Associations of reproductive and chronological ageing with cardiovascular measures including anthropometry, blood pressure, lipids and C-reactive protein and glucose. Figure 3(a) shows the main analysis results: mean difference in cardiovascular measures per year for reproductive and chronological age with mutual adjustment for each age variable and pre-pregnancy body mass index (BMI), age at menarche, parity, maternal education, smoking status, and alcohol intake, undertaken in 1702 women with 4734 observations. The best fitting model between reproductive age and fat mass and non-HDL-c was reproductive age to the power of a half and the addition of a cubic term respectively; however, the linear term is shown here for completeness. The non-linear models are graphically shown in Supplementary Figure S1. The p-value for interaction, a test for any difference in the linear association between any of the three menopause stages, is reported in Table S6 (and by age in Table S7). Figure 3(b) shows the additional analyses undertaken to explore possible selection bias in the main analysis. This shows the mean difference in cardiovascular measures per year in chronological age by menopausal stage (pre, peri or post), undertaken in all eligible women (total N = 3892 women with 9841 observations, with 2313 (4118), 1666 (2388) and 1550 (3335) women (observations) contributing data to pre, peri, and post-menopause, respectively).