Plant-based diet quality, fat mass, and cardiovascular disease: a mediation analysis of mid-aged adults in the UK Biobank

Evidence supports plant-based dietary patterns for preventing cardiovascular diseases (CVD)(1). Fat mass is a strong predictor of CVD(2), however it’s unclear whether this mediates the relationship between plant-based dietary patterns and CVD. Hence, the aim of this study was to determine if longitudinal associations between plant-based dietary patterns and incidence of CVD events, CVD mortality and all-cause mortality are mediated by fat mass in mid-life. Dietary data (Oxford WebQ) from 14,247 adults (median 56 years [IQR 49–61]) in the UK Biobank cohort study were used to derive diet quality index scores for an overall plant-based diet (PDI), a healthy plant-based diet (hPDI), and a less healthy plant-based diet (uPDI). Health registries and national records provided CVD event and mortality data. Percentage fat mass was measured by dual X-ray absorptiometry. Cox proportional hazard ratios (95% CI) identified associations between each diet quality index and CVD events, CVD mortality or all-cause mortality. Regression-based mediation analysis was used to identify the direct effect (plant-based diet quality indices on CVD mortality, CVD events, or all-cause mortality), and the indirect effect, which was mediated by fat mass. New CVD events (n = 364), CVD mortality (n = 52) and all-cause mortality (n = 220) were identified with mean follow up of 11.6 (SD ± 0.4) years for CVD events, and 11.5 (SD ± 0.7) years for mortality. The mean score for PDI was 50.5 (SD ± 5.9), hPDI 52.8 (SD ± 7.2), and uPDI 54.0 (SD ± 6.8). The PDI and hPDI were inversely associated with fat mass, and the uPDI was positively associated with fat mass (p < 0.001). There was no association between the diet quality indices and health outcomes, with (direct effect) or without (total effect) the fat mass mediator for males and females (p ≥ 0.1). Fat mass was associated with risk of mortality in some models, after controlling for the indices, such as a lower risk of all-cause mortality after controlling for the hPDI (p < 0.1). There was a significant negative indirect effect of hPDI on CVD mortality via fat mass for females only (observed coefficient 0.81; 95% CI 0.60–0.99). Overall, there was limited evidence of a mediating effect from fat mass in the association between plant-based dietary patterns and incidence of CVD events, CVD mortality and all-cause mortality. Studies with larger samples and longer follow up are needed to determine whether the mediating effect of fat mass on hPDI and CVD mortality in females is reproducible.