Association Between Change In Leisure-time Physical Activity During Post-retirement Period And Incident Stroke: A Prospective Cohort Study

Objective:To examine the association between change in leisure-time physical activity（LTPA）and risk of incident stroke among retired workers,and the joint effect of baseline and change in LTPA on stroke risk.Methods:The Dongfeng-Tongji（DFTJ）cohort recruited 24175 retired workers who participated in both 2008 baseline and the 2013 follow-up visit.After excluding participants with cardiovascular disease（CVD;including coronary heart disease and stroke）,severely abnormal electrocardiogram or cancer in 2013,and those with missing or unreliable data for LTPA,12644 participants remained for the present study.According to standardized questionnaires,we defined change in LTPA as the difference of the self-reported hours spent on LTPA each week between two visits（i.e.,self-reported hours spent on LTPA each week in 2013 minus that in 2008）and further categorized participants into three groups as decrease(<-5 hours/week;20^th percentiles),maintain(-5 to 7 hours/week;20^th-80^th percentiles),and increase(>7 hours/week;80^thpercentiles).Multivariable-adjusted Cox proportional hazards regression models were used to yield hazard ratios（HRs）and 95%confidence intervals（95%CIs）for the association between change in LTPA and risk of incident stroke.Baseline LTPA level was classified into three categories as low（<3.5 hours/week;20th percentiles）,medium（3.5 to 14 hours/week;20th-80th percentiles）,and high（>14 hours/week;80th percentiles）.We then created joint exposure categories by cross-classifying categories of each exposure categories,with low-maintain（consistently inactive individuals with insufficient LTPA at both visits）as reference.In addition,change in waist-to-height ratio（WHt R）was defined as decrease（<-0.02;20th percentiles）,maintain（-0.02 to 0.05;20th-80th percentiles）and increase（>0.05;80th percentiles）.We then created joint exposure categories by cross-classifying categories of each exposure categories,with low-maintain（participants with maintained WHt R levels and LTPA levels）as reference.Results:During 68476 person-years of follow-up,we documented 549 incident stroke cases,including 434 incident ischemic stroke（IS）cases and 115 incident hemorrhagic stroke（HS）cases.Compared with participants who maintained their LTPA levels,those with increased LTPA levels（>7 hours/week）had a significant lower risk of incident total stroke（HR=0.75;95%CI:0.59,0.95）and IS（HR=0.74;95%CI:0.57,0.98）.When stratified by age,the association of increasing LTPA with incident stroke was more evident among individuals who were<65 years of age compared with their older counterparts[HR:0.40（0.22,0.73）versus 0.84（0.65,1.10）,P_interaction=0.035].In addition,compared with consistently inactive individuals（insufficient LTPA at both visits）,those reported low LTPA levels at baseline but increased their LTPA levels at the2013 survey experienced a lower risk of incident stroke（HR=0.58;95%CI:0.34,0.98）.The analysis of the joint association of change in WHt R and change in LTPA showed that among participants with decreased WHt R levels（change in WHt R<-0.02）,those with increased LTPA levels had a 43%lower risk of stroke（HR=0.57;95%CI:0.32,0.99）compared with participants who maintained their LTPA levels.Similar association was found in participants who maintained their WHt R levels（change in WHt R:-0.02～0.05）but increased their LTPA levels（HR=0.68;95%CI:0.46,1.00）.Conclusions:Increasing LTPA levels during post-retirement period was associated with a lower risk of incident stroke.Baseline LTPA and change in LTPA might reduce stroke risk jointly.Retired individuals should be encouraged to participate in LTPA more frequently to lessen future risk of incident stroke.