Methods and results

From a representative population sample of 5368 men aged 45–59 years in South Wales and South West England 4861 (90.5%), participated in a baseline survey in 1979–83. Detailed questionnaire and clinical examination data were collected. Height was measured using a Holtain stadiometer. Study subjects have been flagged on the National Health Service Central Register in Southport and this analysis is based on events from enrolment up to the end of 1997. Self reports of possible cerebrovascular events were supplemented by medical records, including postmortem reports. Strokes were classified as either ischaemic or haemorrhagic by two independent assessors using standard methods with all available clinical and pathological information (unpublished data).

Analyses were carried out using Cox's proportional hazards models.

Full data were available for 4826 men of whom 287 had a stroke (264 ischaemic and 25 haemorrhagic). In univariate analyses men without stroke were taller than men with stroke 171.7 cm compared with. 170.3 cm (p = 0.0003). In age adjusted analyses, height as a continuous variable, was inversely associated with risk of stroke—both ischaemic and haemorrhagic, and fatal and non-fatal. After adjusting for diastolic blood pressure, diabetes, angina, body mass index, social class, smoking, and cholesterol, there was mild attenuation of the associations, with the inverse associations for all stroke no longer attaining significance, but there were statistically significant inverse associations for ischaemic, and fatal ischaemic strokes (table1). Smaller non-significant inverse associations were found for non-fatal ischaemic and haemorrhagic stroke (data not shown).

With height as a categorical variable subjects in the tallest quartile had a 30% and a more than 50% lower risk of ischaemic and fatal ischaemic stroke respectively compared with those in the shortest quartile (table 1). Trends were significant for all strokes, ischaemic strokes and fatal ischaemic strokes. The hazard ratios for haemorrhagic stroke were unstable as they were based on small numbers, and are not presented.

Comment

Our results show that tall men, as well having less risk of stroke, particularly ischaemic stroke, have a much reduced risk of a fatal ischaemic cerebrovascular event. The results are consistent with recently reported findings on all stroke subtypes combined.1 2 However, they contrast with a recent British study that found only a weak and non-significant inverse association.3 In that study the authors hypothesised that the lack of association was because height is more strongly related to reduced risk of haemorrhagic rather than ischaemic stroke, and the association was not apparent as data on stroke subtypes were not available. Although the Norwegian study did find such an association, data for both sexes were combined and adjustment for socioeconomic position was not possible.1 Our findings do not support this hypothesis.

We have shown previously that deprivation during childhood increased the risk of stroke mortality in adulthood, independent of later socioeconomic circumstances.4 Here we show that height is inversely associated with ischaemic stroke risk, and that other risk factors in adulthood and socioeconomic position do not explain this association. Fetal growth, early life socioenvironmental circumstances, infection and nutrition in childhood are related to adult height and thus may determine stroke risk. Trends in these early life factors may explain the downward secular trends in stroke mortality observed in many countries.