Fig. 2

A. Associations between total bilirubin concentrations and PC_{anthropometry} and PC_lipids among men in the UK Biobank. BMI: body mass index, WHR: waist-hip ratio, ApoA-I: apolipoprotein A, ApoB: apolipoprotein B, Lp (a): lipoprotein (a), LDL-C: low-density lipoprotein cholesterol, HDL-C: high-density lipoprotein cholesterol. PC1_{anthropometry}: General adiposity, PC2_{anthropometry}: Tall with low waist-to-hip ratio, PC3_{anthropometry}: Tall with high waist-to-hip ratio, PC4_{anthropometry}: High BMI and weight, with relatively small hip and waist circumference. PC1_lipids: Dyslipidemia, PC2_lipids: Anti-atherogenic, PC3_lipids: High levels of lipoprotein (a), PC4_lipids: High levels of triglycerides. Associations were estimated for each outcome by modelling log-transformed bilirubin with restricted cubic splines (3 knots at 10^th, 50^th, and 90^th percentile) allowing for non-linear associations (adjusted for age at recruitment, ethnicity, alcohol consumption, alcohol consumption frequency, smoking status, physical activity, liver enzyme (alanine transaminase), chronic diseases (heart problems and diabetes), medications (for cholesterol, blood pressure, diabetes, or exogenous hormones), qualifications, and ever use of hormones among women). P-values for non-linearity were computed with log-likelihood ratio tests comparing the spline model to a linear model. P-values < 0.001 were judged as evidence against linearity. Associations are presented as predicted mean values and 99% confidence intervals (CI). Levels of (total) bilirubin among men at 5^th, 10^th (1^st knot), 50^th (2^nd knot), 90^th (3^rd knot), and 95^th were 5.45, 6.08, 9.13, 15.51, and 19.42 on original scale (umol/L) and 1.70, 1.81, 2.21, 2.74, and 2.97 on log-scale, respectively. B Associations between total bilirubin concentrations and PC_{anthropometry} and PC_lipids concentrations among women in the UK Biobank. BMI: body mass index, WHR: waist-hip ratio, ApoA-I: apolipoprotein A, ApoB: apolipoprotein B, Lp (a): lipoprotein (a), LDL-C: low-density lipoprotein cholesterol, HDL-C: high-density lipoprotein cholesterol. PC1_{anthropometry}: General adiposity, PC2_{anthropometry}: Tall with low waist-to-hip ratio, PC3_{anthropometry}: Tall with high waist-to-hip ratio, PC4_{anthropometry}: High BMI and weight, with relatively small hip and waist circumference. PC1_lipids: Dyslipidemia, PC2_lipids: Anti-atherogenic, PC3_lipids: High levels of lipoprotein (a), PC4_lipids: High levels of triglycerides. Associations were estimated for each outcome by modelling log-transformed bilirubin with restricted cubic splines (3 knots at 10^th, 50^th, and 90^th percentile) allowing for non-linear associations (adjusted for age at recruitment, ethnicity, alcohol consumption, alcohol consumption frequency, smoking status, physical activity, liver enzyme (alanine transaminase), chronic diseases (heart problems and diabetes), medications (for cholesterol, blood pressure, diabetes, or exogenous hormones), qualifications, and ever use of hormones among women). P-values for non-linearity were computed with log-likelihood ratio tests comparing the spline model to a linear model. P-values < 0.001 were judged as evidence against linearity. Associations are presented as predicted mean values and 99% confidence intervals (CI). Levels of (total) bilirubin among women at 5^th, 10^th (1st knot), 50^th (2nd knot), 90^th (3rd knot), and 95^th were 4.46, 4.94, 7.26, 12.07, and 15.02 on original scale (umol/L) and 1.50, 1.60, 1.98, 2.49, and 2.71 on log-scale, respectively