Diet and cardiovascular disease risk in Chinese

Abstract

An unhealthy diet is a risk factor for cardiovascular disease (CVD). Most evidence concerning the cardiovascular effect of diet is from observational studies in Western settings, where dietary reporting is open to under-reporting. Bias caused by residual confounding is also possible as higher socio-economic position is often associated with a different diet and a lower CVD risk. Furthermore, people may adjust their diet in response to poor health. Evidence from non-Western settings with potentially different dietary reporting and patterns of confounding may help identify whether these observed associations in the West are biologically based targets of intervention or contextually specific results stemming from information bias and unmeasured confounding. This thesis took advantage of a large study of older adults (50+ years) from the developing non-Western setting of Guangzhou in Southern China to examine the associations of key aspects of diet that feature heavily in dietary guidelines (milk, fruit, vegetable and nut consumption) with CVD risk. In the Guangzhou Biobank Cohort Study (GBCS), first from a food frequency questionnaire (FFQ), I assessed the proportion of dietary under- and over-reporters based on the ratio of reported energy intake to basal metabolic rate. Then I examined the adjusted associations of milk, fruit, vegetable and nut consumption with CVD risk and risk factors cross-sectionally and prospectively. CVD risk was proxied by Framingham score, calculated from sex, age, LDL-cholesterol, HDL-cholesterol, blood pressure, and history of diabetes. CVD risk factors considered were components of the Framingham score. Multivariable linear regression was used to assess the adjusted associations of dietary items with Framingham score. Multivariable censored linear regression was used to assess adjusted associations with CVD risk factors. Given eating habits may change in response to ill-health, I also considered whether associations varied by health status. Only 10.9% of GBCS participants were classified as under-reporters, and 9.1% as over-reporters. Cross-sectionally: (1) whole cow’s milk consumption was associated with lower Framingham score among all participants, (2) fruit consumption was associated with lower Framingham score among participants in poor health, while associated with higher Framingham score among participants in good health, (3) vegetable consumption was associated with higher Framingham score among all participants and (4) nut consumption was associated with slightly lower Framingham score among participants in good health. Prospectively, adjusted for baseline CVD risk or risk factors, milk, fruit, vegetable or nut consumption was barely associated with Framingham score or its components. In this recently developed Chinese population, few were obviously under- or over-reporting energy intake assessed from a FFQ, which might make this setting a particularly suitable setting to assess the association of diet with health. Milk, fruit, vegetable and nut consumption were not consistently associated with lower CVD risk, cross-sectionally or prospectively, in contrast to findings from most observational studies from the West, which might indicate that the associations observed in the West may be contextually specific, and that contextually specific recommendations, based preferably on experimental evidence, are required.