Plant-based diets (PBD) are generally promoted as beneficial for health. However, whether this is also the case at older ages, when energy deficits, muscle loss and frailty affect health, is unclear. Research has shown that among older adults, particularly in men, a healthful PBD is associated with a lower frailty risk. This relation was however, not studied in the context of socio-economic status (SES), a major factor influencing the risk of frailty. Therefore, we aim to assess whether plant-based diets associate with frailty risk at older ages and whether this association is moderated by sex and income in a large population-based dataset.

We investigated baseline data from the UK Biobank cohort study (UKB) cross-sectionally (n = 73 180, mean age = 55.48 ± 7.87). We applied a plant-based diet index [range 17−85], differentiating between a healthful (hPDI) and unhealthful plant-based diet (uPDI). Frailty was assessed by the Fried frailty phenotype and categorized into 0–4 symptoms of frailty. Average annual household income was divided into three categories: low (<18.000 £), medium (18.000–52.000 £) and high (>52.000 £). We applied an ordinal logistic regression model with frailty as the categorical outcome and PDI as continuous predictor while adjusting for age, sex, ethnicity, education, BMI and UKB assessment center. Secondly, we included an interaction term (PDI*sex*income). To identify subgroups driving any interactions, we stratified by sex and subsequently by income group to determine the effect of PDI in subgroups while additionally adjusting for lifestyle factors.

A 10-unit increase in hPDI was associated with 3.4% lower odds for frailty (OR = 0.966, 95%CI [0.946, 0.987]), whereas a 10-unit increase in uPDI was associated with 7.7% greater odds for frailty (OR = 1.077, 95%CI [1.054, 1.101]). The association between uPDI and frailty was moderated by income and sex (uPDI*income*sex, p = 0.002), whereas no such moderation was found for hPDI (p = 0.602). Subsequent stratification reveals a significant effect of uPDI on frailty particularly among men with low income (OR = 1.177, 95% CI [1.069, 1.298]), but not for women. This association in men largely persisted after adjustment for additional lifestyle factors (OR = 1.119, 95%CI [0.995, 1.258]).

We observed that adherence to an unhealthful plant-based diet was associated with a higher risk for frailty. This relation was especially observed for men with lower incomes and not explained by other lifestyle factors. While future research may investigate more specific determinants of health and diet behavior in men of low household income, this group in particular may profit from diet intervention improving diet quality