Smoking and inactivity dominate dementia risk; caffeine's case stays thin

Dementia prevalence climbs steeply with age, from under 1% below 60 to over 40% above 85 for Alzheimer's disease, and there is no cure to lean on. That makes prevention the main lever available, which in turn makes it important to know which everyday factors genuinely track with cognitive outcomes and which merely get talked about. Education, smoking, alcohol, physical activity, caffeine, antioxidants, homocysteine and n-3 fatty acids have all been studied, with mixed results. This review set out to compare how consistently each one shows up and to size the strongest ones.

The authors searched MEDLINE for literature published from January 1990 through October 2012, including both snapshot studies and studies that followed people over time, restricted to samples larger than 300. In total 247 studies were retrieved for systematic review, and the consistency of positive findings was compared across factors, study designs and study characteristics. A subset of 31 studies reporting new cases of Alzheimer's disease was then pooled to estimate risk ratios and the percentage of cases attributable to each factor at population level. Publication bias was formally assessed.

Consistency of positive findings ranged from about 38.9% for caffeine to about 89% for physical activity, with education also significantly more likely than caffeine, smoking or antioxidant studies to yield a positive finding. In the pooled analysis of new Alzheimer's cases, low education carried a risk ratio of 1.99, high homocysteine 1.93 and current or ever smoking 1.37, while higher physical activity and n-3 fatty acids pointed towards protection. The estimated share of cases attributable to the factor was particularly high for physical activity at 31.9% and smoking at 31.09%. No significant publication bias was detected.

The findings reinforce the now familiar message that vascular and lifestyle factors sit close to the centre of dementia risk, and they add a useful ranking by how consistently each factor replicates. They also complicate the popular caffeine story, which emerged as the least consistent of the factors examined and the one the authors singled out as needing more work. Because all the underlying data come from following or surveying people rather than assigning them to different lives, reverse causation remains possible: early cognitive decline could reduce activity rather than the other way round. The literature searched also ends in 2012, so more recent trials are not represented.

This is a reason to treat regular physical activity and avoiding smoking as brain-relevant habits rather than purely cardiovascular ones, since they were both the most consistent and the largest contributors here. Low education and high homocysteine were also strongly associated with new Alzheimer's cases, though education is not something a reader can change retrospectively. The caffeine result is a reminder that a widely repeated claim can rest on the shakiest evidence in the set. None of this promises that changing a habit changes your own risk; it describes what large populations show.