Association of plant-based diets with total and cause-specific mortality across socioeconomic deprivation level: a large prospective cohort

Purpose Current evidence on the association between plant-based diet indices (PDIs) and mortality is inconsistent. We aimed to investigate the association of PDIs with all-cause and cause-specific mortality and to examine whether such associations were modified by socioeconomic deprivation level. Methods A total of 189,003 UK Biobank participants with at least one 24-h dietary assessment were included. All food items were categorised into three groups, including healthy plant foods, less healthy plant foods, and animal foods. Three PDIs, including the overall PDI (positive scores for all plant-based food intake and inverse scores for animal-based foods), the healthful PDI (hPDI) (positive scores only for healthy plant food intake and inverse scores for others), and the unhealthful PDI (uPDI) (positive scores only for less healthy plant food intake and inverse scores for others), were calculated according to the quantities of each food subgroup in three categories. The Townsend deprivation index was used as the indicator of socioeconomic deprivation level. Cox proportional hazard models were used to estimate the hazard ratios (HRs) of PDIs for all-cause and cause-specific mortality. The modification effects of socioeconomic deprivation levels on these associations were evaluated. Results During a median follow-up of 9.6 years, 9335 deaths were documented. Compared with the lowest quintile, the highest quintile of overall PDI was associated with adjusted HRs of 0.87 (95% CI 0.81–0.93) for all-cause mortality and 0.77 (0.66–0.91) for cardiovascular mortality. Compared with the lowest quintile, the highest quintile of hPDI was associated with lower risks of all-cause mortality (0.92, 0.86–0.98), and death caused by respiratory disease (0.63, 0.47–0.86), neurological disease (0.65, 0.48–0.88), and cancer (0.90, 0.82–0.99). Compared with the lowest quintile, the highest quintile of uPDI was associated with an HR of 1.29 (1.20–1.38) for all-cause mortality, 1.95 (1.40–2.73) for neurological mortality, 1.54 (1.13–2.09) for respiratory mortality, and 1.16 (1.06–1.27) for cancer mortality. The magnitudes of associations of hPDI and uPDI with mortality were larger in the most socioeconomically deprived participants (the highest tertile) than in the less deprived ones ( p- values for interaction were 0.039 and 0.001, respectively). Conclusions This study showed that having a high overall PDI and hPDI were related to a reduced risk of death, while the uPDI was linked t