Epidemiologic studies of modifiable factors associated with cognition and dementia: systematic review and meta-analysis

Abstract

Background: Cognitive impairment, including dementia, is a major health concern with the increasing aging population. Preventive measures to delay cognitive decline are of utmost importance. Alzheimer's disease (AD) is the most frequent cause of dementia, increasing in prevalence from <1% below the age of 60 years to >40% above 85 years of age.

Methods: We systematically reviewed selected modifiable factors such as education, smoking, alcohol, physical activity, caffeine, antioxidants, homocysteine (Hcy), n-3 fatty acids that were studied in relation to various cognitive health outcomes, including incident AD. We searched MEDLINE for published literature (January 1990 through October 2012), including cross-sectional and cohort studies (sample sizes > 300). Analyses compared study finding consistency across factors, study designs and study-level characteristics. Selecting studies of incident AD, our meta-analysis estimated pooled risk ratios (RR), population attributable risk percent (PAR%) and assessed publication bias.

Results: In total, 247 studies were retrieved for systematic review. Consistency analysis for each risk factor suggested positive findings ranging from ~38.9% for caffeine to ~89% for physical activity. Education also had a significantly higher propensity for "a positive finding" compared to caffeine, smoking and antioxidant-related studies. Meta-analysis of 31 studies with incident AD yielded pooled RR for low education (RR = 1.99; 95% CI: 1.30-3.04), high Hcy (RR = 1.93; 95% CI: 1.50-2.49), and current/ever smoking status (RR = 1.37; 95% CI: 1.23-1.52) while indicating protective effects of higher physical activity and n-3 fatty acids. Estimated PAR% were particularly high for physical activity (PAR% = 31.9; 95% CI: 22.7-41.2) and smoking (PAR% = 31.09%; 95% CI: 17.9-44.3). Overall, no significant publication bias was found.

Conclusions: Higher Hcy levels, lower educational attainment, and decreased physical activity were particularly strong predictors of incident AD. Further studies are needed to support other potential modifiable protective factors, such as caffeine.

Figure 1

Flowchart of study selection for systematic review and meta-analysis. Notes: MEDLINE searches (1990–2012) included the following: (1) “Risk factor” as MESH term AND “Dementia” in title; (2) “Risk factor” as MESH term AND “Alzheimer” in title; (3) “Risk factor” as MESH term AND “Alzheimer” in title; (4) “Risk factor” as MESH term AND “cognitive” in title; (5) “Risk factor” in title and “cognitive” in title. Given that each search is not mutually exclusive of other searches, there were duplicates which were deleted from the final number of included studies. The following notations are defined follows: N1 = Studies excluded from all searches combined due to small sample size; N2 = Studies excluded from all searches combined due to design being neither cross-sectional nor cohort; N3 = Studies excluded from all searches combined due to being a review or a letter to the editor; N4 = Studies excluded from all searches combined due to lack of relevance to topic or hypothesis; N5 = Studies excluded from all searches combined for other reasons (e.g. special group of people); N6 = Final included studies; N6a = Final included cohort studies; N6b = Final included cross-sectional studies.

Figure 2

Main findings (%) of selected studies, given hypothesis: (A) Cohort Studies (B) Cross-sectional studies.

Figure 3

Meta-analysis of selected risk and protective factors for incident AD (n = 31). (A) Education. (B) Smoking status. (C) Physical activity. (D) Homocysteine. (E) n3 fatty acids.