Complexity of the relation between physical activity and stroke: a meta-analysis.

Abstract

OBJECTIVE: To explore the associations of recreational and occupational physical activity with ischemic and hemorrhagic stroke by means of meta-analysis. DATA SOURCES: PUBMED was searched for studies published in English up to December 2001. Keywords used were physical activity, exercise, cerebrovascular disease, stroke, ischaemic stroke, and haemorrhagic stroke. Reference lists were reviewed for additional studies. STUDY SELECTION: Cohort and case-control studies investigating the relation between physical activity and stroke were identified (n = 36). The article reporting the longest follow-up time was included if there was > or =1 report of the same study (4 exclusions). Studies had to report a measure of physical activity. A study that measured physical fitness was excluded. DATA EXTRACTION: Details of study methods, samples, levels of physical activity, types of stroke, and estimates of risk (relative risk, RR) were extracted independently by 3 reviewers from the 24 cohort and 7 case-control studies that were included. Authors of relevant studies were contacted to obtain missing data. Risk estimates were standardized. Risks for subgroups within the samples (e.g., women and men) were treated as separate study units. Physical activity was separated into occupational and leisure activity. MAIN RESULTS: In studies of occupational activity, risk of hemorrhagic stroke (RR, 0.31; 95% CI, 0.13-0.76; 1 study) and risk of ischemic stroke (RR, 0.57; CI, 0.43-0.77; 5 studies) were lower for people most active versus inactive at work, risk of ischemic stroke was lower for people who were most active versus moderately active (RR, 0.77; CI, 0.60-0.98; 5 studies), and risk of total stroke was lower for people moderately active versus inactive (RR, 0.64; CI, 0.48-0.87; 4 studies), but none of the other 8 comparisons of level of occupational activity showed significant differences. In studies of leisure-time activity, risk of total stroke (RR, 0.78; CI, 0.71-0.85; 19 studies), risk of hemorrhagic stroke (RR, 0.74; CI, 0.57-0.96, 9 studies), and risk of ischemic stroke (RR, 0.79; CI, 0.69-0.91; 11 studies) were lower for people most active versus inactive. Risk of total stroke was lower for people moderately active versus inactive (RR, 0.85; CI, 0.78-0.93; 15 studies), but none of the other 7 comparisons showed significant differences. In regression analysis weighted by methodologic quality, RR of stroke for active persons was slightly lower, with narrower confidence intervals. Leisure-time activity compared with inactivity was associated with a greater reduction in risk of hemorrhagic stroke in men (RR, 0.54; CI, 0.36-0.81) than in women (RR, 0.76; CI, 0.67-0.86). Studies conducted in Europe found more effect of activity (RR, 0.47; CI, 0.33-0.66; 3 studies) than US studies (RR, 0.82; CI, 0.75-0.90). Study results did not differ by type of study or year of publication. CONCLUSIONS: Recreational and occupational physical activity were both associated with reductions in risk of stroke. Studies varied widely in their estimates of risk; lower risks were found for active versus inactive men than women and for European versus US studies.