Unhealthful plant-based diet associates with frailty risk predominantly in men with low income from the UK Biobank cohort

1. Introduction

Plant-based diets, characterized by a high intake in plant foods and comparatively low intake of animal foods, have gained attention in recent years as a shift towards more plant foods is recommended by the EAT-Lancet committee [1] as well as dietary guidelines worldwide. Further, plant-based diets are recommended for the prevention of age-related diseases, in particular metabolic diseases. However, for vulnerable populations strict plant-based diets may also lead to energy deficits and inadequate protein intake, contributing to a loss of muscle mass, which poses a risk factor for frailty [2]. Frailty has been defined as increased vulnerability resulting from an age-associated decline in function and may put older adults at an increased risk for adverse outcomes, such as hospitalization and mortality [3]. Meanwhile, studies have shown that a healthful plant-based diet may actually lower the risk for frailty, although not in both sexes equally [4, 5]. The role of diet in the prevention of frailty, may be explained by anti-inflammatory properties of plant foods [4, 6] and men in particular may respond more to the effects of an inflammatory diet [6, 7]. At the same time, previous research has shown that women are generally more likely to become frail over the life course [8], indicating differences between men and women exist in the prevalence of frailty and response to lifestyle factors [9]. It is therefore important to further investigate the interplay of diet and sex on frailty risk.

Sex-differences also exist regarding socioeconomic background and its effect on frailty. Higher socioeconomic status (SES) has been linked to a healthier lifestyle and reduced frailty risk, while low SES groups have benefitted more from improving their lifestyle [10]. In addition, sex differences were observed: women’s frailty risk was determined more by education, whereas in men income had a stronger influence [11]. Similarly, individuals with a lower SES and an unhealthy lifestyle have been found to have a higher mortality risk compared to individuals with high SES and a comparable unhealthy lifestyle, with men with low income being a particular at-risk group [12]. Thus, these findings point towards a potential interaction of lifestyle, sex, and income. It is currently not known whether each of these characteristics associate additively and independently with the development of frailty, or whether the interplay of these factors impacts frailty risk.

We therefore aim to identify at-risk groups for frailty which may be target groups for future dietary and lifestyle interventions. We focus on the question whether plant-based diets are differently associated with frailty in men and women and whether this association dependent on income. To quantify adherence to a plant-based diet, we apply a plant-based diet index, differentiating between adherence to a healthful and unhealthful plant-based diet to a population of 73,180 adults (mean age=55.48 years) from the UK Biobank.

2. Methods

3. Results

The complete sample consisted of n=73,180 mostly white adults with a mean age of 55.5 years. A slight majority of the sample consisted of women (52.50%). Women had slightly lower adherence to uPDI, were more likely to be in the low-income group compared to men. Further, women had a lower BMI, consumed less alcohol and were less likely to be smokers than men. Despite their seemingly healthier lifestyle choices, women reported higher medication intake and were more likely to be frail (Table 1).

3.3. Effect of unhealthful plant-based diet primarily in men of low income

To further pinpoint the interaction effects of sex, uPDI and income on frailty, we performed sex-stratified analyses (model 5a+b). In women uPDI was not associated with frailty and neither did it interact with income (Suppl. Table 7). Meanwhile, in men, we observed an association of uPDI with frailty (OR=1.249, p<0.001) with a significant interaction between uPDI and medium (p=0.007) and high income (p<0.001).

The interaction observed in men may be explained by lifestyle and health-related factors such as physical activity, smoking, alcohol consumption and vitamin D levels. Therefore, we further adjusted the association for these additional lifestyle and health-related factors (model 5c). Interestingly, the interaction term between uPDI and income remained significant, indicating there is a difference in the association between uPDI and frailty by income which is not explained by lifestyle and health-related factors tested (model 5c, Suppl. Table 8). Subsequently, to visualize the interaction effect, we further stratified by income group (models 6a-c). We observed that for men with a high income, uPDI was not associated with frailty (OR=0.994, 95% CI [0.933, 1.059]), while in contrast every 10-unit increase in uPDI was associated with 18% (OR=1.177, 95% CI [1.069, 1.298]) greater odds for frailty in men with a low income (Fig. 1). This association persisted after adjustment for additional health- and lifestyle factors (OR=1.119, 95%CI [0.995, 1.258], Fig. 1).

• Download figure
• Open in new tab

Figure 1. Association between unhealthful plant-based diet index (uPDI) and frailty in men of different income groups

OR per 10-unit increase in uPDI

− Basic model --- fully adjusted model ● low income ▲ medium income ■ high income

4. Discussion

In a large dataset of over 70 thousand individuals, we dissected how plant-based diets, sex, and income associate with frailty. We observed for both men and women that greater adherence to a healthful plant-based diet was associated with a lower risk of frailty, and greater adherence to an unhealthful plant-based diet was associated with higher risk of frailty. In addition, we established that higher income and male sex were associated with lower risk of frailty. Because sex, income and diet are known related factors [4, 8, 11], we tested for a three-way interaction of these variables in their association with frailty. While we observed no interaction for hPDI, sex and income on their effect with frailty, we did observe interaction of uPDI, sex and income on frailty. Specifically, men with low income and a high adherence to unhealthful plant-based diet had remarkable higher risk for frailty, which was not explained by physical activity, alcohol consumption or smoking. Remarkably, no effect of uPDI on frailty was observed in high income men or women in either income group.

Our findings, that a healthful plant-based diet was associated with a lower frailty risk, while an unhealthful plant-based diet was associated with a higher risk for frailty, is supported by other research [4, 5, 7]. Our observations of an interaction between income and unhealthful dietary pattern were also in line with previous research. A study in the UK Biobank concluded that the association between unhealthy lifestyle behaviors (including e.g. diet, physical activity, sleep duration and smoking) and adverse health outcomes were stronger among deprived populations [29]. Similarly, a more recent large population-based study reported that a healthy lifestyle benefited low socio-economic status (SES) populations more by having a stronger association with mortality risk compared to groups with high socio-economic status [12] In Chinese older adults a protective effect of high income on frailty was reported [11]. Our study is unique in that we investigate the interplay of all three factors, income, sex, and plant-based diet adherence combined, on frailty risk. We identified men with low income as an at-risk group for frailty when consuming an unhealthful plant-based diet, even adjusting for lifestyle factors, suggesting a unique effect of diet on frailty.

Several hypotheses have been proposed for the higher morbidity observed especially in men with low income. First, men are generally assumed to engage more in unhealthy behaviors [30]. Similarly, low SES groups are more likely to engage in unhealthy behaviors in extreme ways compared to higher SES groups. This has been discussed for alcohol consumption, with low income groups being more likely to engage in excessive drinking, but may also extend to other health behavior such as diet [31]. When comparing uPDI scores of the frailest individuals, we observe that, men with a low income had a higher uPDI (55.05) than high income men in the same frailty category, suggesting men with low income engaged more in unhealthy dietary habits (50.43). While a clear explanation for our findings is lacking, we observed that women in our sample had consistently higher intakes of vegetables, fruit, tea, and coffee as well as dairy compared to men. All these food groups have been associated with a lower frailty risk, supposedly due to concomitant higher intakes of antioxidants, protein, and calcium [32-34]. Further, men with low income and frailty were the subgroup with the highest intake of soda (data not shown), potentially explaining the association with frailty in this population group [35]. Future research may explore how dairy and a reduction of soda consumption may fit into a healthful plant-based diet to contribute to lower frailty risk.

At the same time research suggests that low income groups have less access to the medical system or are less inclined to make use of medical care [29]. Further, low income often goes hand in hand with increased stress and mental burden associated with the challenge to make a living. In men, the mental distress of low socioeconomic status may intersect and be amplified by societal expectations to provide, leading to a greater importance of income in this group [36]. Men with a low income may therefore be a particularly vulnerable group and may be targeted by dietary interventions.

Surprisingly, we observed no interaction effect of income and plant-based diet adherence with frailty risk in women. Women are generally more likely to be frail yet have a lower mortality risk. This is known as the male-female health-survival paradox [8]. It has been proposed, that in women, other factors contribute more to frailty, shifting the focus away from lifestyle. For example, women accumulate more disabling conditions, such as arthritis, which can limit physical functioning [6, 37]. It has also been suggested, that inflammation may contribute more to the development of frailty in women [38]. At the same time, men have been proposed to benefit more from an anti-inflammatory diet, potentially contributing to lower frailty risk [39]. Previously, differences in the gut microbiome between sexes has been suggested as a potential mechanism [6], however more studies are needed to investigate the interplay of sex and frailty, the gut microbiome and risk factors such as diet and inflammation.

Our analysis has several limitations. Firstly, the UK Biobank sample was still relatively young with a mean age of 55 (range 40-70 years) years. Subsequently, cases of frailty, especially those who showed more than three symptoms of frailty, were relatively low in our sample and would be considered early frailty. Secondly, participants of the UK Biobank study tend to live in less deprived areas [40]. This is also supported by the fact that the majority of our sample had a medium or high income. In our analysis, we included participants who had filled at least two 24h diet recalls. It can be argued that two questionnaires may be not enough to accurately reflect usual intake. However, according to the median of questionnaires filled, we had 3 or more questionnaires available from 50% of our sample, a number that has previously been identified as sufficient for estimating energy intake [34]. While we adjusted for ethnicity, the vast majority of the sample was white, leading to limited transferability of our results to other ethnic groups. We did not assess the impact of biological factors, such as levels of inflammatory markers, the gut microbiome, or pre-existing conditions, although we approximated pre-morbidities by adjusting for medication intake. Further research may therefore extend our findings by taking into account these factors. Lastly, as we conducted a cross-sectional analysis, causality cannot be inferred and there is a possibility that reverse confounding contributed to the results. Our findings on the association between diet and frailty are however in line with other, longitudinal research, supporting our results. Future studies may replicate our findings in older, more deprived, and diverse populations and take into account markers of metabolic health.

The strength of our analysis lies predominantly in the large sample size of the UK Biobank making robust observations and allowing to explore the relative contribution of clustering factors influencing frailty. Despite potentially limited representativeness, it has been suggested, that exposure-disease relationships are still extrapolatable to the general population [40]. Further, the extensiveness of the UK Biobank database allowed us to adjust our analysis for a variety of lifestyle and health related factors. Additionally, the Oxford WebQ that was used to collect dietary data has been validated and shown to produce reliable results [14].

5. Conclusion

We conclude that adherence to an unhealthful plant-based diet contributes to frailty risk predominantly among men with low income (<£18k). We therefore identified an at-risk group, which may benefit from dietary and lifestyle interventions. Future research may focus on identifying potential dietary drivers of the identified association in older and more deprived populations.

7. Funding and Conflict of Interest

P.E. Slagboom, M. Beekman, and L.C.P.G.M de Groot have received funding from the Vitality Oriented Innovations for the Lifecourse of the Ageing Society (VOILA) Consortium [ZonMw 457001001]. All authors declare no conflict of interest.

8. Author Contributions

KS, MB, PES were involved in study concept and design, acquisition, analysis, and interpretation of data. NvB and MRG provided statistical and methodological advice. LCPGMdG provided critical input. KS performed statistical analysis and prepared the manuscript. All authors read and approved the final version.