Abstract
Background
Social stimulation from having large networks may build cognitive reserve and contribute to cognitive resilience. However, limited research has assessed whether this relationship differs across the adult life course. We examined the relationship between social network size and resilience to neurodegeneration in memory, language, and executive function among mid-to-late-life adults.
Methods
The sample included 1,602 participants from the Framingham Heart Study Offspring cohort who completed the Berkman-Syme Social Network Index (SNI) questionnaire, had available plasma total tau (t-tau) measures at Exam 8, and had neuropsychological (NP) factor scores in memory, language, and executive function. Social network size was operationalized through a calculated SNI score and categorized into small, medium, and large networks. Cognitive resilience was operationalized using a residual approach by regressing each NP factor score on the plasma t-tau measure. Linear regression was used to assess associations between social network size and cognitive resilience, adjusting for age, sex, education, APOE ε4 allele carrier status, time between the most recent NP exam visit date and Exam 8 date, physical activity, smoking status, and depressive symptoms. Models were stratified by midlife (age 47–64 years) and late-life (age 65–91 years) groups.
Results
Average age was 67 years with most participants identifying as female (n = 880; 55%). By age group, more participants were in the late-life group (n = 918; 57%) compared to the midlife group (n = 684; 43%). Large social network size was significantly associated with greater resilience to neurodegeneration in memory compared to small social network size (β = 0.07; 95% CI = 0.01–0.13). There were no significant associations between large social networks and resilience to neurodegeneration in language or executive function. No significant interactions between social network size and age group on resilience to neurodegeneration were observed in any of the domains examined. In the stratified models, large social networks were only associated with resilience to neurodegeneration in executive function among the midlife group (β = 0.11; 95% CI = 0.01–0.21).
Conclusions
Large social networks may promote greater resilience to neurodegeneration in memory among mid-to-late-life adults. Specific to middle-aged adults was a significantly greater resilience in executive function associated with large networks. These findings should be interpreted with caution, however, as estimate sizes are relatively small and potentially not robust, given the current sample size. Additional studies in larger and more diverse samples are needed to replicate and generalize our findings to other groups.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12883-026-05080-4.
Keywords: Social networks, Cognition, Neurodegeneration, Resilience
Background
In the United States, approximately seven million older adults currently live with clinical Alzheimer’s disease and related dementias (ADRD) [1], which is characterized by decline in cognitive function that impedes the ability to carry out activities of daily life [2]. This number is expected to grow to 14 million by 2060 [1]. Given the rising burden of ADRD and an aging population, it is important to identify factors that allow individuals to be resilient against disease pathology or neurodegeneration to delay the onset of clinical dementia [3]. Identifying modifiable factors, such as social engagement and connectedness, that change across the life course can allow for more precise and targeted interventions to prevent ADRD [4, 5].
Cognitive reserve is the ability for cognitive processes to adapt to brain changes or damages, and can help explain differences in cognition and function between individuals, whereas brain reserve is the structural attributes of the brain (e.g., number of neurons, synapses), and can explain differences in cognitive ability relative to pathology level [6]. When individuals exhibit cognitive resilience, which is the ability to maintain cognitive function despite brain atrophy and disease pathology progression, it may be due to cognitive reserve, brain reserve, or both [6, 7]. Cognitive resilience may act to prevent or delay onset of ADRD, and understanding the factors that enhance cognitive resilience, such as social connectedness, is crucial for developing and promoting interventions that can reduce the incidence of ADRD.
Systematic reviews and meta-analyses found positive associations between social network size and cognitive function [8, 9]. Moreover, greater social network size has been observed to be associated with decreased dementia risk [10, 11], yet one systematic review and meta-analysis found no association between social network size and incident dementia [12]. Social network size may be a factor that promotes cognitive resilience through increasing social stimulation, which in turn enhances cognitive reserve [13, 14]. Sharifian et al. found that social network size moderates the association between cortical thickness and language and executive function [15].
Current research relating to social networks and cognitive resilience, however, is limited to findings based on clinical samples, which are not as generalizable to the broader population [16, 18]. In addition, previous studies focused on older adults and measured social networks in late life, which means that evidence to show whether the impact of social networks differs across the life course, particularly during midlife, is limited. Understanding whether the impact differs across the life course may inform when it is most critical to intervene to prevent ADRD. The current study addresses these limitations by using a community-based sample from the Framingham Heart Study (FHS) and included information on social networks from mid-to-late-life adults to examine whether larger social networks are associated with greater cognitive resilience. The objective of the study was to assess the relationship between social network size among middle- and older-aged adults and resilience to neurodegeneration in multiple cognitive domains, including memory, language, and executive function.
Methods
Study setting and sample
FHS is a longitudinal cohort of participants from Framingham, Massachusetts, with regular health exams conducted since 1948 [17]. The present study used data from the FHS Offspring cohort, which began assessments between 1971 and 1975 with 5,124 participants who were 5–70 years of age and had, on average, health exams every four to eight years [18]. The current study sample included participants who completed the Berkman-Syme Social Network Index (SNI) questionnaire at Exam 8 (2005–2008) and a neuropsychological (NP) exam visit within five years after their Exam 8 visit. All participants provided written informed consent. The institutional review board at the Boston University Medical Center approved the study protocols and consent forms. Clinical trial number: not applicable.
Social network size and social support constructs
The self-reported Berkman-Syme SNI questionnaire (Supplemental Fig. 1), which was administered at Exam 8, measures social network size, and degree and type of social support [19]. Similar to previous studies [11, 19, 20], the Berkman-Syme SNI score was calculated using four questions from the instrument that indicate the degree and type of social contact: number of (1) close friends and (2) relatives (with answer options as “None,” “1 or 2,” “3–5,” “6–9,” or “10 or more”, scored as ≤ 2 friends and ≤ 2 relatives = 0 and all other scores = 1); (3) participation in religious meetings or services (less than or equal to every few months = 0; greater than or equal to once or twice a month = 1); or (4) group participation in other community organizations (no = 0; yes = 1). The scores were summed together and ranged between 0 and 4. We created three categories of scores: low (SNI score = 0 or 1), moderate (SNI score = 2), and high (SNI score = 3 or 4) that reflected small, medium, and large social networks, respectively. Small social network size (e.g., low SNI score) served as the reference group.
Quantification of plasma total tau level
Current research has identified plasma total tau (t-tau) as a biomarker that represents neurodegeneration in AD research frameworks [21]. Plasma t-tau was found to be associated with endophenotypes of dementia and incident clinical AD dementia risk in the FHS sample [22]. Fasting blood samples were obtained at Exam 8. Samples were immediately centrifuged, aliquoted, and stored at -80 degrees Celsius. Plasma samples were analyzed from February to March 2017 using a Simoa Tau 2.0 Kit and a Simoa HD-1 analyzer (Quanterix). This assay is a single-molecule ELISA (digital ELISA) validated as fit-for-purpose research use only, with the limit of detection as 0.019 pg/mL. The assay uses a set of monoclonal antibodies reacting to both normal and phosphorylated tau and can detect all tau isoforms. Samples sent for analysis had never been thawed. The analytical range was between 0.06 and 360 pg/mL. The intra-assay coefficient of variation was 4.1%, whereas the inter-assay coefficient of variation was 7.5%. Additional details on the full assay methods have been previously published [22].
Neuropsychological factor scores
FHS participants completed multiple cognitive assessments, including screening instruments and more comprehensive NP batteries at various study visits. Factor scores were previously derived to calibrate cognitive assessment data across multiple waves in FHS. Experts assigned each cognitive test item to a single cognitive domain to address study design considerations, ceiling effects, and measurement precision. As part of a larger cross-study harmonization effort [23], each domain was calibrated separately using bi-factor confirmatory factor analysis models, which incorporated fixed item parameters for anchor items previously calibrated from other studies and freely estimated item parameters from FHS-specific items. A description of the methods used to develop the harmonized and calibrated cognitive data from FHS, including the scores and scales, has been published [24]. Factor scores for memory, language, and executive function were used in this analysis.
APOE genotyping
APOE genotypes were determined from blood DNA samples through the process previously described by Hixson and Vernier, or with a standard TaqMan assay [25, 26]. Participants were coded as ε4 carriers (e.g., ε2/ε4, ε3/ε4, and ε4/ε4) or non-carriers.
Statistical analysis
To show distributions of baseline participant characteristics, descriptive statistics were used, with mean and standard deviation (SD) for continuous variables, or count and percentage for categorical variables.
Cognitive resilience was operationalized as the discordance between predicted cognition and neurodegeneration using plasma t-tau. To quantify resilience as a continuous measure, we entered NP factor scores, which were standardized first, as a dependent variable and plasma t-tau as an independent variable in a linear regression model. This was repeated for each NP factor score (memory, language, executive function) to create residual scores based on each NP factor score. Because pathology and cognition are continuous variables, this approach minimizes the need for artificial dichotomization, and keeps continuous variables continuous [27]. The residual, calculated as 
, measures the difference between an observed value ( y) and the predicted value (
) from the linear regression model. The calculated residual value serves as a measure of the discordance between pathology and expected cognition based on the central regression line, where, positive scores on the residual indicate better cognitive function than expected based on the level of pathology, and negative scores indicate worse cognitive function than expected. Key assumptions of the residual-based approach include linearity, independence of errors, homoscedasticity, normality of errors, and absence of unmeasured confounding. The residual-based approach also assumes a direct and stable relationship between plasma t-tau and cognition, which may not fully capture underlying pathology, and any residual variance may reflect unmeasured factors, such as vascular burden, stress or depression, and other neuropsychological conditions and pathologies, rather than pure resilience. Finally, it is important to acknowledge that the residuals represent statistical resilience relative to plasma t-tau, rather than a comprehensive measure of neurodegeneration, since plasma t-tau may not be as sensitive compared with other biomarkers, such as phosphorylated tau (p-tau) 217.
After the residuals were calculated, we conducted three separate multiple linear regressions to assess associations between social network size and resilience to neurodegeneration in memory, language, and executive function, respectively, adjusting for age (years), sex (female versus male), education (college graduate versus non-graduate), APOE ε4 allele carrier status (none versus at least one ε4 allele), time between the most recent NP exam visit date and date of the eighth health exam, physical activity level (physical activity index score), smoking status (former or current smoker versus non-smoker), and depressive symptoms (≥ 16 on the Center for Epidemiologic Studies Depression Scale). Scores and scales for the physical activity index and the Center for Epidemiologic Studies Depression Scale have been previously published [28, 29]. All models were stratified by baseline age groups (midlife age < 65 years and late-life age ≥ 65 years), sex (female and male), and APOE ε4 allele carrier status (at least one ε4 allele and no ε4 allele). Statistical interactions between social network size and age group, social network size and sex, as well as social network size and APOE ε4 allele carrier status, were also assessed. Statistical significance was determined by an alpha level of 0.05. No correction for multiple testing was applied. All statistical tests were two-sided, and analyses were performed using SAS™ software version 9.4 [30].
Results
The sample included 1,602 participants, with an average age of 67.24 years (SD = 8.29) and most participants identifying as female (54.93%, n = 880) and not being an APOE ε4 allele carrier (79.86%, n = 1249) (Table 1). Approximately half of the sample reported being a college graduate (50.69%, n = 812). By age group, most participants were in the late-life age group (≥ 65 years; 57.30%, n = 918) compared to the midlife age group (< 65 years; 42.70%, n = 684). Average number of days between the Exam 8 date and NP exam date in the sample was 1448.68 (SD = 1577.35). Differences in the average age of the sample, and proportion of individuals in each age group (e.g., midlife, late-life), education categories, and self-reported depressive symptoms and smoking status, were observed across each social network size group.
Table 1.
Participant sample characteristics in the Framingham Heart Study based on social network size measured at Exam 8 (2005–2008)
| Overall (n = 1602) |
Small social network size (n = 355) |
Medium social network size (n = 475) |
High social network size (n = 772) |
p-value | |
|---|---|---|---|---|---|
| Age (years), mean (SD) | 67.24 (8.29) | 65.14 (7.44) | 66.85 (8.37) | 68.44 (8.40) | < 0.01 |
| Age group, n (%) | < 0.01 | ||||
| Late-life (≥ 65 years) | 918 (57.30) | 165 (46.48) | 259 (54.53) | 494 (63.99) | |
| Midlife (< 65 years) | 684 (42.70) | 190 (53.52) | 216 (45.47) | 278 (36.01) | |
| Sex, n (%) | < 0.01 | ||||
| Male | 722 (45.07) | 190 (53.52) | 241 (50.74) | 291 (37.69) | |
| Female | 880 (54.93) | 165 (46.48) | 234 (49.26) | 681 (62.31) | |
| Education, n (%) | 0.02 | ||||
| High school, did not graduate | 38 (2.37) | 7 (1.97) | 19 (4.00) | 12 (1.55) | |
| High school graduate | 348 (21.72) | 87 (24.51) | 112 (23.58) | 149 (19.30) | |
| Some college | 404 (25.22) | 82 (23.10) | 124 (26.11) | 198 (24.65) | |
| College graduate | 812 (50.69) | 179 (50.42) | 220 (46.32) | 413 (53.50) | |
| APOE ε4 carrier status, n (%) | 0.78 | ||||
| Yes | 315 (20.14) | 67 (19.48) | 91 (19.44) | 157 (20.88) | |
| No | 1249 (79.86) | 277 (80.52) | 377 (80.56) | 595 (79.12) | |
| Depressive symptomsa, n (%) | < 0.01 | ||||
| Yes | 183 (12.01) | 61 (18.60) | 57 (12.61) | 65 (8.74) | |
| No | 1341 (87.99) | 267 (81.40) | 395 (87.39) | 679 (91.26) | |
| Physical activity index, mean (SD) | 35.26 (5.28) | 34.91 (5.74) | 35.10 (5.20) | 35.51 (5.10) | 0.16 |
| Smoking status, n (%) | < 0.01 | ||||
| Non-smoker | 921 (57.53) | 166 (46.76) | 259 (54.64) | 496 (64.25) | |
| Current smoker | 110 (6.87) | 40 (11.27) | 41 (8.65) | 29 (3.76) | |
| Ex-smoker | 570 (35.60) | 149 (41.97) | 174 (36.71) | 247 (31.99) | |
| Plasma total tau (pg/mL), mean (SD) | 4.07 (1.47) | 3.96 (1.54) | 4.05 (1.32) | 4.13 (1.52) | 0.16 |
| Time difference between Exam 8 and NP exam (days), mean (SD) | 1448.68 (1577.35) | 1373.46 (1561.33) | 1485.64 (1596.73) | 1460.53 (1573.59) | 0.57 |
| Memory factor score, mean (SD) | 0.66 (0.41) | 0.66 (0.42) | 0.65 (0.42) | 0.67 (0.40) | 0.82 |
| Language factor score, mean (SD) | 0.36 (0.45) | 0.37 (0.48) | 0.35 (0.42) | 0.36 (0.45) | 0.80 |
| Executive function factor score, mean (SD) | 0.30 (0.52) | 0.24 (0.58) | 0.30 (0.51) | 0.32 (0.50) | 0.07 |
SD Standard deviation, APOE Apolipoprotein E, pg/mL picogram/milliliter, NP Neuropsychological
aDepressive symptoms was defined as a score of 16 or greater on the Center for Epidemiologic Studies Depression scale
Overall, large social network size was significantly associated with greater resilience to neurodegeneration in memory compared to small social network size (β = 0.07; 95% CI = 0.01–0.13; p = 0.03) when adjusted for age, sex, APOE ε4 allele carrier status, education, time between exam date where plasma t-tau was measured and NP exam date, physical activity, depressive symptoms, and smoking status (Table 2). As a reference to better contextualize the magnitude of these observed associations, the magnitude of the association between educational attainment, which is a well-known contributor of cognitive resilience, and each cognitive domain examined was around 0.15 in the overall sample. There were no significant associations between large social network size and resilience to neurodegeneration in language (β = 0.03; 95% CI = -0.04–0.10; p = 0.39) or executive function (β = 0.07; 95% CI = -0.01–0.14; p = 0.07) in the adjusted models. There were also no significant associations between medium social network size and resilience to neurodegeneration in any of the domains (memory, language, and executive function) compared to small social network size.
Table 2.
Associations between social network size and resilience to neurodegeneration in memory, language, and executive function
| Memory | Language | Executive Function | ||||
|---|---|---|---|---|---|---|
| β (95% CI) | P | β (95% CI) | P | β (95% CI) | P | |
| Social Network Size | ||||||
| Small | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Medium | 0.02 (-0.05–0.09) | 0.59 | -0.01 (-0.09–0.04) | 0.74 | 0.02 (-0.06–0.09) | 0.66 |
| Large | 0.07 (0.01–0.13) | 0.03 | 0.03 (-0.04–0.10) | 0.39 | 0.07 (-0.01–0.14) | 0.07 |
CI Confidence interval
All models adjusted for age, sex, APOE, time between exam date where plasma t-tau was measured and NP exam date, physical activity, depressive symptoms, and smoking status
When stratifying by age group (e.g., midlife versus late-life), the average social network size was greater in the midlife group compared to the late-life group (midlife average SNI score = 2.56 versus late-life average SNI score = 2.18). Large social network size in the midlife group was significantly associated with greater resilience to neurodegeneration in executive function compared to small social network size (β = 0.11; 95% CI = 0.01–0.21; p = 0.03), but not in the late-life group (β = 0.02; 95% CI = -0.09–0.12; p = 0.78) (Table 3). In the same model, medium social network size was not significantly associated with resilience to neurodegeneration in executive function compared to small social network size in either the midlife or late-life group. There were no significant associations between social network size (large or medium social network size compared to small network size) and resilience to neurodegeneration in memory and language in either the midlife group or late-life group. There were no significant statistical interactions between social network size and age group in the three cognitive domains examined.
Table 3.
Associations between social network size and resilience to neurodegeneration in memory, language, and executive function, stratified by midlife and late-life groups
| Midlife group (n = 684) | ||||||
|---|---|---|---|---|---|---|
| Memory | Language | Executive Function | ||||
| β (95% CI) | P | β (95% CI) | P | β (95% CI) | P | |
| Social Network Size | ||||||
| Small | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Medium | 0.02 (-0.07–0.10) | 0.73 | 0.04 (-0.07–0.15) | 0.47 | 0.07 (-0.04–0.17) | 0.20 |
| Large | 0.08 (-0.01–0.16) | 0.06 | 0.08 (-0.02–0.18) | 0.13 | 0.11 (0.01–0.21) | 0.03 |
| Late-life group (n = 918) | ||||||
|---|---|---|---|---|---|---|
| Memory | Language | Executive Function | ||||
| β (95% CI) | P | β (95% CI) | P | β (95% CI) | P | |
| Social Network Size | ||||||
| Small | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Medium | 0.02 (-0.09–0.12) | 0.72 | -0.08 (-0.18–0.03) | 0.16 | -0.04 (-0.15–0.08) | 0.51 |
| Large | 0.06 (-0.04–0.15) | 0.24 | -0.03 (-0.13–0.07) | 0.59 | 0.02 (-0.09–0.12) | 0.78 |
CI Confidence interval
All models adjusted for age, sex, APOE, time between exam date where plasma t-tau was measured and NP exam date, physical activity, depressive symptoms, and smoking status
There were no significant relationships between social network size and resilience to neurodegeneration in memory, language, and executive function when stratifying by sex (Supplemental Table 1). When stratifying by APOE ε4 allele carrier status, there was a significant association between large social network size and greater resilience to neurodegeneration in memory compared to small social network size among those without the ε4 allele only (β = 0.08; 95% CI = 0.01–0.15; p = 0.02) (Supplemental Table 2). No significant associations between social network size and resilience to neurodegeneration in language and executive function were observed among ε4 allele carriers or non-carriers. There were no significant statistical interactions between social network size and sex on resilience, as well as social network size and APOE genotype on resilience, in the three cognitive domains examined.
Discussion
In this study, we investigated the relationship between social network size and resilience to neurodegeneration in memory, language, and executive function among a community-based sample of mid-to-late-life adults. Based on our primary findings, we found that a large social network size was associated with a small, but significant increase in resilience to neurodegeneration in memory, but not in language or executive function, as compared to a small social network size. This aligns with findings from a systematic review where social networks were also associated with episodic and working memory [31], and from previous studies that examined overall cognitive function as an outcome. One of these studies, which recruited older adults in long-term care, found that greater social networks were associated with better cognitive performance [32]. Findings were also consistent in a cohort of primary care patients 85 years and older, where small social networks and changing social network size over time were each associated with lower cognitive function compared to those with larger social networks [5].
Our overall findings related to cognitive resilience support existing evidence that large social network sizes may be associated with decreased risk of dementia [10]. Large social network size may serve as one factor to mitigate some of the deleterious effects of neurodegeneration on cognitive function by promoting cognitive stimulation in social environments [13] and limiting social isolation, which is associated with cognitive decline [33]. Social stimulation associated with large social networks may also increase cognitive reserve. It is possible that within larger networks, individuals may have more social roles within their network (e.g., being a relative, friend, colleague), which has been found to be associated with higher Mini-Mental State Examination (MMSE) scores [34]. Bennett et al. found that having large social networks modified the relationship between neurofibrillary tangles and cognitive function in multiple domains, with the strongest association observed in semantic memory [35]. This supports the notion that large social networks may be associated with greater resilience.
Our study also found that large social network size was only associated with resilience to neurodegeneration in executive function among the midlife age group (< 65 years). In addition, we did not observe associations between social network size and resilience to neurodegeneration in memory in either the midlife or late-life group, whereas in the overall sample, large social network size was associated with greater resilience to neurodegeneration in memory. There are studies that found a positive relationship between social network size and cognition in late-life adults [5, 15], but only two studies among midlife adults found that larger networks were associated with higher cognitive scores in global cognition [36, 37]. Given that we conducted these age group-specific analyses as an exploratory analysis, interpretation of the findings should be done cautiously and that the discrepant findings should not be interpreted as differential effects across age groups, but could rather reflect reduced statistical power or false positives due to multiple testing. Additional studies with larger samples of midlife and late-life adults are needed to confirm whether there are any relationships between social network size across the life course and resilience to neurodegeneration in specific cognitive domains.
Previous studies have found that the impact of social connectedness on cognition may be differentially associated by APOE genotype, though systematic exploration of social risk factors in genetically at-risk populations is still limited. We found that there was a significant relationship between large social network size and greater resilience to neurodegeneration in memory only among those who were ε4 non-carriers, although no evidence of a statistical interaction between social network size and APOE ε4 carrier status on resilience to neurodegeneration was observed. A study from the Aging, Demographics, and Memory Study did find that the social environment, which included living arrangements, perceived social support, and loneliness, moderated the relationship between APOE ε4 allele and cognitive function in older adults [38]. Another smaller study in 115 older adults found that as social support increased, cognitive function improved among APOE ε4 carriers, but not among APOE ε4 non-carriers [39]. Our results in the context of these previous findings should be interpreted with caution since they were conducted as exploratory analyses. Further studies with larger samples are recommended to confirm whether the relationship between social network size and cognitive resilience to neurodegeneration differs by APOE ε4 carrier status.
We found null associations related to resilience to neurodegeneration in language and executive function in the overall sample, but there are mixed findings on the cognitive domains among studies. Some researchers did find an association between large social network size and executive function [16] while others found null results for memory and executive function [40]. The differences in findings may be due to other underlying mechanisms that should be explored, such as assessing the quality of relationships and how stimulating the social interactions are. One study found that more frequent social interaction, and maintaining this over time, was associated with higher global cognitive function scores [41]. Another study saw associations between engaging in social activities and better cognitive function [42]. Frequent social interaction and engaging in social activities may potentially be associated with cognitive function through other mechanisms, such as perceived social support providing a buffer to stress and verbal interaction providing cognitive stimulation [43]. These types of qualitative aspects of social networks could be protective against dementia [12] and cognitive decline [41–43], but are not captured through social network size.
Strengths of this study include a large, community-based sample, use of a well-validated marker of neurodegeneration in plasma t-tau, and assessment of multiple cognitive domains by including carefully calibrated and harmonized NP factor scores. There were also several limitations in the study. The sample predominantly consisted of White participants, which limits the generalizability of our study since social network size and behavior can vary by culture [44] and socioeconomic status [45]. Another limitation is that our analyses did not include the frequency of social interactions, or examine the quality or strength of social connections, and whether participants engaged in social activities (e.g., senior center, religious group, volunteer group, etc.), which complements social network size and can promote stimulation to increase resilience to neurodegeneration through cognitive reserve. Future analyses will extend our current work focused on social network size and investigate the relationship between the quality or strength of social networks on cognitive resilience by incorporating sub-components of the SNI questionnaire. Given the use of plasma t-tau as a general, non-specific marker of neurodegeneration in the current analysis, we acknowledge that our findings are not reflective of any specific neurodegenerative disease, such as Alzheimer’s disease (AD). Another future direction will be to incorporate more disease-specific biomarkers, such as p-tau for AD, or imaging-based measures or other plasma-based markers, such as neurofilament light chain, to measure resilience to AD-specific pathology and neurodegeneration. Finally, causal relationships cannot be drawn from our findings because of the possibility of residual confounding and unmeasured confounding, as well as lack of clear temporality.
Conclusion
Our study found that having a large social network size was associated with greater resilience to neurodegeneration in memory in a sample of mid-to-late-life adults. As a suggestive preliminary finding, large social network size in midlife adults specifically was associated with greater resilience to neurodegeneration in executive function. These findings contribute to understanding the potential the role that social networks may have in supporting healthy cognitive aging and at what time in adulthood is a critical point to intervene, which has implications on developing social-based interventions to delay the onset of, or possibly prevent dementia. Future studies should consider longitudinal measures of social networks and whether it is associated with resilience, as well as replicate our findings in more diverse populations.
Supplementary Information
Acknowledgements
The authors would like to acknowledge the Framingham Heart Study participants, staff, and study personnel.
Abbreviations
- AD
Alzheimer’s disease
- ADRD
Alzheimer’s disease and related dementias
- APOE
Apolipoprotein E
- CI
Confidence interval
- DNA
Deoxyribonucleic acid
- FHS
Framingham Heart Study
- mL
milliliters
- MMSE
Mini-Mental State Examination
- NP
Neuropsychological
- pg
picograms
- SD
Standard deviation
- SNI
Social Network Index
- p-tau
Phosphorylated tau
- t-tau
Total tau
Authors’ contributions
KH analyzed and interpreted the data, and drafted and revised the work; PH conceived the work, and substantially revised the work.
Funding
Support for this work came from the Alzheimer’s Association (grant number 23AARF-1019048) and from the Framingham Heart Study of the National Heart Lung and Blood Institute of the National Institutes of Health and Boston University School of Medicine. This project has been funded in whole or in part with Federal funds from the National Heart Lung and Blood Institute, Department of Health and Human Services, under Contract No. 75N92019D00031.
Data availability
The data that support the findings of this study are available from the Framingham Heart study, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of the Framingham Heart Study.
Declarations
Ethics approval and consent to participate
The study was conducted in accordance with the Declaration of Helsinki, and approved by the institutional review board of the Boston University Medical Campus (H-43332). Informed consent to participate in the study was obtained from all participants.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Alzheimer’s Association. Alzheimer’s Disease Facts and Figures. Alzheimer’s Disease and Dementia. 2024 [cited 2024 Nov 26]. Available from: https://www.alz.org/alzheimers-dementia/facts-figures
- 2.NIA. What Is Dementia? Symptoms, Types, and Diagnosis. National Institute on Aging. 2022 [cited 2024 Dec 1]. Available from: https://www.nia.nih.gov/health/alzheimers-and-dementia/what-dementia-symptoms-types-and-diagnosis
- 3.Masurkar AV, Marsh K, Morgan B, Leitner D, Wisniewski T. Factors Affecting Resilience and Prevention of Alzheimer’s Disease and Related Dementias. Ann Neurol. 2024;96(4):633–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Wrzus C, Hänel M, Wagner J, Neyer FJ. Social network changes and life events across the life span: a meta-analysis. Psychol Bull. 2013;139(1):53–80. [DOI] [PubMed] [Google Scholar]
- 5.Röhr S, Löbner M, Gühne U, Heser K, Kleineidam L, Pentzek M, et al. Changes in Social Network Size Are Associated With Cognitive Changes in the Oldest-Old. Front Psychiatry. 2020;11:330. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Stern Y, Arenaza-Urquijo EM, Bartrés-Faz D, Belleville S, Cantilon M, Chetelat G, et al. Whitepaper: Defining and investigating cognitive reserve, brain reserve, and brain maintenance. Alzheimers Dement J Alzheimers Assoc. 2020 ;16(9):1305–11. [DOI] [PMC free article] [PubMed]
- 7.Arenaza-Urquijo EM, Vemuri P. Resistance vs resilience to Alzheimer disease: Clarifying terminology for preclinical studies. Neurology. 2018;90(15):695–703. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Evans IEM, Martyr A, Collins R, Brayne C, Clare L. Social Isolation and Cognitive Function in Later Life: A Systematic Review and Meta-Analysis. J Alzheimers Dis. 2019;70(Suppl 1):S119–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Kelly ME, Duff H, Kelly S, McHugh Power JE, Brennan S, Lawlor BA, et al. The impact of social activities, social networks, social support and social relationships on the cognitive functioning of healthy older adults: a systematic review. Syst Rev. 2017;6(1):259. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Crooks VC, Lubben J, Petitti DB, Little D, Chiu V. Social Network, Cognitive Function, and Dementia Incidence Among Elderly Women. Am J Public Health 2008 ;98(7):1221–7. [DOI] [PMC free article] [PubMed]
- 11.Salinas J, Beiser A, Himali JJ, Satizabal CL, Aparicio HJ, Weinstein G, et al. Associations between social relationship measures, serum brain-derived neurotrophic factor, and risk of stroke and dementia. Alzheimers Dement Transl Res Clin Interv. 2017;3(2):229–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Kuiper JS, Zuidersma M, Oude Voshaar RC, Zuidema SU, van den Heuvel ER, Stolk RP, et al. Social relationships and risk of dementia: A systematic review and meta-analysis of longitudinal cohort studies. Ageing Res Rev. 2015 ;1:22:39–57. [DOI] [PubMed]
- 13.Peng S, Roth AR, Apostolova LG, Saykin AJ, Perry BL. Cognitively stimulating environments and cognitive reserve: the case of personal social networks. Neurobiol Aging. 2022;112:197–203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Perry BL, Roth AR, Peng S, Risacher SL, Saykin AJ, Apostolova LG. Social Networks and Cognitive Reserve: Network Structure Moderates the Association Between Amygdalar Volume and Cognitive Outcomes. J Gerontol Ser B. 2022;77(8):1490–500. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Sharifian N, Manly JJ, Brickman AM, Zahodne LB. Social network characteristics and cognitive functioning in ethnically diverse older adults: The role of network size and composition. Neuropsychology. 2019;33(7):956–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Perry BL, McConnell WR, Peng S, Roth AR, Coleman M, Manchella M, et al. Social Networks and Cognitive Function: An Evaluation of Social Bridging and Bonding Mechanisms. Gerontologist. 2022;15(6):865–75. [DOI] [PMC free article] [PubMed]
- 17.Dawber TR, Meadors GF, Moore FE. Epidemiological approaches to heart disease: the Framingham Study. Am J Public Health Nations Health. 1951;41(3):279–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kannel WB, Feinleib M, McNamara PM, Garrison RJ, Castelli WP. An investigation of coronary heart disease in families. The Framingham offspring study. Am J Epidemiol. 1979 Sept;110(3):281–90. [DOI] [PubMed]
- 19.Berkman LF, Syme SLS, Social networks, host resistance. and mortality: A nine-year follow-up study of Alameda county residents. 1979. [DOI] [PubMed]
- 20.Loucks EB, Sullivan LM, D’Agostino RB, Larson MG, Berkman LF, Benjamin EJ. Social networks and inflammatory markers in the Framingham Heart Study. J Biosoc Sci. 2006;38(6):835–42. [DOI] [PubMed] [Google Scholar]
- 21.Jack CR Jr, Bennett DA, Blennow K, Carrillo MC, Dunn B, Haeberlein SB, et al. NIA-AA Research Framework: Toward a biological definition of Alzheimer’s disease. Alzheimers Dement. 2018;14(4):535–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Pase MP, Beiser AS, Himali JJ, Satizabal CL, Aparicio HJ, DeCarli C, et al. Assessment of Plasma Total Tau Level as a Predictive Biomarker for Dementia and Related Endophenotypes. JAMA Neurol. 2019;76(5):598–606. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Mukherjee S, Choi SE, Lee ML, Scollard P, Trittschuh EH, Mez J, et al. Cognitive domain harmonization and cocalibration in studies of older adults. Neuropsychology. 2023;37(4):409–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Scollard P, Choi SE, Lee ML, Mukherjee S, Trittschuh EH, Sanders RE, et al. Ceiling effects and differential measurement precision across calibrated cognitive scores in the Framingham Study. Neuropsychology. 2023;37(4):383–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Hixson JE, Vernier DT. Restriction isotyping of human apolipoprotein E by gene amplification and cleavage with HhaI. J Lipid Res. 1990;31(3):545–8. [PubMed] [Google Scholar]
- 26.Zhong L, Xie YZ, Cao TT, Wang Z, Wang T, Li X, et al. A rapid and cost-effective method for genotyping apolipoprotein E gene polymorphism. Mol Neurodegener. 2016;11(1):2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.MacCallum RC, Zhang S, Preacher KJ, Rucker DD. On the practice of dichotomization of quantitative variables. Psychol Methods. 2002;7(1):19–40. [DOI] [PubMed] [Google Scholar]
- 28.Kannel WB, Belanger A, D’Agostino R, Israel I. Physical activity and physical demand on the job and risk of cardiovascular disease and death: the Framingham Heart Study. Am Heart J. 1986;112:820–5. [DOI] [PubMed] [Google Scholar]
- 29.Radloff LS, The CES-D, Scale. A Self-Report Depression Scale for Research in the General Population. Appl Psychol Measure. 1977;3(1):385–401. [Google Scholar]
- 30.SAS Institute Inc. SAS. SAS Institute Inc; 2023.
- 31.Yoo I. Social Networks as A Predictive Factor in Preserving Cognitive Functioning During Aging: A Systematic Review. Home Health Care Manag Pract. 2022;34(1):72–80. [Google Scholar]
- 32.Dodds L, Brayne C, Siette J. Associations between social networks, cognitive function, and quality of life among older adults in long-term care. BMC Geriatr. 2024;24(1):221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Yu B, Steptoe A, Chen Y, Jia X. Social isolation, rather than loneliness, is associated with cognitive decline in older adults: the China Health and Retirement Longitudinal Study. Psychol Med. 2021;51(14):2414–21. [DOI] [PubMed] [Google Scholar]
- 34.Ellwardt L, Van Tilburg TG, Aartsen MJ. The mix matters: Complex personal networks relate to higher cognitive functioning in old age. Soc Sci Med. 2015;125:107–15. [DOI] [PubMed] [Google Scholar]
- 35.Bennett DA, Schneider JA, Tang Y, Arnold SE, Wilson RS. The effect of social networks on the relation between Alzheimer’s disease pathology and level of cognitive function in old people: a longitudinal cohort study. Lancet Neurol. 2006;5(5):406–12. [DOI] [PubMed] [Google Scholar]
- 36.Green AF, Rebok G, Lyketsos CG. Influence of social network characteristics on cognition and functional status with aging. Int J Geriatr Psychiatry. 2008 ;23(9):972–8. [DOI] [PMC free article] [PubMed]
- 37.Holtzman RE, Rebok GW, Saczynski JS, Kouzis AC, Wilcox Doyle K, Eaton WW. Social Network Characteristics and Cognition in Middle-Aged and Older Adults. J Gerontol Ser B. 2004;59(6):P278–84. [DOI] [PubMed] [Google Scholar]
- 38.Poey JL, Burr JA, Roberts JS. Social Connectedness, Perceived Isolation, and Dementia: Does the Social Environment Moderate the Relationship Between Genetic Risk and Cognitive Well-Being? Gerontologist. 2017;57(6):1031–40. [DOI] [PubMed] [Google Scholar]
- 39.Payen A, Bateman JR, Dhanavanthri Muralidhara S, et al. Social support may buffer the effect of APOE ε4 allele on cognition in older adults. J Alzheimers Dis. 2024;102(4):967–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Rehan S, Phillips NA. Psychosocial Function in Mild Cognitive Impairment: Social Participation is Associated With Cognitive Performance in Multiple Domains. J Appl Gerontol. 2025;07334648241311661. [DOI] [PMC free article] [PubMed]
- 41.Sommerlad A, Sabia S, Singh-Manoux A, Lewis G, Livingston G. Association of social contact with dementia and cognition: 28-year follow-up of the Whitehall II cohort study. PLoS Med. 2019;16(8):e1002862. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Nie Y, Richards M, Kubinova R, Titarenko A, Malyutina S, Kozela M, et al. Social networks and cognitive function in older adults: findings from the HAPIEE study. BMC Geriatr. 2021;21(1):570. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Zuelsdorff ML, Koscik RL, Okonkwo OC, Peppard PE, Hermann BP, Sager MA, et al. Social support and verbal interaction are differentially associated with cognitive function in midlife and older age. Neuropsychol Dev Cogn B Aging Neuropsychol Cogn. 2019;26(2):144–60. [DOI] [PMC free article] [PubMed]
- 44.Ajrouch KJ, Blandon AY, Antonucci TC. Social Networks Among Men and Women: The Effects of Age and Socioeconomic Status. J Gerontol B Psychol Sci Soc Sci. 2005 ;1(6):S311–7. [DOI] [PubMed]
- 45.Igarashi T, Kashima Y, Kashima ES, Farsides T, Kim U, Strack F, et al. Culture, trust, and social networks. Asian J Soc Psychol. 2008;11(1):88–101. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data that support the findings of this study are available from the Framingham Heart study, but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of the Framingham Heart Study.