured distinct constructs. Self-reported memory (AUC=.582, p=.007) and attention (AUC=.564, p=.036) were significant predictors of preclinical AD defined by CSF ptau181/Aß42 ratio. Self-reported spatial navigation (AUC=.592, p=.022) was a significant predictor of preclinical AD defined by hippocampal volume. Additionally, self-reported attention was a significant predictor of the CSF ptau181/Aß42 ratio (p<.001) and self-reported memory was a significant predictor of hippocampal volume (p=.024) when controlling for depressive symptoms. Informant-reports were not significant predictors of preclinical AD (all ps>.074).

There was a nonsignificant trend for the objectively measured executive function AUC to be higher than for self-reported attention in detecting preclinical AD defined by CSF ptau181/Aß42 ratio and was significantly higher than self-reported attention in detecting preclinical AD defined by hippocampal volume (p=.084 and p<.001, respectively). For memory and spatial navigation/visuospatial domains, the AUCs for self-reported and objective measures did not differ in detecting preclinical AD defined by either CSF ptau181/Aß42 ratio or hippocampal volume (ps>.129).

Although the self-reported subsections produced significant AUCs, these were not high enough to indicate clinical utility based on existing recommendations (all AUCs<.60; Mandrekar, 2010). Nonetheless, there was evidence that self-reported cognitive change has promise as a screening tool for preclinical AD but there is a need to develop questionnaires with greater sensitivity to subtle cognitive change associated with preclinical AD.