Abstract
Background:
Some studies have demonstrated an association between low and high body mass index (BMI) and an increased risk of dementia. However, only a few of these studies were performed in rural areas.
Objective:
This cross–sectional study investigated the associations between BMI and cognitive impairment among community–dwelling older adults from rural and urban areas.
Methods:
8,221 older persons enrolled in the Hubei Memory & Ageing Cohort Study (HMACS) were recruited. Sociodemographic and lifestyle data, comorbidities, physical measurements, and clinical diagnoses of cognitive impairment were analyzed. Logistic regression was performed to assess the associations of BMI categories with cognitive impairment. A series of sensitivity analyses were conducted to test whether reverse causality could influence our results.
Results:
Being underweight in the rural–dwelling participants increased the risk of cognitive impairment. Being overweight was a protective factor in rural–dwelling participants aged 65–69 years and 75–79 years, whereas being underweight was significantly associated with cognitive impairment (OR, 1.37; 95% CI: 1.03–1.83; p < 0.05). Sensitivity analyses support that underweight had an additive effect on the odds of cognitive impairment and was related to risk of dementia. Interaction test revealed that the differences between urban/rural in the relationship between BMI and cognitive impairment are statistically significant.
Conclusion:
Associations between BMI and cognitive impairment differ among urban/rural groups. Older people with low BMI living in rural China are at a higher risk for dementia than those living in urban areas.
Keywords
INTRODUCTION
The increasing prevalence of dementia and obesity represents a significant public health concern. Although mounting evidence has shown that higher late–life body mass index (BMI) is associated with reduced risk of dementia and Alzheimer’s disease (AD) [1–4], inconsistent results have been reported regarding the associations between BMI and the risk of cognitive impairment and dementia. A recent meta–analysis showed that being overweight and obese in late–life reduces the risk of dementia and cognitive impairments by 21% and 25%, respectively [5]. However, another meta–analysis concluded that the current evidence on the beneficial impacts of obesity and being overweight in older age on incident dementia is inconsistent [6]. In addition, most of these epidemiological data were from studies conducted in high–income countries; very few studies were conducted in low–income and middle–income countries.
The prevalence of mild cognitive impairment (MCI) among older adults ranges from 13.55% to 21.46% [7]. However, studies in which the association between BMI obesity and cognitive impairment is investigated are relatively scarce. Only a few small population–based studies have been conducted to investigate this association; however, the results of those studies were inconsistent [8–10]. Thus, the effect of obesity on cognitive function in adults, independent of obesity–related comorbidities, remains ambiguous.
Presently, there are approximately 9 million patients with AD in China. In addition, more than half of Chinese senior citizens live in rural areas [11]. A notably higher prevalence of dementia and AD has been recorded in rural areas than in urban ones [12]. The urban–rural disparity in many domains in China is large compared with those in some nations. Nutritional and behavioral assessment should be considered additional tools in the evaluation of urban–rural disparity. However, no adequately designed study of the prevalence of undernutrition and cognitive impairment has been conducted among community–dwelling older adults in China. Therefore, more studies with urban–and rural–dwelling populations are needed to clarify the association between body weight in older age and risk of dementia. In response to this urgent need, this study was conducted to estimate the prevalence of different BMI categories and to evaluate the association between BMI and the risk of dementia in community–dwelling older adults; we also assessed the differences between rural and urban areas in China in this regard. We found that the associations between BMI and cognitive impairment differ among rural–and urban–dwelling older adults. Older people with low BMI living in rural China are at a higher risk for dementia than those living in urban areas.
MATERIALS AND METHODS
Study design and participants
This was a cross–sectional study of the survey data of the Hubei Memory & Ageing Cohort Study (HMACS), which was carried out between 2018 and 2020. HMACS is the first prospective community–based cohort study of cognitive impairment in Central China, with residents sampled from both metropolitan and remote rural areas. A total of 48 villages within a county and 31 neighborhoods within four city communities were sampled in the study. Four city community health centers in Wuhan and four township hospitals in Dawu County were selected as the medical facilities for interviews and clinical examinations.
Flowchart of subject recruitment for the study participants.
Since China has established an electronic health record (EHR) for each old citizen aged≥65 years, all eligible residents living within the sampled villages and neighborhoods were identified using the EHR kept in the health centers and the hospitals. Potential participants were excluded if they 1) had a life–threatening disease, 2) their medical record indicated severe schizophrenia or mental retardation, or 3) had severe problems with vision and hearing and were not able to communicate and undergo physical and cognitive examinations. We excluded residents who did not meet the inclusion criteria (n = 2,916) from the registered residents (n = 13,246), and excluded residents who refused to complete the questionnaire or could not be tracked (n = 2,109) and residents who could not participate in the assessment due to hearing or vision deficiencies. A total of 8,221 residents completed all in-person evaluations were then included in this study. All the procedures used in this study were approved by the ethics committee of the medical school of Wuhan University of Science and Technology, Wuhan, China (protocol code: 201845; approved on October 22, 2018). Written informed consent was obtained from all participants or their legally acceptable representatives.
Interviews, medical examination, and neuropsychological assessment
Data were collected through face–to–face interviews, physical examination, clinical examination, neuropsychological assessment, and laboratory tests. The structured questionnaire included questions on social demographic characteristics, lifestyle habits, medical history, and family history. Data on medical history included information on hypertension, diabetes, cerebrovascular disease, and coronary artery disease diagnosed by a physician and confirmed from the participants’ medical records. The neuropsychological assessments included the following: 1) use of the Chinese version of the Mini–Mental State Examination to evaluate global cognition; 2) use of the Chinese version of the Montreal Cognitive Assessment–basic to assess mild cognitive impairment [13]; 3) use of the Chinese version of the Auditory Verbal Learning Test [14], Trail Making Tests A and B, the forward and backward conditions of the Digital Span Test, the Boston Naming Test [15], and the Clock Drawing Test to assess five cognitive domains. The Clinical Dementia Rating scale was used by professional neurologists to assess cognitive impairment [16]. Depressive symptoms were assessed using the Geriatric Depression Scale [17]. Items from the Lawton and Brody Activity of Daily Living scale were used to determine physical self–maintenance and instrumental activities of daily living [18].
Height and weight were measured when the participants were wearing light clothes. BMI was calculated as weight (kg) divided by the square of height (m). According to the Chinese standard of BMI [19], BMI was divided into four groups: underweight (< 18.5 kg/m2), normal weight (18.5–23.9 kg/m2), overweight (24.0–27.9 kg/m2), and obesity (≥28 kg/m2).
Cognitive diagnoses
After each participant was examined, a panel of two neuroscientists and two neuropsychologists with expertise in dementia diagnosed dementia and MCI according to the criteria outlined in the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM–IV) [20]. The diagnosis of MCI was determined by a panel of experts who reviewed participants’ data according to the Petersen’s criteria [21], namely: 1) patients, insiders, and/or clinicians reported or noted cognitive impairment; 2) objective evidence of impairment in at least one cognitive domain not limited to memory; 3) retention of independent functional ability while instrumental ability may be slightly impaired (assessed using ADL); 4) not diagnosed as dementia (assessed using DSM–IV).
Statistical analysis
SPSS 26.0 was used for all statistical analyses (SPSS, Inc., Chicago, IL, USA). The Kolmogorov–Smirnov test was used to determine the normality of the distribution of continuous variables. Since the continuous variables included in the study were non–normally distributed, the data are presented as medians (interquartile ranges) and classified variables are presented as numbers (percentages). The Mann Whitney U test and chi square test were used to compare the statistical differences between the continuous variables and categorical variables of the participants with normal cognition and those with cognitive impairment. Univariate logistic regression was used to evaluate the correlation between baseline factors and cognitive impairment. A multivariate logistic regression model was used to explore the relationship between BMI and cognitive impairment. Covariates included sociodemographic variables, health behavior variables, and health status variables. Sociodemographic variables included age, sex, and education. Health behavior variables included cigarette smoking, alcohol consumption, and physical activity. Comorbidity variables included hypertension, diabetes, coronary heart disease, and cerebrovascular disease. We further stratified the participants according to age, which is recognized as an important factor of cognitive impairment. We used an age–stratified logistic regression model to evaluate the relationship between BMI and cognitive impairment in different age groups. We also used logistic regression to perform interactions test between BMI categories and rural/urban areas to see whether the urban/rural differences were statistically significant. Finally, we conducted a series of sensitivity analyses to test whether reverse causality could influence our results. Firstly, we separated analyses for MCI versus dementia for individuals aged 80 years and above, and 30% of persons (n = 18) with BMI closer to the low end of normal weight were additionally included into the underweight group. Secondly, we removed individuals with dementia whose ADL scores showed severe loss of activities of daily living and repeated the fully adjusted model to explore the relationship between being underweight and cognitive impairment (include both MCI and dementia). These analyses were also performed for younger urban old adults to check the reverse causality. All p values were bilateral, and the results were considered statistically significant at p < 0.05.
RESULTS
Body mass index, sociodemographic characteristics, lifestyle, and comorbidities
The characteristics of the participants, including BMI, sociodemographic characteristics, lifestyle habits, and comorbidities, are outlined in Table 1. A total of 8,221 eligible older adults aged≥65 years were included in this study; 4,449 (54.1%) were women, 3,164 (38.4%) were from remote rural areas, 2,576 (31.3%) had cognitive impairment, and 5,645 had normal cognition. The mean age of the study cohort was 71.96 years (SD, 5.895) and the mean duration of their education was 7.58 years (SD, 5.359). Among the 8,221 participants included, 396 (4.8%) persons were underweight, 2,808 (34.1%) were overweight, and 836 (10.1%) were obese. Older adults with cognitive impairment accounted for 39.3% and 26.3% of the rural–and urban–dwelling participants, respectively. There were significant differences between the age and education levels of the overall, rural–dwelling, and urban–dwelling participants in the normal cognition and the cognitive impairment groups. The cognitive impairment group had lower mean BMI than the normal cognition group; the differences were statistically significant for all the participants (p = 0.007) and the rural–dwelling participants (p = 0.011), but not for the urban–dwelling participants. The differences between the prevalence rates of diabetes for the overall, rural–dwelling, and urban–dwelling participants (p = 0.001, 0.040, and 0.001, respectively) in the normal cognition and cognitive impairment groups were statistically significant. The difference in the prevalence rate of hypertension for all the participants in the normal cognition and cognitive impairment groups were also statistically significant. The rate of physical activity was higher for all the participants (p < 0.001) and the urban–dwelling participants (p < 0.001) in the normal cognition group than for all the participants and the urban–dwelling participants in the cognitive impairment group. Sex, the rates of smoking and drinking, and the prevalence of cerebrovascular disease and coronary heart disease were comparable between the normal cognition and cognitive impairment groups.
Body mass index, sociodemographic characteristics, lifestyle habits, and comorbidities of the participants
*p < 0.05; **p < 0.01.
Risk factors associated with cognitive impairment
As shown in Table 2, we used univariate logistic regression to evaluate the correlation between risk factors and cognitive impairment. The risk of cognitive impairment increased gradually with increase in age (p < 0.001). This finding was more obvious among the urban–dwelling participants. Higher education was associated with a lower risk of cognitive impairment in urban–dwelling participants (p < 0.001). Surprisingly, lower education was associated with lower risk of cognitive impairment in rural–dwelling participants. Physical activity reduced the risk of cognitive impairment by 36.7% (OR, 0.63; 95% CI: 0.53–0.75; p < 0.001) in urban–dwelling participants.
Univariate analysis of factors associated with cognitive impairment
*p < 0.05; **p < 0.01.
Diabetes was undoubtedly associated with an increased risk of cognitive impairment regardless of whether the participants lived in rural (OR, 1.23; 95% CI: 1.01–1.51; p < 0.05) or urban (OR, 1.31; 95% CI: 1.12–1.53; p < 0.001) areas. Compared with normal weight (18.5–23.9 kg/m2), being underweight (< 18.5 kg/m2) increased the risk of cognitive impairment by 34.2% in rural–dwelling participants (OR, 1.34; 95% CI: 1.01–1.79; p < 0.05), whereas being overweight reduced the risk by 20.9% (OR, 0.79; 95% CI: 0.67–0.93; p < 0.01). There was no significant correlation between BMI classification and the risk of cognitive impairment in urban–dwelling participants.
As shown in Table 3, we further evaluated the correlation between risk factors and cognitive impairment using multiple logistic analyses. After adjustment by controlling sociodemographic characteristics (age, sex, education) and lifestyle habits (smoking, drinking, physical activity), we found that being overweight was inversely correlated with cognitive impairment in rural residents (OR, 0.79; 95% CI: 0.67–0.94; p < 0.01). However, obesity increased the risk of cognitive impairment by 24.6% in urban–dwelling participants (OR, 1.25; 95% CI: 1.00–1.55; p < 0.05). In model 2, we adjusted for comorbidities (hypertension, diabetes, and cerebrovascular disease) and found that being underweight was positively correlated with cognitive impairment in rural residents (OR, 1.37; 95% CI: 1.03–1.83; p < 0.05) and in all the participants (OR, 1.43; 95% CI: 1.16–1.77; p < 0.01), but not in urban–dwelling participants (p > 0.05). Being overweight was inversely correlated with cognitive impairments in rural–dwelling participants (OR, 0.77; 95% CI: 0.65–0.91; p < 0.01) and in all the participants (OR, 0.83; 95% CI: 0.74–0.92; p < 0.001). In model 3, we adjusted all the covariates included in models 1 and 2 (nine variables) and noted similar results in model 2. Being underweight was a risk factor for dementia in all the participants (OR, 1.25; 95% CI: 1.00–1.55; p < 0.05); however, the correlation between being underweight and dementia in rural participants was not statistically significant (p > 0.05). Taken together, our findings revealed that being underweight is associated with a more favorable dementia risk profile in rural–dwelling participants but not in urban–dwelling ones, suggesting that being overweight may not be a candidate risk for prevention of dementia in rural residents.
Multiple logistic linear analyses of factors associated with cognitive impairment after adjustment for covariates
*p < 0.05; **p < 0.01. Model 1 adjusted for age, sex, education level, physical exercise, smoking, and drinking. Model 2 adjusted for hypertension, diabetes, and cerebrovascular disease. Model 3 adjusted for age, sex, education level, physical exercise, smoking, drinking, hypertension, diabetes, and cerebrovascular disease.
Relationship between BMI and cognitive impairment at different ages, adjusted in various ways
*p < 0.05; **p < 0.01. Model 4 adjusted for sex, education level, physical exercise, smoking, and drinking. Model 5 adjusted for sex, education level, physical exercise, smoking, drinking, hypertension, diabetes, and cerebrovascular disease.
Age is the most important risk factor for cognitive impairment. To explore whether the association between BMI classification and cognitive impairment changes with age, we divided the participants into four age groups (65–69 years old, 70–74 years old, 75–79 years, and≥80 years old) and analyzed the association using multivariate logistic regression. In model 4, we found that being underweight increased the risk of cognitive impairment 2.04 times in rural–dwelling participants aged≥80 years (OR, 2.04; 95% CI: 1.01–4.14; p < 0.05). In model 5, being overweight was inversely correlated with cognitive impairments in the groups of rural–dwelling participants aged 65–69 years (OR, 0.73; 95% CI: 0.55–0.96; p < 0.05) and 75–79 years (OR, 0.63; 95% CI: 0.42–0.93; p < 0.05). However, the results for the urban–dwelling participants were the opposite. In models 4 and 5, obesity and being overweight were associated with an increased risk of cognitive impairment in urban–dwelling participants aged 65–69 years, and higher BMI was associated with higher risk of cognitive impairment. We examined whether the urban/rural differences in the relationship between BMI categories and cognitive impairment were statistically significant using the logistic regression analyses, and the results showed that the difference in urban/rural areas and cognitive impairment was statistically significant (p < 0.001), as well as in BMI categories and cognitive impairment (p = 0.016). The interaction between BMI categories and urban/rural areas was statistically significant (p = 0.049).
We further examined whether being underweight for individuals aged 80 years and above was a cause of impairment rather than a consequence. The result indicated that the association between underweight and cognitive impairment remained robust (OR, 1.73; 95% CI: 1.06–2.84; p < 0.05), and a stronger relationship of underweight with dementia (OR, 1.90; 95% CI: 1.04–3.50; p < 0.05) than with MCI (OR, 1.66; 95% CI: 1.02–2.70; p < 0.05). We also examined whether obesity and cognitive impairment had a reverse causal relationship by using the same adjusted model for MCI and dementia in urban old adults aged 65–69 years and found that being obesity was associated with MCI (OR, 1.37; 95% CI: 1.07–1.77; p < 0.05), but not significantly associated with dementia (OR, 1.22; 95% CI: 0.49–3.00; p = 0.672). Therefore, the association of obesity, overweight, and underweight with cognitive impairment were age and area dependent.
DISCUSSION
In this study, we evaluated the associations between BMI and cognitive impairment among older adults living in rural and urban areas. To our knowledge, this is the first comprehensive subnational assessment of the association between BMI and cognitive impairment in Central China to include data on BMI, cognitive diagnoses, and combined lifestyle factors. We demonstrated that cognitive impairment among community–dwelling older adults is affected by age, BMI, and a broad spectrum of time–varying modifiable factors. We saw a relationship between obesity and increased cognitive impairment in the 65–69 sample of urban residents. However, being underweight was associated with a more favorable dementia risk profile in the 80 + sample of rural residents, whereas being overweight was inversely associated with cognitive impairment. The associations were robust to adjustment for a wide array of potentially confounding variables. Several sensitivity analyses did not support the alternate hypothesis of reverse causation, including separating the MCI and dementia, as well as examining differences in the relationship between BMI and cognitive impairment between urban and rural areas. Some of other studies also tested for and found little evidence of reverse causation [4, 22]. This study revealed the differences in the associations between BMI and cognitive impairment among older residents of rural and urban areas. These findings can guide interventions, particularly in rural regions where being underweight is typically more prevalent than in urban regions. Rural community managers can strengthen health management by grouping the weight of older people to prevent the risk of dementia. The HMACS data showed that the prevalence of cognitive impairment among rural dwellers is significantly higher than that among urban dwellers (31.3% versus 26.3%). This finding corroborates those of previous studies in China that suggest that the prevalence rate of dementia is notably higher in rural areas than in urban areas [23]. Although a few epidemiologic studies are available for the estimation of the prevalence of obesity among different ethnic groups in China, the data are from different studies conducted at different times [24–26]. Consistent survey data in terms of BMI categories in old adults are still scarce. This study is the first to provide more recent estimates of the prevalence of obesity and being overweight in Central China. We found that the prevalence of being underweight in rural areas is much higher than that in urban areas (6.7% versus 3.6%). This suggests that undernutrition is currently a serious public health problem and occurs frequently among the rural–dwelling older adults in Central China.
Previous prospective cohort studies on US population imply that maintaining normal weight across adulthood, especially preventing weight gain in early adulthood, is important for preventing premature deaths in later life [27, 28]. To investigate the independent associations between BMI and cognitive impairment, models were adjusted for socioeconomic factors, lifestyle habits, and comorbidities. We replicated the obesity–cognitive impairment findings in the groups of participants aged 65–74 years, as proposed by The Lancet Standing Commission on Dementia Prevention, Intervention, and Care in a life–course model of dementia prevention [29]. Based on this group’s age (65–74 years), the development of obesity likely occurred during the midlife years.
We also evaluated the relationship between physical activity and cognitive impairment. The rate of physical activity was higher among all the participants and the urban–dwelling participants in the normal cognition group than among those in the cognitive impairment group. This finding can further support the implementation of preventive strategies that are focused on increasing physical activity and improving education in urban–dwelling older adults.
The finding of the present study regarding the relationship between body weight and cognitive impairment in rural areas in China is different from those of most previous studies that demonstrated that high middle–age BMI or low late–age BMI is a predictor for the development of dementia [1, 30]. We conducted a series of sensitivity analyses by separating out MCI versus dementia to examine an association between underweight and risk of dementia or cognitive impairment. The results supported the argument that underweight was more likely to have dementia. It is worth noting that the proportion of obese people older than 80 years was not very high in our sample (8.3%, only 87 persons). Thus, we need to validate this finding in bigger samples.
The urban–rural disparities of other potential lifestyle risk factors and comorbidities were investigated in this study. There were significant differences between the age and education levels of all the participants, the rural–dwelling participants, and the urban–dwelling participants in the normal cognitive and cognitive impairment groups. The difference between the prevalence rates of diabetes for all the participants, the rural–dwelling, and the dwelling participants in the normal and cognitive impairment groups were statistically significant. The difference between the prevalence rate of hypertension for all the participants in the normal cognition and cognitive impairment groups were also statistically significant. Sensitivity analyses support this difference. Taken together, the participants in this study were heterogeneous with respect to education and socio–economic factors and could be considered representative of Central China.
This study had certain limitations. First, this was a cross–sectional observational study and thus cannot provide information on a causal relationship. Second, the results cannot be generalized to the whole population in China because the sample population was selected from one province. Third, the prevalence of underweight found by our study may be lower than the actual prevalence because we did not include older adults who were institutionalized. While studies have reported that undernutrition occurs frequently among the hospitalized older adults [31]. Finally, dietary factors related to BMI were not analyzed in this study, and this may have influenced the results. Although this analysis attempted to account for reverse causation and confounding, due to its observational study design, these two alternate scenarios cannot be ruled out completely.
CONCLUSIONS
We conducted a cross–sectional study of data from the HMACS to investigate the association between cognitive impairment and different BMI categories in older adults. The results demonstrated urban–rural disparities in the association between BMI and cognitive impairment in older adults, suggesting that a large proportion of rural–dwelling older adults need nutritional intervention. These findings indicate the need for interventions in rural/urban populations at the highest risk of cognitive impairment.
Footnotes
ACKNOWLEDGMENTS
We thank all the study participants and all the graduate students for their participation. We also thank the doctors, nurses, and clinical supervisors for their contributions, and the field coordinators at the Qinglinjie, Gangduhuayuan, and Liyuan community health centers and Dawu Chinese traditional medicine hospital.
Financial support for the present study was received from National Natural Science Foundation of China (No. 81870901 and 82071272 to Y.Z., No. 81771488 to H.X.D., No. 71774127 to D. L) and Ministry of Science and Technology of China (No. 2020YFC2006000 to Y.Z.). The financial contributors had no role in the design, analysis, or writing of this article.
DATA AVAILABILITY STATEMENT
The data collected for the study, including individual participant data, a data dictionary defining each field in the set, and related documents will be available (e.g., study protocol, statistical analysis plan, informed consent form) will be made available to others according to MDPI Research Data Policies.