Associations between Sedentary Duration and Cognitive Function in Older Adults: A Longitudinal Study with 2-Year Follow-Up

Abstract

Objectives

This study aimed to investigate the association between different forms of sedentary behavior and cognitive function in Chinese community-dwelling older adults.

Design

A longitudinal study with a 2-year follow-up.

Setting and Participants

Data from 5356 participants at baseline and 956 participants at the follow-up of the Anhui Healthy Longevity Survey (AHLS) were analysed.

Measurements

Cognitive function was evaluated by the Mini-Mental State Examination (MMSE), and mild cognitive impairment (MCI) was classified according to education-specific criteria. Self-report questionnaires were used to assess the sedentary behavior of the participants.

Results

The participants who reported longer screen-watching sedentary duration had higher MMSE scores (1–2 hours: β=0.758, 95% CI: 0.450, 1.066; > 2 hours: β=1.240, 95% CI: 0.917, 1.562) and lower likelihoods of MCI (1–2 hours: OR= 0.787, 95% CI: 0.677, 0.914; >2 hours: OR=0.617, 95% CI: 0.524, 0.726). The participants who had played cards (or mahjong) sedentary had higher MMSE scores (β= 1.132, 95% CI: 0.788, 1.476) and lower likelihoods of MCI (OR=0.572, 95% CI: 0.476, 0.687). However, the participants who reported longer other forms of sedentary duration had lower MMSE scores (1–2 hours: β=−0.409, 95% CI: −0.735, −0.082; > 2 hours: β=−1.391, 95% CI: −1.696, −1.087) and higher likelihoods of MCI (1–2 hours: OR=1.271, 95% CI: 1.081, 1.496; > 2 hours: OR=1.632, 95% CI: 1.409, 1.889). No significant association was detected between sedentary duration and MCI incidence.

Conclusion

Variations in the impact of diverse sedentary behaviors on the cognitive function were detected in Chinese older adults. However, such associations were cross-sectional and longitudinal associations were not found in the current study.

Introduction

Alzheimer's disease (AD) is a slowly progressive neurodegenerative disease that contributes to the predominant subtype of dementia. It is characterized by the appearance of multiple cognitive deficits progressively increasing with time and independence in personal daily activities (1). At present, there are approximately 50 million AD patients worldwide, and this number is projected to double every 5 years and will increase to 152 million by 2050, which can ultimately cause substantial increases in healthcare burdens (2, 3). AD has become a major public health problem worldwide, and many treatment methods have been explored for several decades; however, there is still no cure for AD. Therefore, the priority remains prevention, which requires the development of effective lifestyle-based strategies (4, 5). Mild cognitive impairment (MCI) is a transitional stage between healthy cognition and dementia, so it is an important stage for AD intervention (6). Providing effective intervention strategies to maintain cognitive health during this transition period might slow the conversion to dementia.

Accumulating literature suggests a potential link between sedentary behavior and many adverse health outcomes, including cognitive and physical health (7, 8). Sedentary behavior is usually defined as waking behavior in a reclining or seating position with low energy expenditures, and the negative impact of sedentary behavior on health can be observed even in the presence of regular physical activity. One possible reason is that sedentariness involves a lack of physical activity and stimulation provided by other activities (9); therefore, physical activity may only attenuate but does not eliminate the increased risk of sedentary behavior (10). Generally, sedentary behavior is highly prevalent among older adults; for example, a study showed that Canadian older adults spend approximately 80% of their time in a seated posture, and more than half of them were sedentary up to 8.5 hours per day (11). Existing studies on sedentary behavior have primarily been conducted with a focus on sedentary patterns (e.g., total duration, sedentary forms and breaks) and their associations with cognition, although limited literature has drawn inconsistent results. For example, although some studies revealed that longer sedentary duration is associated with poorer cognitive performance (12), several studies concluded the opposite (13, 14). In addition, different forms of sedentary behavior might be related to dementia risk differently; for instance, driving sedentary duration was associated with higher risks of dementia (15); however, longer computer and mobile phone use duration was associated with better cognitive function (16, 17). Generally, individuals with cognitive impairment tend to be physically inactive and have a longer sedentary duration (18); however, recent studies have indicated that different breaking strategies of sedentary may have different impacts on cognition. For example, a study showed that frequently breaking up sedentary periods with active strategies (e.g., breaks every 30-min sitting with 2-min walking) may have important implications for cognitive function (19). The abovementioned evidence highlights the importance of further investigating sedentary behavior in relation to cognitive function; however, available evidence is sparse, especially those obtained from developing countries such as China.

This study aimed to examine the cross-sectional and prospective associations between different forms of sedentary behavior (screen-watching, playing cards, and others) and cognitive function in older adults dwelling in urban and rural communities of China. The elucidation of this association could help develop strategies to reduce the prevalence of dementia.

Methods

Study population and data collection

Participants were recruited from the Anhui Healthy Longevity Survey (AHLS), the details of which were previously reported (20). Briefly, a multistage sampling strategy was used to obtain a representative sample. Four cities (Chuzhou, Lu'an, Xuancheng, and Fuyang) were purposely selected to represent their different geographic characteristics, followed by 3–5 urban and rural communities selected in each city. Older dwellers (aged 60 and older) of the selected communities with basic communication ability were invited to participate in the study. Enrolment occurred from July to August 2019. In total, 6211 participants who were aged 60 or higher were initially enrolled.

Data on demographic characteristics, health-related behaviors and chronic status were self-reported by the participants and collected with questionnaires. Standardized training of the data collection process was provided to each data collector prior to the study. The returned questionnaires were cross-checked on the survey day, and the poorly recorded items were confirmed and revised. Collected data were double entered, and inconsistencies were verified to ensure accuracy.

Fig 1 displays the flowchart of obtaining the analytic sample. The following exclusion criteria were applied: (1) not able to finish the cognitive function assessment or sedentary behavior assessment (n = 379) or (2) missing data for any covariate variable (n = 476). Finally, 5356 participants were included as baseline data. After two years, 2308 participants in the first wave of follow-up areas (rural areas of Chuzhou, Lu'an, and Xuancheng) were included. Among the 2308 rural participants who were investigated at baseline, 1566 participants completed the follow-up in 2021, for a response rate of 67.9% (742 participants were lost to follow-up, including 88 cases of death, 23 cases of loss of communication ability, 201 cases of migration and 430 cases of refusal to participate). The participants were further excluded if they (1) were not able to finish the cognitive function assessment or sedentary behavior assessment (n = 71) or (2) had any missing covariate data (n = 59). Therefore, 1436 participants were included in the follow-up analysis. In addition, for the analysis of the relationship between sedentary behavior and 2-year MCI incidence, participants with preexisting MCI at baseline (n = 480) were excluded, leaving 956 eligible participants for analyses. In summary, the current study analysed data from 5356 participants in the baseline survey (2019) and 956 participants in the follow-up survey (2021). All procedures complied with the ethical standards of the Anhui Medical University committee (No. 2020H011).

Figure 1.

Flow diagram of presenting the study sample for both cross-sectional and longitudinal analyses

Assessment of Cognitive Function

Cognitive function was measured by using the Mini-Mental State Examination (MMSE). It is a widely applied tool for evaluating cognitive function with good validity and reliability (21, 22). The domains of the MMSE include memory, orientation, language, attention, and computation. The maximum score is 30 points, with higher MMSE scores indicating better cognitive function. The participants with mild cognitive impairment (MCI) were classified according to education-specific criteria. It has been reported that the Chinese version of the MMSE indicates MCI if the scores are ≤ 17, 20, and 24 points for people with an educational level of illiteracy (0 years of education), elementary school (1–6 years of education), and secondary school or above (more than 6 years of education), respectively (23).

Assessment of sedentary behavior

Data on total hours of sedentary duration were self-reported by the participants. The classification of sedentary behavior was inspired by the literature (24, 25) and verified by several community visits in the pilot study. Since screen-watching sedentary (e.g., sitting and watching TV or playing cell-phone) and playing cards (or mahjong) are very popular among older adults in the context of Chinese culture, data on total sedentary duration and sedentary duration on screen-watching and playing cards (or mahjong) were collected. The participants were asked “How many hours do you usually spend sitting or reclining watching the screen (TVs, cell phones, computers, etc.)/playing cards (or mahjong) on a typical day?” Duration of other forms of sedentary were obtained by subtracting screen and playing cards (or mahjong) sedentary duration from the total sedentary duration.

Total sedentary duration, screen-watching sedentary duration and other forms of sedentary duration were divided based on approximate tertiles. Specifically, the total sedentary duration was divided into three groups (0–3 hours, 3–5 hours, and more than 5 hours); the screen-watching sedentary duration and other forms of sedentary time were categorized into three groups (0–1 hour, 1–2 hours, and more than 2 hours). However, playing cards (or mahjong) sedentary duration was divided into two categories (with, without) due to the extremely skewed distribution of the data of playing cards (or mahjong) sedentary duration.

Assessment of Covariates

Demographic characteristics and other potential confounders, including city, sex, age, education, marital status, annual income, living alone, body mass index (BMI), drinking status, smoking status, self-rated sleeping quality, physical activity, hypertension, diabetes, and depression status, were considered the covariates in this study. Education was categorized as low (0 years of education), medium (1–6 years of education), and high (>6 years of education). Marital status was divided into two groups: married and not married (divorced, widowed, or never married). Annual income was classified into two groups (lower than 6,500 RMB and 6,500 RMB or higher). BMI was calculated by measured weight (in kilograms) and height (in meters) and introduced as a continuous variable (formula: BMI= weight /height²). Drinking status was categorized as never drinkers, former drinkers, and current drinkers. Smoking status was categorized as current smokers and noncurrent smokers. Self-rated sleep quality was divided into three groups (good, fair, and poor). Participants were asked to answer the question regarding their regular physical activity: “Do you have a target goal (e.g., a certain number of steps, distance or exercise time) of your daily exercise?” The participants who answered “yes” were grouped into with physical activity. Depression symptoms of the participants were evaluated by the Patient Health Questionnaire-9 (PHQ-9), a 9-item self-report scale of depression symptoms with higher scores indicating more severe depression symptoms. The depression status was divided into two groups according to the existence of the depression symptoms identified by PHQ-9 scores (without: 0–4; with: 5 and higher) (26). In addition, participants' chronic conditions (diabetes and hypertension), which were diagnosed by physicians in the facilities at the county level or higher, were recorded.

Statistical Analyses

Multivariable linear regression and logistic regression strategies were employed to analyse the associations between different forms of sedentary duration and cognitive function. Categorical sedentary duration was treated as an independent variable. Specifically, total sedentary duration, screen-watching sedentary duration, playing cards (or mahjong) sedentary duration, and other forms of sedentary duration were introduced as independent variables. Cross-sectional associations between different forms of sedentary duration and MMSE scores and prevalent MCI were analysed by linear regression and logistic regression, respectively.

A multiple logistic regression strategy was employed to assess the association between different forms of sedentary duration and 2-year MCI incidence. The dependent variable was set as 2-year incident MCI, which referred to the number of participants who developed MCI within the 2-year follow-up. Independent variables, i.e., different forms of sedentary duration, were introduced into the models with all the covariates. Pearson correlation tests and collinearity tests were used prior to the main analysis to avoid collinearity between the independent variables.

In addition, three additional analyses were also conducted. First, stratified analyses were performed to assess the possible influence of several confounding factors (age, gender, urban or rural settings, education level, physical activity, diabetes, and hypertension), and the interactions were tested via the interaction term approach. Specifically, multiplicative interaction terms were added to each multivariable model to assess effect modification by the potential confounders. Second, the analyses were conducted by using the imputed dataset, which was generated by mean imputing each variable with missing values. Third, continuous hours of sedentary duration were used to test the robustness of the results.

The regression coefficient (β), odds ratio (OR), and 95% confidence interval (CI) were used to measure the strength of the associations. Statistical analyses were conducted by Stata 16.0 (College Station, TX, United States). A two-sided P value<0.05 was considered statistically significant.

Results

Participant characteristics are shown in Table 1. In total, 5356 participants (46.3% of them were male) were included in the main analysis. The mean age of the participants was 70.91 years, and the mean BMI value of the participants was 24.2 kg/m². Nearly half of the participants (48.8%) were illiterate. More than half of the participants (50.4%) reported having high blood pressure, and a total of 15.7% of the participants reported having diabetes. Approximately one-third of the participants (31.5%) were classified as having depression symptoms. The mean MMSE scores was 21.68.

Table 1.

Basic characteristics of the sample

			N	Mean (SD)/Percentage
City
Fuyang (North)			1133	21.2
Lu'an (West)			1541	28.8
Chuzhou (East)			1527	28.5
Xuancheng (South)			1155	21.6
Gender
Male			2482	46.3
Female			2874	53.7
Age			5356	70.91 (7.0) *
Education
Low (0 year of education)			2615	48.8
Medium (1∼6 years of education)			1505	28.1
High (more than 6 years of education)			1236	23.1
Marital status
Married			3891	72.6
Not married			1465	27.4
Annual income
<6500			3228	60.3
≥6500			2128	39.7
BMI			5356	24.2 (3.7) *
Underweight (BMI<18)			275	5.1
Normal (BMI 18∼24)			2368	44.2
Overweight (BMI >24)			2713	50.7
Drinking status
Never			3035	56.7
Former			218	4.1
Current			2103	39.3
Current smoker
Yes			1154	21.5
Physical activity
Yes			2348	43.8
Self-rated sleeping quality
Very good			1092	20.4
Good			2967	55.4
Not good			1297	24.2
Hypertension
Yes			2697	50.4
Diabetes
Yes			841	15.7
Living alone
Yes			966	18.0
Depression (PHQ-9)
Yes			1688	31.5
MMSE scores			5356	21.68 (6.0) *
MCI
Yes			3678	68.7
Sedentary Duration
Total	<3h		2147	40.1
	3∼5 h		1716	32.0
	>5h		1493	27.9
Screen-watching		<1h	2593	48.4
		1∼2h	1428	26.7
		>2h	1335	24.9
Playing cards		(Yes)	948	17.7
Other forms		<1h	2428	45.3
		1∼2h	1227	22.9
		>2h	1701	31.8

* The data are expressed as means with standard deviation in parentheses.

Fig 2 shows the cross-sectional and longitudinal associations between different forms of sedentary duration (total sedentary duration, screen-watching sedentary duration, playing cards (or mahjong) sedentary duration, and other forms of sedentary duration) and cognitive function. The participants who reported longer sedentary duration had higher MMSE scores (1–2 hours: β=0.758, 95% CI=0.450, 1.066; > 2 hours: β=1.240, 95% CI=0.917, 1.562) and lower likelihoods of MCI (1–2 hours: OR= 0.787, 95% CI: 0.677, 0.914; > 2 hours: OR=0.617, 95% CI=0.524, 0.726). The participants who had play cards (or mahjong) sedentary duration had higher MMSE scores (β=1.132, 95% CI=0.788, 1.476) and lower likelihoods of MCI (OR=0.572, 95% CI=0.476, 0.687). The participants who reported longer other forms of sedentary duration had lower MMSE scores (1–2 hours: β=−0.409, 95% CI=−0.735, −0.082; > 2 hours: MMSE scores: β=−1.391, 95% CI: −1.696, −1.087) and higher likelihoods of MCI (1–2 hours: OR=1.271, 95% CI=1.081, 1.496; > 2 hours: OR=1.632, 95% CI=1.409, 1.889). No significant association was detected between different forms of sedentary duration and 2-year MCI incidence.

Figure 2.

The associations between different forms of sedentary duration and cognitive function in multivariable linear regression and logistic regression analysis

The model adjusted city (Fuyang, Chuzhou, Lu'an and Xuancheng), age (≤70 and >70), sex (male and female), education (0, 1‘-6 years and more than 6 years), marital status (married and not-married), annual income (lower than 6,500 RMR and 6,500 RMR or higher), BMI (continuous), drinking status (never, former and current), current smoker (yes and no), physical activity (yes, no), self-rated sleeping quality (very good, good and not good), diabetes (yes and no), hypertension (yes and no), living alone (yes and no), depression at baseline (yes and no). a Multivariable linear regression and logistic regression analysis included 5356 participants. b Multivariable logistic regression analysis included 956 participants.

The results obtained from the additional analyses were generally consistent with the main analysis. Fig S1–Fig S7 show the results of the first way of additional analyses stratified by selected possible confounders (age, gender, urban or rural settings, education level, physical activity, diabetes, and hypertension). The test for interaction suggested possible interactions for gender, urban or rural settings, education level, and physical activity (p for interaction <0.05) but not for the other factors. The second method of additional analysis, i.e., the analyses conducted by using the imputed dataset, also revealed similar results to the main analysis (data not shown). The results of the third additional analysis conducted using continuous sedentary duration are shown in Fig S8. Every 1-hour increase in total sedentary duration was associated with a 0.053-unit decrease in MMSE scores; however, every 1-hour increase in screen-watching and playing cards (or mahjong) sedentary duration was associated with an approximately 0.3-unit increase in MMSE scores and less like to have MCI. Additionally, increased other forms of sedentary duration were associated with poorer cognitive performance.

Discussion

The current study investigated the relationship between sedentary behavior and cognitive function among older adults dwelling in urban and rural communities in Anhui, China. It has been revealed that different forms of sedentary behavior might have different influences on cognitive health based on cross-sectional data. Therefore, the study highlighted the great importance of differentiating the forms of sedentary behavior when providing health recommendations regarding sedentary behavior management for older adults. To date, the current study is the first to investigate the influence of different forms of sedentary behavior on cognitive health among older Chinese adults, and the findings imply the need to refine health recommendations regarding sedentary behavior, although the evidence was derived from cross-sectional data only.

Existing literature has revealed that prolonged uninterrupted sitting may reduce the automatic regulation and blood flow of the brain, which may in turn lead to a decline in cognitive performance (19). However, recent findings in many recent studies suggest that different forms of sedentary behavior are differentially associated with cognitive function (27, 28). For example, accumulated evidence suggests that sedentary screen-watching (including computer and smart-phone use and watching TV) duration might be associated with better cognition in older adults (17, 29, 30, 31, 32). Screening-related activities (e.g., computer and smartphone use) may increase brain sensory stimulation, which may be helpful for improving visual-spatial function and attention (33), and may also activate brain regions that are important for decision-making and complex reasoning to strengthen cognition (34).

Sedentary duration spent playing cards might also have potential benefits for maintaining cognition. Cognitive leisure activities, such as playing cards, require active cerebral effort, including sustained memory and calculation tasks, and thereby contribute to higher cognitive functions (28, 35). Another possible explanation may be that playing cards enhances social engagement, which may improve speech fluency, numeracy and memory (36, 37). However, some forms of sedentary behavior, especially passive sedentary behavior with less cognitive activity (e.g., leisure sitting), might have a negative impact on cognition, as those behaviors may weaken executive functions and memory tasks, which lead to worse cognitive performance (38).

Moderators were also found in the current study. Gender-specific associations might be partially explained by physiological and behavioral reasons. Different susceptibilities to cognition impairment exist between genders (39, 40), and the levels of exposure to behavioral risk factors (such as smoking) have significant gender differences. Sedentary patterns may also vary among genders (41). Older adults who live in rural areas are recognized as a vulnerable group for cognitive decline (42). A possible explanation is the relatively low socioeconomic development in rural areas, and lifestyle (including sedentary behavior) is very different among urban and rural older dwellers in China (43, 44). Education influences cognitive function in late life primarily by inducing individual differences in cognitive skills in the whole life span (45). A previous study also suggested that well-educated older adults tended to benefit more in cognition than undereducated individuals from leisure activities (46). Physical activity is believed to be beneficial to cognitive health; however, its competing nature with sedentary behavior should also be taken into consideration. A recent study indicated that physical activity may attenuate the association between sedentary behavior and cognitive function among older adults (47). However, there is no denying that it was difficult to clearly clarify the reasons for the moderators at the current stage, so future studies with more complex designs are needed to further illustrate this phenomenon. In addition, an association between different forms of sedentary duration and the 2-year incidence of MCI was not detected in the study. This may be due to the relatively short follow-up period, which is not sufficient to observe the true associations.

This study has the following advantages. First, this study was a prospective study with a 2-year follow-up and an appropriate follow-up response rate. A large-size and representative sample was recruited, which increased the generalizability of the current study. Second, multiple covariates were introduced in this study to minimize the impact of confounding. However, this study still has some limitations. First, the current study only specified popular sedentary leisure activities, i.e., screen-watching and playing cards, making it difficult to illustrate other forms of sedentary behavior. Moreover, sedentary patterns (e.g., sedentary period, difference in weekdays and weekend, break) were not considered in the current stage. Future studies should incorporate other types of sedentary behavior (for example, reading and listening to music) as well as different sedentary patterns. Second, self-reported sedentary duration was used for analysis, which might be subjective and influenced by recall and response problems (48, 49) and therefore may contribute to the overreporting of sedentary behaviors (50). Third, the current study could not include many potential biological covariates, such as vascular function and inflammation (51), although we have considered a wide range of covariates. Fourth, the MMSE is currently a commonly used screening tool for cognitive impairment rather than a diagnostic tool. This limitation makes it difficult to capture the true association between sedentary behavior and cognitive function. Future studies may consider using a standard diagnostic procedure to increase the accuracy of the measurement. Fifth, despite the longitudinal design of the current study, the follow-up period is short, which may not be able to demonstrate the true associations between sedentary behavior and cognitive performance. Future longitudinal studies with longer follow-up should be conducted for further investigation. Additionally, although a previous study indicated that medication use for chronic disease may be related to sedentary duration and cognition (52, 53), such a variable was not involved in the current study and thus needs to be further investigated in the future. Finally, the excluded participants due to missing data seemed to have a potential influence on the findings; however, the robust results obtained from the sensitivity analysis suggested that the impact might not be significant.

Conclusion

The findings of the present study highlighted the potential variations in the impact of diverse sedentary behaviors on the cognitive function of Chinese older adults. However, it is important to note that these associations were observed in a cross-sectional design, necessitating further validation through a multicenter longitudinal study encompassing a substantial sample size. Future health recommendations regarding sedentary behavior should be provided in detail, with full consideration of the specific forms of sedentary behavior for older adults, to meet the future challenges of AD.

Electronic supplementary material

Supplementary material is available for this article at https://doi.org/10.1007/s12603-023-1963-4 and is accessible for authorized users.

Supplementary material, approximately 1.93 MB.

Author Contributions: SZ drafted the manuscript. GS and YZ framed the concept and designed the study. Data collection and material preparation were conducted by SZ, JZ, QinW, and QioW, and data analysis was performed by WZ and CD. All authors meet the criteria for authorship according to their contributions to the manuscript. All authors contributed to the article and approved the submitted version.

Funding: This research was funded by the National Natural Science Foundation of China, Grant Number 72004003 to YZ, the Key Project of Science and Technology of Anhui Province, Grant Number 202004b11020019 to GS, and the Hefei Municipal Natural Science Foundation, Grant Number 2021005 to GS.

Ethics approval and consent to participate: The study protocol was approved by the Ethics Review Committee of Anhui Medical University, Hefei, China (No. 2020H011). Informed consent was obtained from all participants included in the study.

Conflict of Interest: None declared. The research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.