Association Between Combined Multiple Behaviors and Cognitive Impairment in Older Adults: A 10-Year Cohort Study

1. Introduction

Population aging is becoming a global problem, particularly acute in China with the world's largest population, and brings the challenges for the elderly care system and medical system. A recent report from the United Nations predicted that the population of people aged 60 years old and above will continue to grow in China, both in absolute ways and as a proportion of the total [1]. By 2050, China's elderly population aged 60 and above will exceed 500 million, accounting for 38.81% of the total, and the proportion of elderly people aged 80 and above will exceed 10% [2] . Dementia and cognitive impairment issues, in general, are major health concerns among older adults. It is estimated by United Nations that the cases of dementia among older adults is expected to reach one billion in 2050 [3,4]. Cognitive disorders are also responsible for millions of deaths every year.

Prevention is the most important strategy to address the challenge of dementia because there is no effective treatment for this disease yet [5]. Cognitive impairment is the early stage of dementia and provides a time-window for taking preventive measures [6]. Therefore, identifying potentially modifiable risk factors for cognitive impairment is of particular importance for effective preventions of dementia and precise public health interventions [7,8].

Several single lifestyle factors have been associated with dementia risk, but the combined impact of multiple unhealthy lifestyle factors on cognitive impairment remains largely unknown in China [9,10]. This study aims to investigate whether the increased risk of cognitive impairment among Chinese adults can be offset by a broad combination of healthy lifestyle behaviors. We performed a prospective study to examine the long-term association between combined multiple modifiable lifestyle factors and cognitive impairment, utilizing a nationwide population-based data from a Chinese cohort survey. Cox proportional hazards models were used to examine the associations between multiple unhealthy lifestyle behaviors and cognitive impairment.

2. Materials and Methods

2.1. Study Design and Participants

We used the newest data from four waves (2008-2018) of the Chinese Longitudinal Healthy Longevity Survey (CLHLS). The CLHLS established in 1998 is a nationwide survey done in a randomly selected half of the counties and cities in 22 of the 31 provinces in China. Trained interviewers did the surveys at participants’homes following a structured questionnaire collecting socio-demographic date lifestyle, cognitive function, psychological status, physical capacity and chronic disease diagnosed. More details on the sampling procedure and assessment of data quality can be found in previous publications [11,12]. This study analysed data in members of the population aged over 65 years old from the fifth wave as baseline, and followed-up data from sixth wave, seventh wave and eighth wave to explore association between co-factors and cognitive impairment.

Data collected in 2008/2009 included 16,954 Chinese elderly adults (aged 65 years and above). For the present study, exclusion criteria were (1) cognitive function missing at baseline (n=3,257); (2) age (<65 years old) (n=395); (3) impaired cognitive function at baseline (n = 4,716); (4) key variables and covariables missing (n=265). Total 8,321 participants were included finally. Figure. 1 shows the full process of the inclusion and exclusion of participants in this study.

2.2. Exposures

For single lifestyle activity, this study selected sleep duration, housework. Sleep duration: The individuals were asked to answer “how long do you sleep normally?”. Based on previous literature [13], the study participants were classified into three duration groups: normal (5-10 hours/day) and abnormal (<5 or > 10 hours/day). Housework: The individuals were asked to answer “do you do housework at present?”. Exercise: The individuals were asked to answer “exercise or not at present?”. Finally variable “housework” and “exercise” were aggregated into dichotomies, which were “yes” and “no”. Social connection: “yes” (at least one of the two answers is “yes”, including “do you take part in some social activities at present?” and “do you play cards/mah-jongg at present?”) versus “no” (both “no” for the above two questions).

Figure 1. Flow chart of the study population.

Figure 1. Flow chart of the study population.

For the measurement of multiple lifestyle behaviors, this study created dietary behavior, multiple lifestyle activity and combined lifestyle behaviors. 1) Dietary behavior: The individuals were asked to answer “how often eat fish/vegetables/fruits/meat/eggs or drink milk at present?”. Based on the “Dietary Guidelines for Chinese Residents (2022)” [14], we assigned different scores according to different options. The option “every day” was given two points; The option “not everyday, but at least once per week or per month or occasionally” was given one point; The option “rarely or never” was given zero point. The sum of all the above scores were the nutritional score, which was divided into two levels: good (>6 points), and poor (≤6 points). 2). The multiple lifestyle activity integrated “sleep”, “housework”, “exercise” and “social connection”. If the individual preferred “normal” for “sleep”, “yes” for “housework”, “exercise” and “social connection”, “multiple lifestyle activity” was defined as the “favorable” level. If the individuals preferred “abnormal” for sleep, “no” for “housework”, “exercise” and “social connection”, “multiple lifestyle activity” was defined as the “unfavorable” level. Other combinations were defined as the “intermediate” level. 3). The combined multiple behavior was further created by including dietary habits based on multiple lifestyle activity. If the individuals preferred “normal” for “sleep”, both “yes” for “housework”, “exercise” and “social connection”, and “good” for “dietary”, then “multi” was defined as the “favorable” level. If the individual preferred “abnormal” for “sleep”, both “no” for “housework” and “exercise”, and “poor” for “dietary”, then “combined multiple behavior” was defined as the “unfavorable” level. Other combinations were defined as the “intermediate” level.

2.3. Outcome Measurement

Cognitive status was measured according to the Chinese version of the Mini-Mental State Examination (MMSE) at each survey [15], which has been widely used to evaluate cognitive function and turned out to be reliability and validity [16,17]. The total scores on the MMSE ranges from 0 to 30, which included 24 items that assess orientation, naming, registration, attention and calculation, copying a figure, recalling, and language capability, with higher scores representing higher cognitive function. Based on the previous study [18]. We treated responses of “unable to answer” as “wrong”. Individuals with MMSE scores of 24 or higher were considered to have normal cognitive function and MMSE scores of < 24 were considered to suffer cognitive impairment [19].

2.4. Covariates

Sociodemographic variables, including age (“>80” versus “65-80”), sex (“male” versus “female”), marital status (“married” versus “divorced or widowed” versus “never married”), education (“no schooling” versus “1-5 years schooling” versus “>5 years schooling”), residence (“urban” versus “rural”), income (“rich” versus “medium” versus “poor”) was collected and social. Living habits: Three main terms were self-reported: smoking currently (“yes” versus “no”), drinking currently (“yes” versus “no”). Health conditions: The body health status variables were body mass index (BMI) (“underweight (≤18.5 kg/m2”) versus “normal (18.5–23.9 kg/m2) ” versus “overweight or obesity (≥ 24 kg/m2) ”) [20], and self-reported medical diagnoses of non-communicable chronic diseases (NCD) (“yes” (suffering at least one disease of cancer, hypertension, diabetes, heart disease and stroke or cerebrovascular disease) versus “no” (suffering none of the above diseases)).

2.5. Statistical Analysis

Continuous data are presented as mean ± standard deviation, and categorical data are expressed as the count and percentage. One-way ANOVA was applied to compare those continuous respondents, and χ2-tests for categorical variables were conducted to evaluate the differences in sample characteristics according to the status of healthy behaviors in the baseline.

Cox proportional hazards models were performed to assess the relationship between various exposures, including [single lifestyle activity (“sleep duration”, “housework”, “exercise” and “social connection”), multiple lifestyle behaviors (“lifestyle activities”, “dietary behaviors”, and “combined multiple behaviors”] and cognitive impairment in respondents. The crude cumulative incidence rate per 100 person and incidence rate per 100 person-years of cognitive impairment were both estimated. We conducted the unadjusted model and the ultivariable-adjusted model for outcome, adjusting for age, gender, education, martial status, income, residence place, smoking, alcohol drinking, other NCDs, baseline MMSE score and other confounders. The results for associations between exposures and outcomes were presented as hazard ratios (HRs) and 95% CIs.

To comprehensively assess the relationship between sleep duration and cognitive impairment, instead of dividing sleep duration into two groups, we used sleep duration as a continuous variable to evaluate the association by logistic regression, adjusting for sociodemographic characteristics. Restricted cubic spline analysis [21] was performed to examine the pattern of the association between sleep duration (as a continuous variable) and cognitive impairment after adjusting for covariates mentioned above, with 5.0 hours of sleep duration as a reference point. Meanwhile, we also performed the same restricted cubic spline analysis, with nutrition scores as a continuous variable and 6 points as a reference to further understand the relationship between nutrition level and cognitive impairment.

Subgroup analyses were subsequently conducted in different exposure groups, including age, sex, residence, education, NCD and income. To evaluate whether interactions with age, sex, residence, education, NCD and income had an influence on the associations between various exposures and the onset of cognitive impairment, interaction terms were added to ultivariable-adjusted model.

We also conducted sensitivity analyse to evaluate the robustness of our main findings. A competing risk model was performed to assess the relationship between various exposures and cognitive impairment, in which death without cognitive impairment was treated as a competitive event. Data analysis was performed with SAS version 9.3 (SAS Institute, Cary, NC) for Windows (IBM Corp) except for the restricted cubic spline analysis, which was performed using R version 3.4.2 (R Foundation for Statistical Computing). The significance level was set at p < 0.05.

3. Results

Our analysis included 8,321 participants aged 65 years or older without cognitive impairment at baseline (mean age 82.0 [SD 10.5] years). Among the participants, 47.1% of the participants were females, 48.9% were illiterate, 84.8% were residing in rural areas and 24.61% individuals at the baseline reported NCDs. Among Chinese older adults, the proportion of overweight or obesity was 16.6%. The median cognition score of included participants was 29 (IQR: 27–29) in 2008 and 28 (IQR: 25-29) in 2018, respectively. Participants with multiple healthy behaviors, showed a higher value of MMSE score than those with medium and bad behaviors. During a median follow-up of 9.3 years, 2,070 developed all-cause dementia. (Table 1)

Table 2 showed a significant association of combined multiple behaviors with incident cognitive impairment [the hazard ratio (HR) 0.55, 95%CI 0.35 - 0.86]. Among participants with good dietary behaviors, the HR for cognitive impairment comparing poor dietary behaviors was 0.91 (95% CI 0.82 - 0.99). Compared with participants who had poor lifestyle behaviors, the HR for cognitive impairment was 0.65 (95% CI 0.45 - 0.94) in those who had good lifestyle behaviors, respectively. In terms of single lifestyle activity, the findings also revealed protective effects of doing housework, taking exercise and social connection on cognitive performance, while there was not a significant association for sleep after adjusting for covariates.

Subgroup analyses revealed a significant relationship of combined multiple behaviors with cognitive impairment among rural residence (HR = 0.56, 95% CI =0.30 - 0.90) and individuals with lowest education (HR = 0.53, 95% CI =0.35 - 0.91). Furthermore, the association was statistically significant among people with medium income and without NCDs in China. (Figure 2) Similar results were also showed for associations of dietary behaviors with cognitive impairment among older Chinese adults across socio-demographic groups. (Figure 3) However, there was no statistically significant results related to the subgroup analysis of relationships between multiple lifestyle activity and cognitive impairment. (Figure 4)

The Cubic spline curve described the association between healthy dietary score (nutrition score) and incident cognitive impairment events. Compared with 6 points, high healthy dietary scores (>6) were associated with lower risks of cognitive impairment; corresponding, people with a low healthy dietary score (<6) had higher risks of cognitive impairment. (Figure 5) Compared with 10 hours, the risk of cognitive impairment increased with the length of sleep. (Figure 6)

4. Discussion

4.1. Principal Findings

Based on nationally representative data from the CLHLS from 2008 to 2018, we found a significant relationship between combined multiple behaviors and incidence of cognitive impairment among older Chinese adults aged ≥65 years, with consistent results observed for the dietary behaviors, single and multiple lifestyle activities. Subgroup analyses suggested significant associations of combined multiple behaviors with cognitive impairment in females and individuals with lowest education. Furthermore, the association was statistically significant among people living in rural areas and those individuals with medium income and without NCDs in China.

4.2. Literature Comparisons

Numerous studies have explored the potential avenues for the prevention of cognitive impairment/dementia, and the Lancet's 2020 Commission has proposed 12 modifiable risk factors and recommended strategies for reducing the risk of dementia. [5] Among these, the suggestion of increasing social and physical activity for later life population [>65 years old] is relatively consistent with the result of housework, exercise, and social connection in our study. There is clear evidence that social isolation [22,23,24,25] and physical inactivity [26,27,28] as well as reducing housing work are risk factors for dementia, and even intervention during the later life period may have a delayed/preventive effect on dementia. [29,30,31] Accordingly, the Lancet Commissions suggested to sustain midlife and possibly later life physical activity and to increase social activity to prevent dementia. However, research on the relationship between diet and sleep and the risk of cognitive impairment has generated some controversies. Studies have indicated that excessive or insufficient sleep duration [32,33,34,35,36] and disturbance [37] are associated with an increased risk of dementia in the elderly. Nevertheless, research has not identified any causal relationships between sleep disorders and Alzheimer's disease. [38] Furthermore, there is insufficient evidence to support the effectiveness of sleeping medication for preventing cognitive decline, and such medications may have harmful effects. [35,39] Our study, which focused on the relationship between sleep duration and cognitive impairment, did not find any significant association between sleep duration less than 5 hours or more than 10 hours and an increased risk of cognitive impairment. It is worth noting that the relationship between sleep and cognitive impairment is complex and may not be fully explained by a single sleep duration. Further research is needed to elucidate this relationship. As for diet habits, observational study suggests that malnutrition [40] and unhealthy diet with low plant intake [41] are related to accelerated cognitive decline. However, a Cochrane review did not find sufficient evidence that supplementing vitamins or minerals could improve cognition. [42] Although the Mediterranean diet appears to improve global cognition scores, it does not significantly reduce the risk of cognitive impairment or dementia in healthy populations. [43,44] Our study also found that good dietary habits are a protective factor against cognitive impairment, thereby adding evidence in this field for healthy elderly populations in China.

Our study proposed a combined multiple behavior approach by integrating various activities and dietary habits and proposed criteria for good behavior. Our findings demonstrated a significant association between good behavior and reduced risk of cognitive impairment. Similar studies have been conducted among elderly [45,46] and middle-aged populations [47], but these studies generally combined different lifestyle activities, such as smoking and alcohol intake. Meta-analysis on multiple lifestyle activities and cognitive impairment have indicated that these factors have cumulative effects on cognitive impairment risk. [48] Given that individual risk factors account for only a small proportion [2-4%] of the explanation for dementia risk, [5] research or intervention addressing multiple risk factors/behaviors may be more meaningful. Recent studies have been conducted on multiple interventions for lifestyle activities using smartphone apps and other digital methods to improve lifestyle activities and potentially improve brain health outcomes. [49,50] Although the lifestyle behaviors involved in these studies may differ from those in our study, it can be inferred that interventions of a combination of multiple lifestyles could have great prospects in preventing dementia.

4.3. Public Health and Clinical Implications

In terms of public health and clinical implications, this study indicates the strong potential cognitive benefits of adherence to multiple ideal behavioral lifestyle factors regardless of socioeconomic characteristics. Therefore, preventive policies should promote stricter adherence to multiple favourable behavioral lifestyles (such as a healthy diet, good sleep, housework, engaging in regular physical activity and social connection) for all individuals. In clinical practice, more attention needs to be paid to the prevention, identification, and treatment of cognitive impairment in individuals with a low number of healthy lifestyle habits. Furthermore, decreasing the cognition risk could potentially be achieved through promotion of modifiable lifestyles. This study could contribute to the establishment of personalized preventative measures and population-level interventions against cognitive impairment. In addition, a long-term plan of health behavior promotion can be a target for preventing cognitive decline, or even cognitive impairment. Further studies are needed to consider the combined impact of multiple lifestyle factors on dementia as well as neurodegenerative diseases, so that early identification of behavioral risks and intervention could suggest an initial intervention in preserving cognitive impairment in the general population.

4.4. Study Limitations

There are several limitations to this study. First, lifestyle information was collected through self-reports in CLHLS, which may have caused measurement errors. Participants’ lifestyles are not static, and analyses of repeated measures should be considered in the future. Second, a limited number of lifestyle factors were mainly included in this study, and other environmental factors may also influence cognitive impairment. Third, the heterogeneity might come from differences in geography, social demographics, measuring approach and other unmeasured factors. Due to space constraints, further analyses were not performed. Fourth, considering the nature of this observational study, we could not rule out the influence of unmeasured confounding factors on observed associations and it is difficult to derive causality. Well-designed interventional studies are needed. Finally, this study only included older populations in China. The combined effect of long-term multiple behavioral lifestyles on cognition performance among younger and middle-aged adults should be considered in future studies.

5. Conclusions

Adherence to a broad range of healthy lifestyle factors was associated with a significantly lower risk of cognitive impairment. Behavioral lifestyle modification through multifactorial approaches should be a priority for prevention and delayed onset of cognitive impairment and dementia in middle-aged and older adults.