Plant-based dietary patterns and age-specific risk of multimorbidity of cancer and cardiometabolic diseases: a prospective analysis

Abstract

Background: It is currently unknown whether plant-based dietary patterns influence disease progression to multimorbidity after an initial non-communicable disease, and whether the associated risk of multimorbidity varies with age. This study aimed to investigate associations of plant-based diets with the risk of multimorbidity, defined as the co-occurrence of at least two chronic diseases in an individual (either cancer at any site, cardiovascular disease, or type 2 diabetes).

Methods: This prospective cohort study used data from EPIC and UK Biobank across six European countries, with participants aged 35-70 years at recruitment. We excluded participants from these cohorts who had cancer, cardiovascular disease, or type 2 diabetes at baseline or those with missing data on diet or health outcomes. Data on dietary habits were assessed either at baseline through a validated dietary questionnaire about habits in the previous 12 months or through several 24-h recall questionnaires during approximately a year of follow-up. Multistate modelling with Cox regression was used to estimate the risk of multimorbidity according to a healthful plant-based diet index (hPDI) and, separately, an unhealthful plant-based diet index (uPDI). Risk differences in adults younger than 60 years and those age 60 years and older were estimated.

Findings: 407 618 participants (226 324 from EPIC and 181 294 from UK Biobank) were included in this study. During a median follow-up time of 10·9 years in EPIC and 11·4 years in UK Biobank, 6604 cancer-cardiometabolic multimorbidity events occurred in both cohorts combined. A ten-point increment of the hPDI score was associated with a lower risk of multimorbidity, with a hazard ratio (HR) of 0·89 (95% CI 0·83-0·96) in EPIC and 0·81 (0·76-0·86) in UK Biobank. This inverse association was marginally weaker in older adults than in middle-aged adults in both cohorts. In UK Biobank, a ten-point increment of the hPDI score was associated with multivariable-adjusted HRs of 0·71 (95% CI 0·65-0·79) in adults younger than 60 years and 0·86 (0·80-0·92) in those aged 60 years and older (p_interaction=0·0016). The respective HRs in EPIC were 0·86 (95% CI 0·78-0·95) and 0·92 (0·84-1·02; p_interaction=0·32). A higher adherence to an unhealthy plant-based diet was positively associated with multimorbidity risk in UK Biobank (HR per ten-point increment of uPDI 1·22, 95% CI 1·16-1·29), but this was not replicated in EPIC (1·00, 0·94-1·08).

Interpretation: A healthy plant-based diet might reduce the burden of multimorbidity of cancer and cardiometabolic diseases among middle-aged and older adults.

Funding: The Korean Government (Ministry of Science and ICT).

Figure 1

Transitions from baseline to cancer, cardiovascular disease, and type 2 diabetes, and subsequent cancer-cardiometabolic multimorbidity Transitions are shown separately for the EPIC (A) and UK Biobank (B) cohorts. Cancer refers to first malignant tumour at any site excluding non-melanoma skin cancer. Deaths were censored and not modelled as a separate outcome. State-specific number of events are reported in boxes, and transition-specific number of events and incidence rates per 1000 person-years (within brackets) are reported on arrows. Dashed line indicates the direct transition from baseline to multimorbidity (at least two diseases in an individual). Censoring due to loss to follow-up was 1·5% (n=3395) in the EPIC cohort and 0·3% (n=544) in UK Biobank.

Figure 2

Associations between a plant-based diet and risks of cancer, cardiovascular disease, type 2 diabetes, and subsequent cancer–cardiometabolic multimorbidity Data are from the EPIC (n=226 324) and UK Biobank (n=181 294) cohorts, and associations are with either the hPDI score (A) or the uPDI score (B). Cox proportional hazard regression models were stratified by age categories at recruitment, sex, centre (or geographical region of recruitment in UK Biobank), and transitions in a clock forward multistate analysis with age as primary time variable. Models were adjusted for physical activity, smoking status, alcohol intake, energy intake, an indicator for socioeconomic status, and in women further adjusted for menopausal status and use of menopausal hormone therapy. hPDI=healthy plant-based diet index. uPDI=unhealthy plant-based diet index.

Figure 3

Associations between a plant-based diet and risk of multimorbidity, stratified by age Data are from the EPIC (n=226 324) and UK Biobank (n=181 294) cohorts, and associations are with either the hPDI score (A) or the uPDI score (B). Cox proportional hazard regression models were stratified by age categories at recruitment, sex, centre (or geographical region of recruitment in UK Biobank), and transitions in a clock forward multistate analysis with age as primary time variable. Models were adjusted for physical activity, smoking status, alcohol intake, energy intake, an indicator for socioeconomic status, and in women further adjusted for menopausal status and use of menopausal hormone therapy. The risk difference (multiplicative interaction) for the hPDI score (A) in EPIC was 0·93 (95% CI 0·81–1·07, p value 0·32) and in UK Biobank was 0·83 (95% CI 0·74–0·93, p value 0·0016). The risk difference (multiplicative interaction) for the uPDI score (B) in EPIC was 1·10 (95% CI 0·96−1·26, p value 0·19) and in UK Biobank was 1·07 (95% CI 0·96−1·20, p value 0·23). hPDI=healthy plant-based diet index. uPDI=unhealthy plant-based diet index.