Interaction between plant-based dietary pattern and frailty on cognitive decline: a longitudinal analysis of the Chinese Longitudinal Healthy Longevity Survey cohort

Abstract

Objectives

Frailty is a risk factor for faster cognitive decline, while plant-based dietary patterns are associated with decreased risk of cognitive decline. We aimed to explore their interaction with cognitive function among older adults.

Methods

We used data from the Chinese Longitudinal Healthy Longevity Survey between 2008 and 2018. Frailty was evaluated based on the frailty index (FI), and the plant-based diet index (PDI) was calculated using food frequency questionnaire at baseline. Repeated measures of the Mini-Mental State Examination (MMSE) were utilised to assess cognitive function. We used linear mixed models to estimate regression coefficients (β) and 95% confidence intervals (CI).

Results

We included 7,166 participants with a median follow-up of 5.8 years. Participants in pre-frail (β = −0.18, 95% CI: −0.24, −0.13) and frail (β = −0.39, 95% CI: −0.48, −0.30) groups experienced an accelerated decline in MMSE score compared with the robust group. The PDI modified the above association, with corresponding associations with frailty being much more pronounced among participants with a lower PDI (frail vs. robust β = −0.44, 95% CI: −0.56, −0.32), compared with those with a higher PDI (frail vs. robust β = −0.27, 95% CI: −0.40, −0.13). In addition, A combination of frailty and a low PDI was strongly associated with a faster decline in MMSE score (β = −0.52, 95% CI: −0.63, −0.41).

Conclusion

Adherence to plant-based dietary patterns attenuates the association between frailty and cognitive decline. If the observed association is causal, promoting plant-based dietary patterns may be a strategy to reduce the effects of frailty on neurological health.

Key Points

• Frailty is associated with faster cognitive decline.

• Adherence to plant-based dietary patterns attenuates frailty-associated cognitive decline.

• Combined frailty and low consumption of plant foods predict faster cognitive decline in an additive manner.

Introduction

The incidence rates of dementia and cognitive impairment are rising due to global ageing population. It is estimated that 57.4 million individuals worldwide were affected by dementia in 2019 and this number is projected to reach 152.8 million by 2050 [1]. Dementia poses significant challenges to individuals, their families, healthcare systems and societies, with global costs estimated at US$1 trillion annually [2]. In light of the limited success of current dementia treatments, there is a critical need to identify risk factors that can predict disease progression and develop effective prevention strategies to delay or prevent dementia onset.

One promising avenue of research focuses on the correlation between physical conditions and cognitive measures, with a growing interest in the association between frailty and cognitive decline. Frailty describes an age-related state of deterioration in functioning across multiple physiological systems, accompanied by an increased vulnerability to stressors [3]. Numerous epidemiological studies have demonstrated a close relationship between frailty and cognitive decline, with potential mechanisms involving inflammation, mitochondrial dysfunction and oxidative stress, epigenetic changes and hypothalamic–pituitary–adrenal axis dysfunction [4, 5]. Meanwhile, plant foods, such as whole grains, vegetables, fruit and nuts could reduce inflammation and oxidative stress in the central nervous system [6] and have been linked to increased brain volume and decreased risk of cognitive decline over time [7, 8]. Nevertheless, whether plant-based diets modify the association between frailty and cognitive function or how these factors interact to influence cognitive function remains poorly understood.

To address this knowledge gap, we used data from the Chinese Longitudinal Healthy Longevity Survey (CLHLS) to explore the interaction between frailty and plant-based dietary patterns on trajectories in cognitive function, measured from repeated assessments over time.

Methods

Study design and participants

The CLHLS is a nationally representative cohort study focusing on the older population in mainland China [9]. Data collection began with interviewing individuals aged 80 years or older in 1998, followed by subsequent interviews of both the original sample and newly enrolled participants aged 65–79 years, at intervals of 2–3 years in 2000, 2002, 2005, 2008, 2011, 2014 and 2018. Details of the survey design and collected data are available elsewhere [10].

In the current study, we used the data from the 2008 wave of the CLHLS as our baseline (n = 16,954). Exclusions were made for participants below 65 or above 106 years (n = 920), as these individuals lacked adequate data for age validation. Furthermore, we excluded participants with missing information on at least one item of frailty (n = 2,786), diet (n = 7) and covariates (n = 293) at baseline. Because the overall missing rate of cognitive function items was very low (less than 1%), we imputed the missing data using the median of the corresponding item to maximise the sample size. We also excluded participants without cognitive function reassessment because of death or loss of follow-up (n = 5,782). Ultimately, 7,166 participants were included in the analysis (Supplementary Figure S1 is available in Age and Ageing online). A comparison between the included participants and those excluded due to the absence of repeated MMSE measurements can be found in Supplementary Table S1 is available in Age and Ageing online.

The CLHLS study received ethical approval from the Research Ethics Committee of Peking University (IRB00001052-13074) and obtained written informed consent from participants or their proxy respondents before the survey. No further ethics approval was required for the present study.

Assessment of frailty

The baseline frailty status of participants was assessed based on a 44-item frailty index (FI) proposed by Liu and colleagues [11]. This comprehensive index encompasses a spectrum of health deficits, encompassing psychological functions, self-reported health status, activities of daily living, chronic diseases and interviewer-rated health status. Each deficit variable was discretely classified and mapped onto an interval of 0 to 1, where a higher score denoted a greater severity of the corresponding deficit (Supplementary Table S2 is available in Age and Ageing online). The FI was then computed by aggregating all deficit scores and dividing by the total number of deficits, with the corresponding value ranging from 0 to 1. In accordance with previous studies [11, 12], we categorised all participants as three levels based on proposed cut-off scores identified using stratum-specific likelihood ratios [13]: robust (FI ≤ 0.10), pre-frail (0.10 < FI ≤ 0.21) or frail (FI > 0.21).

Assessment of dietary pattern

A simplified 22-item food frequency questionnaire (FFQ) was used to assess participants’ regular consumption of foods and beverages at baseline. This simplified FFQ has shown satisfactory reliability and validity for generating dietary patterns for Chinese populations [14] and has been widely employed in previous studies [15–17]. In the current study, we calculated the plant-based diet index (PDI) utilising data from this simplified FFQ, based on 16 food groups encompassing the most common food consumed within the Chinese daily diet. These food groups included both plant-based foods and animal-derived foods. We assigned a score from 1 to 5 for each food group based on its frequency of consumption (Supplementary Table S3 is available in Age and Ageing online), which has been described in detail in prior publications [16, 17]. The cumulative PDI score was obtained by summing the scores assigned to each constituent food group, with a potential range from 16 to 80. Based on the median PDI level, participants were dichotomised into two groups: lower PDI (lower half scores) and higher PDI (higher half scores).

Assessment of cognitive function

Cognitive function was evaluated at each interview using the Chinese version of the Mini-Mental State Examination (MMSE) [18]. Participants were asked to finish a 30-point examination, encompassing five distinct dimensions: orientation, registration, attention and calculation, recall and language. We first scored 1 point for a correct response and 0 points for an incorrect response or inability to respond to each question. Then, we summed scores on each test, with total scores ranging from 0 to 30. Lower scores are more indicative of impaired cognitive function.

Covariates

Information on covariates including demographic characteristics, lifestyle factors and depressive symptoms was collected through a face-to-face questionnaire survey administered by an interviewer. Demographic characteristics included age (years), sex (men/women), education years and residence area (urban/rural). Lifestyle factors included tobacco smoking (current or former/never), alcohol drinking (current or former/never) and regular physical activity (yes/no). For smoking, individuals were classified as smokers if they had smoked in the past or during the interview and as non-smokers if they had neither smoked in the past nor during the interview. Similar criteria were applied to alcohol consumption. Height and weight were measured with lightweight clothing and no shoes. The body mass index (BMI) was generated by dividing weight in kilograms by the square of the height in metres. Depressive symptoms were assessed using a five-item scale [19, 20]. Two items measured positive emotions including optimism and happiness, while three items addressed negative affect including anxiety, loneliness and useless. The total score ranges from 5 to 25, with higher scores indicating more severe depressive symptoms.

Statistical analysis

Baseline characteristics of participants were tabulated by frailty status and presented as mean ± standard deviation (SD) or median (interquartile range, IQR) for continuous variables and n (%) for categorical variables. Comparative analyses were conducted using analysis of variance or Kruskal–Wallis tests for continuous variables and χ²-tests for categorical variables.

In the primary analysis, we employed linear mixed models with years since baseline as the time scale and a random intercept at the individual level, incorporating all available data on cognitive function, to examine the longitudinal association of baseline frailty with consequent cognitive decline during follow-up. Primary models included frailty, time in years since baseline, the interaction term of frailty and time and confounding variables based on previous evidence of their links with frailty and cognitive performance: age at baseline, sex, education years, urban/rural residence, tobacco smoking, alcohol drinking, physical activity, BMI and depressive symptoms. When considering frailty as a categorical variable (robust, pre-frail, frail), we reported yearly differences in cognitive decline compared with the reference group (robust). Alternatively, when considering frailty as a continuous variable, we reported yearly differences in cognitive decline for each 0.1-point increase in FI. We also used this model to predict cognitive trajectories for different FI groups over 10 years from baseline, with reference values (men, urban residence, non-smokers, non-alcohol drinkers, physical inactivity) and mean values (age at baseline, education years, BMI and depression score) of covariates.

To examine the interaction between PDI and FI in relation to cognitive decline, we introduced a three-way interaction term among PDI, FI and the time scale into the primary models. Because the interaction term was significant (P < 0.001), we stratified the primary analysis by PDI, to examine the effect modification of PDI on frailty–cognitive decline associations. We also examined the joint associations of frailty and plant-based dietary patterns with cognitive decline, by categorising participants into four groups based on frailty status (robust; pre-frail/frail) and PDI (lower half; higher half).

We performed sensitivity analyses to assess the robustness of our findings. First, we excluded participants with a baseline MMSE score < 18 to account for the potential of inaccurate reporting of FI items and dietary habits due to low cognition. Second, we used another cutoff for FI score that was suggested in previous studies [21, 22]: non-frail (FI ≤ 0.1), very mildly frail (0.1 < FI ≤ 0.2) and mildly/moderately/severely frail (FI > 0.2).

All statistical analyses were performed using R software version 4.1.3 (2022; R Development Core Team). We used the ‘lmer’ command in package ‘lme4’ (version 1.1.28) to perform linear mixed models. All statistical tests were two-sided, and P < 0.05 was considered statistically significant.

Results

Among the 7,166 participants, the mean ± SD age was 81.46 ± 10.18 years and 53.08% were women. Supplementary Figure S2, available in Age and Ageing online, displays the distribution of FI, PDI and MMSE score, revealing a baseline median (IQR) FI of 0.11 (0.07, 0.18), a mean ± SD PDI of 49.63 ± 6.05 and a median (IQR) MMSE score of 28 (24, 29). Throughout, participants underwent an average of 2.94 ± 0.81 cognitive assessments, with twice for 2,553 participants, three times for 2,463 participants and four times for 2,150 participants. Compared with the robust group, the frail group tended to be older, women, less educated, live in urban areas and have a lower MMSE score at baseline, but less likely to smoke, drink or engage in physical activity (Table 1).

During a median (IQR) follow-up of 5.81 (3.12, 9.69) years, participants identified as pre-frail (β = −0.18, 95% CI: −0.24, −0.13, P < 0.001) and frail (β = −0.39, 95% CI: −0.48, −0.30, P < 0.001) experienced accelerated decline in MMSE score compared with the reference group (robust). For each 0.1-point increase in FI, there was an additional 0.16 (95% CI: −0.20, −0.13, P < 0.001) decrease in MMSE score annually. Notably, plant-based dietary patterns acted as an effect modifier in the association between frailty and cognitive decline. Among individuals with lower PDI, the association between frailty and cognitive decline was substantially more pronounced (frail vs. robust β = −0.44, 95% CI: −0.56, −0.32, P < 0.001; per 0.1 increase in FI β = −0.17, 95% CI: −0.22, −0.13, P < 0.001), compared with those with higher PDI (frail vs. robust β = −0.27, 95% CI: −0.40, −0.13, P < 0.001; per 0.1 increase in FI β = −0.14, 95% CI: −0.18, −0.09, P < 0.001) (Table 2,Figure 1). When exploring the joint associations (Table 3), the β (95% CI) were − 0.08 (−0.15, −0.01) in the robust and higher PDI group, −0.21 (−0.28, −0.15) in the pre-frail/frail and higher PDI group and − 0.52 (−0.63, −0.41) in the pre-frail/frail and lower PDI group, compared with the robust and higher PDI group.

The results of the main analyses were consistent in sensitivity analyses by excluding participants with baseline MMSE < 18, although the β estimates were generally magnified (Supplementary Tables S4 and S5 are available in Age and Ageing online). For example, the frail vs. robust β (95% CI) decreased to −0.67 (−0.77, −0.58) for all participants, −0.78 (−0.90, −0.66) for those with lower PDI and − 0.46 (−0.60, −0.32) for those with higher PDI, and for the pre-frail/frail and lower PDI group vs. the robust and higher PDI group, the β (95% CI) reduced to −0.85 (−0.97, −0.74). When using another cutoff for FI score (0.1 and 0.2), the results remained similar (Supplementary Tables S6 is available in Age and Ageing online).

Discussion

In this cohort study involving 7,166 participants followed for a median duration of nearly 6 years, we observed that frailty was significantly associated with an accelerated rate of cognitive decline when compared with robust individuals. The associations were modified by plant-based dietary patterns. Specifically, frailty was associated with an additional annual decrease of 0.44 in MMSE score among participants with lower PDI, whereas this association was attenuated to 0.27 for those with higher PDI. In addition, frailty and PDI jointly predict cognitive decline in an additive manner, with the combined association being more significant than the sum of their individual associations. Taken together, these findings highlighted the important role of plant-based dietary patterns in shaping the trajectory of frailty-related cognitive decline.

Frailty and cognitive impairment constitute two prevalent health challenges in the older population [23, 24], and their potential interconnectedness and underlying mechanisms have been widely explored [4, 25]. However, longitudinal evidence regarding the association between frailty and subsequent cognitive decline remains relatively limited [5]. A study including 7,439 older US adults from the National Health and Ageing Trends Study indicated that frail individuals experienced steeper declines in global cognitive function (−0.01 SD/year, 95% CI: −0.012, −0.005) over a 5-year follow-up compared with their non-frail counterparts [26]. Similarly, recent data comprising 385 participants from the Tasmanian Study of Cognition and Gait demonstrated an independent relationship between frailty and a decline in global cognition (β = −0.001, 95% CI: −0.003, −0.001), even after accounting for brain atrophy and cerebral small vessel disease [27]. Importantly, these studies did not specifically differentiate between pre-frailty and frailty, nor did they compare pre-frailty to robustness. Our study extends these insights by revealing that even pre-frailty, a potentially reversible stage, is associated with accelerated cognitive decline. This was consistent with the findings from the Health and Retirement Study involving 15,454 participants, where mild frailty was significantly linked to subsequent cognitive decline (β = −0.22 for inverted U-shaped and U-shaped frailty trajectories) [28]. Additionally, our analysis using continuous FI indicated a positive relationship between FI score and cognitive decline, indicating that any increase in frailty severity was related to accelerated decline in cognitive function.

Dietary patterns and other modifiable lifestyle factors have been associated with cognitive outcomes, suggesting potential targets for slowing the progression of cognitive decline and preventing dementia [2]. Previous studies have reported the interaction between lifestyle factors and frailty in relation to cognition. For example, an analysis of the 2014–16 Nutrition and Health Survey in Taiwan (n = 1,115) showed a combined association of frailty and a lower dietary diversity score with worsening cognitive function [29]. The Dongfeng–Tongji cohort of 3,279 participants indicated that the frailty–cognitive impairment association was attenuated by smoking cessation and regular physical exercise [30]. Notably, these studies were cross-sectional, thus precluding the assessment of cognitive decline. To the best of our knowledge, no prior studies have specifically evaluated the modification effects of plant-based dietary patterns on the association between frailty and cognitive decline, despite substantial work investigating the link between plant-based diets and age-related cognitive decline or dementia [8, 31]. In the current study, we observed that the association of frailty with a faster decline in MMSE score was more pronounced among participants with lower PDI, yet markedly reduced among those with higher PDI. Together with the joint effects of frailty and PDI, our findings expand on these previous findings and provide new insight into the potential of plant-based diets to ameliorate frailty-associated cognitive decline. Nevertheless, it is important to acknowledge that the present evidence was observational and no causal inference can be drawn.

The complex associations of plant-based dietary patterns and frailty probably reflect joint mechanisms through which these factors are thought to influence cognitive function. First, the abundance of specific nutrients inherent in plant-based diets could explain their potential in modifying the frailty–cognition relationship. These nutrients like antioxidants, unsaturated fatty acids, dietary fibre, vitamins and carotenoids have the capacity to mitigate inflammation and oxidative stress within the central nervous system, which are mechanisms that strongly link both frailty and cognitive decline [5, 32]. Second, unhealthy dietary habits, encompassing high consumption of full-fat dairy, red and processed meats, coupled with low intake of fruits and vegetables, are recognised risk factors for frailty [33], which notably align with factors associated with dementia. Given these potential interconnections, we may hypothesise that frailty might not only influence cognitive function through independent pathways but also serve as a mediator between dietary patterns and cognition (about 15% in this study, data not shown). In scenarios where low plant-based dietary patterns co-occur with frailty, their combined effect may predict more adverse cognitive outcomes than the individual sum of effects.

A major strength of the present study is our use of longitudinal data in a large-scale national cohort. Our study improves on previous cross-sectional studies because the results are less likely to be influenced by reverse causation; we excluded participants with MMSE scores < 18 at baseline, and the results were consistent. However, our study has several limitations. First, akin to many longitudinal studies involving older participants, a large attrition proportion due to non-random missing in the current study raises concerns about selection bias. Indeed, participants excluded from the original cohort due to a lack of repeated MMSE measurements tended to be older and have poorer cognitive performance at baseline compared with those who were included. Second, the single assessment of frailty and PDI at baseline restricts our ability to track transitions in frailty status and dietary patterns with cognitive function. Third, although we adjusted for multiple confounders, residual confounding cannot be ruled out. Finally, our study comprises only Chinese old adults; thus, generalisation to other populations should be done cautiously.

In summary, this study provided evidence that adherence to plant-based dietary patterns attenuated the association between frailty and cognitive decline. Furthermore, we elucidated the combined effect of plant-based dietary patterns and frailty on cognitive function. If the observed association is causal, promoting plant-based dietary patterns may be a strategy to reduce the effects of frailty on neurological health.

Acknowledgements

We thank the staff and the participants of the CLHLS study.

Declaration of Conflicts of Interest

None.

Declaration of Funding

None.

Data Availability

The analytical dataset used in this study is a publicly available dataset released by the CLHLS. Information about the data source and available data are found at [https://opendata.pku.edu.cn/dataverse/pku?q=CLHLS]. Researchers can obtain these data after submitting a data use agreement to the CLHLS team.

References

Author notes