A cross-country analysis of episodic memory and (potentially) modifiable risk factors of dementia

The widely cited Lancet Commission concluded that 40% of dementia cases may be preventable through interventions targeting what they refer to as modifiable risk factors. These risk factors have been widely studied individually, but rarely investigated collectively and across many countries. If these factors are "true" (i.e., impactful) modifiable risk factors, then their independent relationship should be robust across countries and comorbidities. We analyze the cross-country consistency of relationships between these modifiable risk factors and episodic memory, a common predictor of cognition and dementia. Using internationally comparable aging studies in 31 countries including the United States, England and Europe, we estimate regressions of combined immediate and delayed word recall with modifiable risk factors and demographic characteristics. Cross-country differences in culture, policies, economy, and other collective experiences lead to significant variation in lifecycle outcomes, including cognitive decline and modifiable risk factors. Our approach does not conclusively affirm a causal relationship but can identify relationships that are weak or nonexistent. We find a limited number of robust relations: education, depression, and hearing loss show clear, consistent associations with our cognition measure. The evidence for other factors, including obesity, smoking, diabetes, and hypertension is weaker and becomes almost non-existent when correcting for multiple hypotheses testing. The inconsistent relationships across countries between episodic memory and obesity, smoking, diabetes, and hypertension suggest the lack of a causal mechanism leading to cognitive decline - a necessary condition for these risk factors to be modifiable and effective targets for policy interventions aimed at controlling dementia prevalence and cost.