Arterial stiffness is an important marker of cardiovascular health. Recent evidence from cross-sectional research has suggested it is associated with alcohol consumption. Research that employs a longitudinal perspective may be better equipped to evaluate the nature of this relationship and in particular to determine whether alcohol consumption is linked to the progression of arterial stiffness over time. The current study will consequently implement a longitudinal cohort design to evaluate the association between long-term alcohol consumption patterns and changes in arterial stiffness. Data will be drawn from the Whitehall II cohort study of British civil servants, in which participants completed repeat pulse wave velocity (PWV) assessments of arterial stiffness across a four-to-five year interval. Repeat measurements of volume of alcohol intake were also recorded for participants, extending back across more than two decades. Intake will be categorised in such a way as to distinguish between different alcohol consumer types, including non-drinkers and former drinkers. Linear mixed effects models will be used with adjustment for potential confounds, such as age, diabetes, mean arterial pressure and heart rate. Results from the modelling work will illustrate the extent and form of the association between alcohol intake and PWV. This work will provide useful insights into the role that alcohol intake plays in the longitudinal progression of an important cardiac marker, and it will have implications for our understanding of alcohol's relationship to cardiovascular health in the general population.

Arterial stiffness occurs where the vessel wall lacks elasticity. It is known to be independently associated with both cardiovascular morbidity and mortality, such that it can act as a surrogate end point for studies of cardiovascular disease. Pulse wave velocity (PWV) is an index of this stiffness and has been described as its 'gold standard' measurement.

A number of other clinical factors have been found to show meaningful association with PWV, including heart rate and mean arterial pressure. These findings have highlighted the need for studies of new potential determinants of PWV to account for the potentially confounding influence of these known factors. Additional demographic and lifestyle characteristics have also been shown to act as moderators of PWV. In particular, research has demonstrated that PWV increases with age, and this stiffness is consequently deemed an indicator of vascular ageing.

Alcohol intake is another lifestyle characteristic that has been considered as a potential factor in the loss of arterial elasticity. Evidence to date suggests that a J-shaped association may exist between such intake and PWV, but this work has mostly relied on cross-sectional data or looked at short-term intake patterns only. By adopting a longitudinal perspective, further insight may be garnered into this important epidemiological issue. Consequently, the new study herein proposed will address this research need and investigate how different patterns of alcohol consumption relate to arterial stiffness and to the progression of this stiffness over time. The primary objectives of this study are to (1) examine whether long-term patterns of alcohol consumption are independently associated with a baseline assessment of carotid-femoral PWV and (2) whether alcohol intake is likewise associated with longitudinal changes in this important cardiovascular marker.

Longitudinal alcohol exposure will be determined from reported consumption at phases 1 (1985-1988) through 9 (2007-2009) of the Whitehall II Cohort Study and categorised according to the latest UK recommendations on alcohol intake limits. These guidelines define moderate consumption as weekly ethanol intake volumes of 112g or under for both males and females. As alcohol consumption is recorded in UK units in the Whitehall II database, conversion will be performed using the standard ratio of 1 unit to 8 grams of ethanol prior to categorisation. The subsequent categorisation will draw distinctions between non-drinkers and former drinkers, as well as between moderate and heavy consumers. Those with unstable patterns of consumption will also be identified as there is emerging evidence that such consumers may have a different cardiovascular risk profile compared to consumers with more longitudinally consistent intake patterns.

Demographic and clinical characteristics will be investigated to provide context for the subsequent inferential analyses. As research has also suggested that participant sex can moderate the relationship between alcohol and PWV, this will be tested in the current analysis (through preliminary modelling that includes sex as part of an interaction term with level of alcohol consumption). If this moderation effect is replicated, then our core result reporting will be stratified according to sex.

Linear mixed effects models will then be used to examine the effect of long-term alcohol consumption on PWV and its effect on subsequent longitudinal changes in this cardiovascular marker. This modelling approach is particularly suited to analysing the complex data structure under investigation, including the correlation across individual participants' paired measurements. A detailed analytic protocol is available upon request.

Supplementary linear mixed effects modelling will also be performed to look at the association between short-term alcohol consumption patterns and PWV. This will employ the same modelling strategy as used in the primary analysis above, with the categorisation of consumption utilising reported intake at the most recent assessment phase only. This supplementary analysis will offer additional context for the findings of the primary analysis and provide further detail on the relationship that different forms of alcohol consumption have with arterial stiffness.

Through this comprehensive and detailed analysis, it is hoped that new insights on this important issue of cardiovascular health will be obtained.

Inclusion Criteria:

• Participated in Whitehall II cohort study
• Reported alcohol consumption on a minimum of one occasion between phase 1 and phase 9 of the Whitehall II cohort study

Exclusion Criteria:

• Left Whitehall II cohort study prior to phase 9 or did not participate in PWV assessment at Phase 9
• Had history of coronary heart disease

Data will be drawn from the Whitehall II cohort study database. This incorporates longitudinal assessments of 6,895 male (67%) and 3,413 female British civil servants, originally recruited between 1985 and 1988. The overall age range was 34 to 56 years. Of these, 897 were excluded due to having a history of cardiovascular disease. Clinical assessments were repeated at 4-to-5 year intervals. PWV measurement was introduced to the protocol at phase 9, with assessments taking place during 2008 and 2009. 3,869 of the sample participants had a successful PWV assessment during this time. Follow-up testing took place at phase 11, during 2012 and 2013. Of the participants who underwent a PWV assessment at phase 9, 3,130 also had a successful PWV assessment at phase 11. This group included 2,350 males (75%) and 780 females.

Researchers can request access to anonymised data: http://www.ucl.ac.uk/whitehallII/data-sharing