Late-life Cognitive Activity and Dementia: A Systematic Review and Bias Analysis

Epidemiology. 2016 Sep;27(5):732-42. doi: 10.1097/EDE.0000000000000513.


Background: Engaging in late-life cognitive activity is often proposed as a strategy to delay or prevent Alzheimer's disease (AD) and other dementias. However, it is unclear to what extent the available evidence supports a causal effect of cognitive activity in dementia prevention.

Methods: We systematically searched PubMed and EMBASE through June 2014 to identify peer-reviewed epidemiologic studies of cognitive activity and incidence of AD or all-cause dementia. Eligible articles analyzed data from cohort or nested case-control studies, explicitly defined cognitive activity, evaluated participants for AD or all-cause dementia using clearly defined criteria, and provided effect estimates adjusted for at least age and sex. We describe methodologic issues and biases relevant to interpretation of these studies, and quantify the degree of bias due to confounding and reverse causation required to nullify typically observed associations.

Results: We reviewed 12 studies involving 13,939 participants and 1,663 dementia cases, of which 565 were specifically evaluated as AD. Most studies found associations between late-life cognitive activity and lower AD and/or all-cause dementia incidence. Differences in cognitive activity operationalization across studies precluded meta-analysis of effect estimates. Our bias analysis indicated that the observed inverse associations are probably robust to unmeasured confounding, and likely only partially explained by reverse causation.

Conclusion: Our systematic review and bias analyses provide support for the hypothesis that late-life cognitive activity offers some reduction in AD and all-cause dementia risk. However, more data are needed to confirm this relationship and on the optimal type, duration, intensity, and timing of that activity.

Publication types

  • Review
  • Systematic Review

MeSH terms

  • Alzheimer Disease / epidemiology*
  • Bias
  • Case-Control Studies
  • Causality
  • Cognition*
  • Cohort Studies
  • Confounding Factors, Epidemiologic
  • Dementia / epidemiology
  • Humans
  • Incidence
  • Proportional Hazards Models
  • Protective Factors