Risk of stroke and lifestyle

BMJ 2009; 338 doi: https://doi.org/10.1136/bmj.b14 (Published 20 February 2009) Cite this as: BMJ 2009;338:b14
  1. Matthew F Giles, senior research fellow and consultant physician
  1. 1Stroke Prevention Research Unit, Biomedical Research Centre, John Radcliffe Hospital, Oxford OX3 9DU
  1. matthew.giles{at}clneuro.ox.ac.uk

    Smoking, exercise, alcohol, and diet are predictive factors

    The current mortality and morbidity attributable to cerebrovascular disease is of concern nationally and globally. Moreover, the overall burden of stroke is expected to increase over the next two to four decades. More people will reach older age, when the risk of stroke rises sharply,1 and this will offset the fall in the age specific incidence of stroke reported by recent epidemiological studies.2 3 In the linked study (doi:10.1136/bmj.b349), Myint and colleagues prospectively assess the relation between lifestyle behaviours and incident stroke in a cohort of 20 040 men and women aged 40-79 years.4

    How should individuals, policy makers, and society respond to this challenge? Benjamin Franklin said that an ounce of prevention is worth a pound of cure, and fortunately stroke is both predictable and preventable. More than 15 years ago, the Framingham stroke risk score—based on age, blood pressure, use of antihypertensives, diabetes mellitus, cigarette smoking, previous cardiovascular disease, atrial fibrillation, and left ventricular hypertrophy—was derived to calculate an individual’s risk and “provide the impetus for risk factor modification.”5 Such vascular risk scores have underpinned the concept of targeting high risk people with preventive treatments.

    However, the contribution of lifestyle behaviours such as physical activity and diet to the risk of stroke is less clear. This is because they contribute less to the risk of stroke than other factors and require large long term studies to show an association, and also because such studies may be confounded by the heterogeneity of stroke, with risk factors working in different directions for ischaemic stroke and haemorrhagic stroke. Indeed, until two years ago, no studies had looked at the combined effect of lifestyle behaviours on the risk of stroke. However, the contribution of lifestyle behaviours to stroke is important because modifying these weak risk factors across a population has a greater potential for overall reduction in stroke than modifying more powerful risk factors (such as carotid stenosis and atrial fibrillation) in a smaller number of people.6

    Myint and colleagues assessed the relation between lifestyle behaviours and incident stroke in men and women participating in the European Prospective Investigation into Cancer Study (EPIC).4 At baseline, between 1993 and 1997, patients completed a health and lifestyle survey, were examined by a nurse, and gave a blood sample. Over an average follow-up of 11.5 years, 599 strokes were ascertained from centrally collected death certificates and hospital discharge coding data. The authors constructed a five point lifestyle behaviour score on the basis of smoking, physical activity, alcohol intake, and plasma vitamin C concentrations, which are indicative of fruit and vegetable intake. An increasing score indicated more protective behaviour.

    The risk of stroke decreased linearly with increasing scores in three models, with differing degrees of multivariate adjustment. In the final model, the risk of stroke was 2.3 times greater in people with a score of zero compared with those with a score of four (P trend <0.001), after adjusting for age, sex, body mass index, blood pressure, cholesterol, diabetes, use of aspirin, and social class. The associations between stroke and physical activity, stroke and alcohol consumption, and stroke and vitamin C concentrations differed between the sexes—each reached statistical significance in women but not in men.

    The authors acknowledge some of the limitations of the study. Ischaemic and haemorrhagic strokes were not differentiated; vitamin C concentration was only a surrogate marker for fruit and vegetable intake; the elements of the score were dichotomised, which would remove any subtle association between individual factors and risk; and 9000 participants were excluded from the analysis because of missing data. Despite these limitations, the simple conclusion that the risk of stroke is associated with combined health behaviour is important. Importantly, although this association is probably causal, so that lifestyle modifications will modify the risk of stroke, only interventional studies can prove this.

    These findings should be interpreted in conjunction with other recent reports. The Health Professionals’ Follow-up Study and Nurses’ Health Study reported that the risk of stroke was associated with a similar lifestyle score to that used in EPIC, but that used body mass index instead of vitamin C concentrations.7 Among the 43 865 men and 71 243 women who had 1559 strokes during follow-up, the risk of stroke was 3.2 and 4.7 times greater in those with all risk factors than in those with none. The Women’s Health Study was a randomised controlled trial of 39 876 women who had 450 strokes,8 and risk was 2.2 times higher in those with the highest lifestyle scores than in those with the lowest scores. The composition of these scores was similar to the EPIC score but used more detailed dietary information and body mass index.

    It is encouraging that the association between the risk of stroke and combined health behaviour is consistent across different populations, and between observational and randomised controlled trials. The conclusion that lifestyle predicts the risk of stroke should help to inform individuals’ choices and policy makers’ decisions. However, what is also consistent but less encouraging is the small proportion of participants with a lifestyle that protects against stroke—although lifestyle interventions could be of great benefit, a huge shift in behaviour will be needed to achieve this.


    Cite this as: BMJ 2009;337:b14



    View Abstract