Predictors and prognosis of population-based subjective cognitive decline: longitudinal evidence from the Caerphilly Prospective Study (CaPS).

OBJECTIVES: To understand associations between the subjective experience of cognitive decline and objective cognition. This subjective experience is often conceptualised as an early step towards neurodegeneration, but this has not been scrutinised at the population level. An alternative explanation is poor meta-cognition, the extreme of which is seen in functional cognitive disorder (FCD). DESIGN: Prospective cohort (Caerphilly Prospective Study). SETTING: Population-based, South Wales, UK. PARTICIPANTS: This men-only study began in 1979; 1225 men participated at an average age of 73 in 2002-2004, including assessments of simple subjective cognitive decline (sSCD, defined as a subjective report of worsening memory or concentration). Dementia outcomes were followed up to 2012-2014. Data on non-completers was additionally obtained from death certificates and local health records. PRIMARY AND SECONDARY OUTCOME MEASURES: The primary outcome measure was incident dementia over 10 years. Secondary outcome measures included prospective change in objective cognition and cross-sectional cognitive internal inconsistency (the existence of a cognitive ability at some times, and its absence at other times, with no intervening explanatory factors except for focus of attention). RESULTS: sSCD was common (30%) and only weakly associated with prior objective cognitive decline (sensitivity 36% (95% CI 30 to 42) and specificity 72% (95% CI 68 to 75)). Independent predictors of sSCD were older age, poor sleep quality and higher trait anxiety. Those with sSCD did not have excess cognitive internal inconsistency, but results suggested a mild attentional deficit. sSCD did not predict objective cognitive change (linear regression coefficient -0.01 (95% CI -0.13 to 0.15)) nor dementia (odds ratio 1.35 (0.61 to 2.99)) 10 years later. CONCLUSIONS: sSCD is weakly associated with prior objective cognitive decline and does not predict future cognition. Prior sleep difficulties and anxiety were the most robust predictors of sSCD. sSCD in the absence of objective decline appears to be a highly prevalent example of poor meta-cognition (ie, poor self-awareness of cognitive performance), which could be a driver for later FCD.