Association between social connectedness and mild cognitive impairment: A case-control study in aging rural Indians

Abstract

There is a scarcity of evidence on the association between social connectedness and cognitive impairment in the rural Indian population. This cross-sectional study included rural Indians aged 45+ years (n = 5805) from an ongoing aging cohort, the CBR-SANSCOG study. Based on Clinical Dementia Rating (CDR), participants were classified into cognitively normal (CN, CDR = 0) and mild cognitive impairment (MCI, CDR = 0.5). Social networks, assessed across three dimensions (network diversity, size, and embeddedness), were compared between the two groups, adjusting for potential confounders. Individuals with MCI had significantly lower mean scores across all three social network dimensions than CN individuals.

Introduction

Prior studies have demonstrated that poor social connectedness is associated with a higher risk of cognitive decline and dementia.^1,2 On the other hand, persons with robust social networks have been shown to have better cognitive functioning.^3,4 Nevertheless, such evidence predominantly stems from studies conducted in high-income countries or among Western populations.

It is essential to consider that the nature and extent of social relationships are primarily culture-dependent and that specific cultural values or communication styles could play a substantial role in shaping social relationships. Therefore, the effect of social connectedness on cognitive health may vary from population to population.

However, studies examining the association between social connectedness and cognitive impairment in the Indian population are scarce. Further, though over two-thirds of Indians are rural-dwelling (the rural population of India itself is nearly three times the entire US population), there are hardly any such studies conducted among rural Indians.

Mild cognitive impairment (MCI) is an intermediate stage between normal cognitive aging and dementia, characterized by noticeable cognitive deficits (both subjective and objective), not severe enough to significantly impair daily functioning. ⁵ Though individuals with MCI have a heightened risk for dementia, not all progress to developing dementia. Thus, understanding MCI and the factors associated with it may be crucial for potentially introducing preventive strategies to prevent further cognitive decline.

Social connectedness is a complex concept that encompasses quantitative and qualitative aspects. The quantitative aspect can be assessed through three dimensions of social networks, namely network diversity, network size, and network embeddedness. ⁶ Network diversity refers to the variety in the active social roles; network size refers to the total number of active social contacts, whereas network embeddedness refers to how closely knit the social networks are. So, it is crucial to examine which of these social network dimensions differ between the MCI and cognitively normal groups.

This study aimed to cross-sectionally examine social networks in the above three dimensions, using the Cohen's Social Network Index (SNI) among rural Indians with MCI compared to their cognitively normal counterparts. We hypothesized that individuals with MCI have lower scores in all three network dimensions than cognitively normal individuals.

Methods

This was a cross-sectional (case-control) study wherein baseline clinical assessment data was obtained from an ongoing large community-based, prospective aging cohort from rural India, namely the Centre for Brain Research – Srinivaspura Aging NeuroSenescence and COGnition (CBR-SANSCOG) study. The CBR-SANSCOG cohort recruits non-demented individuals aged 45+ years, from a predominantly agricultural community in the villages of Srinivaspura in Kolar district, Karnataka, southern India. According to the latest Census of India (2011), Karnataka compares with India in terms of average life expectancy: 69 years versus 67 years, proportion of males: 51% versus 51%, proportion of rural population: 61% versus 69%, and literacy rate: 67% versus 63%. In the CBR-SANCSOG cohort, individuals with psychosis, bipolar disorder, substance dependence (excluding nicotine), severe/ terminal medical illnesses, and severe hearing or vision impairments that would limit study assessments are excluded. Full details on the CBR-SANSCOG study recruitment and study protocol have been published as a separate manuscript. ⁷ A comparison of the demographic and clinical characteristics between the CBR-SANSCOG cohort and the rural Indian population (45+ years) from the nationally representative Longitudinal Study of Aging in India (LASI, Wave 1, 2017–18) is presented in Supplemental Table 1.

The present study's sample included 5805 CBR-SANSCOG cohort participants (2766 males, 3039 females) recruited between January 2018 and December 2023, with complete data on the sociodemographic, clinical, and cognitive variables. The cases were participants with MCI (n = 942; 320 males, 622 females), and the controls were cognitively normal (CN) participants (n = 4863; 2446 males, 2417 females).

Ethics and consent

The CBR-SANSCOG study has obtained ethical clearance from the Institutional Human Ethics Committee of the Centre for Brain Research. All participants provided voluntary, written informed consent for the study, including specific consent to undergo the clinical and cognitive assessments.

Assessments

• (i) Sociodemographic details: Information such as age, sex, education, occupation, and annual income was collected using a sociodemographic proforma.

• (ii) Clinical Dementia Rating (CDR): The CDR instrument ⁸ was used to diagnose MCI. CDR is an extensively validated scale that assesses six cognitive and functional domains: memory, orientation, judgment and problem solving, community affairs, home and hobbies, and personal care. The total CDR score is calculated using a predetermined algorithm, and a score of ‘0.5’ is categorized as MCI, whereas a score of ‘0’ is interpreted as cognitively normal (CDR ≥ 1 indicates a diagnosis of dementia; in this study, these individuals were excluded)

• (iii) Social Networks: The Cohen's Social Network Index (SNI) ⁹ was used to assess social networks, wherein three distinct dimensions are evaluated: (a) Network diversity, (b) Network size, and (c) Network embeddedness. The SNI considers twelve possible social relationships (spouse, parents, in-laws, children, other relatives, neighbors, friends, work colleagues, school/college-mates, volunteer groups, religious groups, and non-religious social groups) and eight possible social networks (family, neighbors, friends, work colleagues, school/college-mates, volunteer groups, religious groups, and non-religious social groups). In any relationship, a person is a “regular contact” if the frequency of contact is at least once in two weeks. The scoring is done separately for each of the three dimensions as follows:

  1. Network diversity: Number of relationships where the individual has regular contact with at least one person (one point for each of the twelve relationships listed above)
  2. Network size: Total number of persons in all the twelve possible relationships with whom the individual has regular contact.
  3. Network embeddedness: Number of networks (one point for each of the eight possible networks listed above) where the individual is actively engaged (at least four regular contact persons should be within each network).

• (iv) Covariates: Age, sex, education, marital status, depression, vitamin B12 deficiency, vitamin D deficiency, frailty, vision problems, hearing problems, and history of stroke were used as covariates.

Statistical analysis

Statistical analyses were conducted using SPSS 25.0. Armonk, NY: IBM Corp. ¹⁰ Continuous variables, such as age and education, were presented as means with standard deviations (Mean ± SD), whereas categorical variables, such as sex, marital status, depression, vitamin B12 deficiency, vitamin D deficiency, frailty, vision problems, hearing problems, and history of stroke were presented as counts and percentages (n, %). The independent samples t-test was used to compare the continuous variables, whereas the Chi-Square test was used to compare the categorical variables between the MCI and CN groups. Further, multivariate analysis of covariance (MANCOVA) was used to investigate differences in the three social network dimensions between MCI and CN individuals, adjusting for the same set of covariates. A sex-stratified analysis was also performed. A p-value of <0.05 was considered statistically significant.

Results

We found that participants with MCI were significantly older and had higher proportions of frailty, hearing problems, and a history of stroke compared to the CN group. We found no significant differences between the two groups for the other sociodemographic and clinical variables (Table 1).

Table 1.

Comparison of demographic and clinical characteristics between mild cognitive impairment (MCI) and cognitively normal (CN) groups.

Characteristics	Groups		p
	Cases: MCI (n = 942)	Controls: CN (n = 4863)
Age, y (Mean ± SD)	63.19 ± 9.38	58.38 ± 9.57	0.001*
Sex
Male, n (%)	320 (33.97)	2446 (50.29)	0.65
Female, n (%)	622 (66.03)	2417 (49.71)
Education, y (Mean ± SD)	3.52 ± 3.78	4.41 ± 3.99	0.25
Marital status
Living with a partner, n (%)	682 (72.40)	3987 (81.98)	0.82
Not living with a partner, n (%)	260 (27.60)	876 (18.02)
Depression
Normal, n (%)	687 (72.92)	4133 (84.98)	0.25
Depressed, n (%)	255 (27.08)	730 (15.02)
Vitamin B12 level
Normal, n (%)	602 (63.90)	2917 (59.98)	0.75
Deficient*, n (%)	340 (36.10)	1946 (40.02)
Vitamin D level
Normal, n (%)	637 (67.62)	3384 (69.58)	0.25
Deficient@, n (%)	305 (32.38)	1479 (30.42)
Frailty
Not frail, n (%)	766 (81.31)	4473 (91.98)	0.04*
Frail$, n (%)	176 (18.69)	390 (8.02)
Vision problems
No, n (%)	588 (62.43)	2915 (59.94)	0.07
Yes, n (%)	354 (37.57)	1948 (40.06)
Hearing problems
No, n (%)	881 (93.53)	4665 (95.94)	0.03*
Yes, n (%)	61 (6.47)	198 (4.06)
History of stroke
No, n (%)	924 (98.14)	4821 (99.13)	0.03*
Yes, n (%)	18 (1.91)	42 (0.87)

^@ <200 pg/mL; ^#<20 ng/mL; ^$Fried's frailty phenotype.

The independent samples t-test was used to compare continuous variables such as age and education, whereas the Chi-Square test was used to compare categorical variables such as sex, marital status, depression, vitamin B12 deficiency, vitamin D deficiency, frailty, vision problems, hearing problems, and history of stroke.

*Indicates a p-value of <0.05.

The mean scores of each of the three social network dimensions were significantly lower in the MCI than the CN group: (i) Network diversity (4.78 ± 1.67 versus 5.57 ± 1.67, p < 0.001), (ii) Network size: (20.32 ± 9.36 versus 21.16 ± 8.63, p = 0.04), and (iii) Network embeddedness (1.47 ± 1.20 versus 1.76 ± 1.23, p = 0.03) as shown in Figure 1.

Figure 1.

This figure compares the mean scores across the three dimensions of the social network Index (SNI) between individuals with mild cognitive impairment (MCI, n = 942) and cognitively normal (CN, n = 4863) individuals. ***p < 0.001, *p < 0.05.

After adjusting for the covariates, we found a significant association between all three social network dimensions and cognitive status (Table 2). The sex-stratified analysis showed a significant difference between the MCI and CN groups in the network size dimension among males whereas network diversity differed significantly between the MCI and CN groups among females (Table 2). There were no differences in network embeddedness between the two groups among males and females.

Table 2.

Multivariate analysis of covariance for the three social network dimensions among mild cognitive impairment (MCI) and cognitively normal (CN) individuals – overall and sex-stratified.

	Model 1 a					Model 2 b
Social network dimensions	F	95% CI		p	η²	F	95% CI		p	η²
Overall
Network Diversity	60.22	0.38	0.64	0.001*	0.01	5.78	−0.35	−0.04	0.02*	0.02
Network Size	5.62	0.14	1.52	0.02*	0.10	4.67	−1.88	−0.09	0.03*	0.03
Network Embeddedness	1.92	−0.03	0.17	0.17	0.003	8.12	−0.32	−0.06	0.004*	0.004
Male c
Network Diversity	19.91	0.22	0.56	0.001*	0.001	3.71	0.01	0.04	0.05	0.05
Network Size	1.07	−0.43	1.38	0.30	0.30	7.04	−0.38	−0.06	0.01*	0.01
Network Embeddedness	0.82	−0.19	0.07	0.37	0.37	1.42	−1.80	0.44	0.23	0.23
Female c
Network Diversity	28.99	0.35	0.75	0.001*	0.03	1.72	0.04	0.09	0.04*	0.04
Network Size	5.67	0.24	2.42	0.02*	0.06	3.12	−2.80	0.15	0.08	0.08
Network Embeddedness	7.93	0.07	0.39	0.01*	0.08	1.36	−0.34	0.09	0.24	0.24

^a Unadjusted model.

^b Adjusted for age, sex, education, marital status, depression, vitamin B12 deficiency, vitamin D deficiency, frailty, vision problems, hearing problems, and history of stroke.

^c For the sex-stratified analysis, Model 2 was adjusted for all the above covariates except sex.

*Indicates a p-value of <0.05.

Discussion

Our study is one of the first large-scale studies comparing social networks between individuals with MCI and CN individuals from an aging, rural Indian population. Our findings reveal that the MCI group had poorer network size, diversity, and embeddedness than the CN group after controlling for age, sex, education, marital status, depression, vitamin B12 and vitamin D deficiency, frailty, vision and hearing problems, and history of stroke.

Our findings are in line with that from a similar aging cohort study in the United Kingdom, the Cognitive Function and Ageing Study Wales (CFAS Wales), where older adults (>65 years) with MCI had lesser social networks than the CN individuals. ¹¹ Another large national study in the United States, which serially followed up older adults aged 65+ years, found that better social networks and specific roles in the network decreased the risk of developing MCI as well as conversion of MCI to dementia. ¹² A recent study based on data from over thirty-two thousand individuals aged ≥50 years from six low- and middle-income countries (LMICs), including India, revealed that social participation was associated with lower odds of having MCI. ¹³ Thus, our results add strength to prior studies demonstrating that social connectedness is negatively associated with MCI.

Our study focused on two unique aspects. One is that we examined the different dimensions of social relationships separately since they could have differential relationships with cognitive impairment. However, our findings revealed that all three social network dimensions were associated with MCI. Secondly, the sex-specific analysis revealed that men with MCI had smaller network sizes than CN men, whereas women with MCI had lesser network diversity than CN women. Such nuances could be crucial when developing targeted social interventions for individuals with MCI. ¹⁴

There could be several putative mechanisms for the association between poorer social connectedness and cognitive impairment. Social participation could enhance cognitive reserve and resilience to brain insults. ¹⁵ A recent study revealed that network diversity was associated with better cognitive performance and seemed to attenuate the adverse cognitive effects on amygdalar volumes on brain MRI. ¹⁶ Large and diverse social networks have been shown to have a protective effect against depression, ¹⁷ which, in turn, is a known risk factor for dementia. Better social connectedness is likely to promote healthier lifestyles and positive health-related behaviors.

The strengths of our study include a large and relatively homogenous sample size and the use of well-validated tools for assessing cognitive status and social networks, namely CDR and Cohen's SNI, respectively. Additionally, our study's results from this rural community hold significant relevance for India, where though two-thirds of its population is rural-dwelling, the current urbanization trends, the rapidly expanding digital connectivity, and the ongoing cultural changes are expected to impact the rural lifestyle and social relationships.

Our study's main limitation is the cross-sectional design, so it is not possible to conclude that decreased social connectedness is a risk or predictive factor for MCI. Moreover, reverse causality is also possible, wherein individuals with MCI may find it more challenging to maintain social connections due to their cognitive or functional problems. Further, the self-reported measure of social networks is prone to recall bias, particularly among individuals with MCI due to their cognitive impairment. The area sampling technique utilized in the study is another limitation.

This potential bidirectional relationship between social connectedness and cognitive health makes it vital to follow up these study participants with serial cognitive monitoring to determine if individuals with MCI and having poorer social networks are at greater risk of progressing to dementia. Nevertheless, it is essential to promote awareness among older persons on the importance of healthy social relationships for better cognitive health and quality of life.

Future research could explore the effectiveness of including social interventions as a significant component in the management of individuals with MCI to ascertain whether such interventions could delay or even prevent the onset of dementia. However, for such social intervention strategies to be feasible in rural communities of resource-limited countries such as India, they should be culturally tailored, cost-effective, and easily scalable to the community level. These could involve creating more opportunities for social interaction through community centers, public parks, support groups, etc., and improving the built environment to promote mobility and connectivity.