Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 May 1.
Published in final edited form as: J Women Aging. 2021 Jul 10;34(3):394–410. doi: 10.1080/08952841.2021.1945368

The Influence of Social Support on Cognitive Health in Older Women: A Women’s Health Initiative Study

Georgina L Moreno 1, Eric Ammann 2, Erin T Kaseda 3, Mark A Espeland 4, Robert Wallace 5, Jennifer Robinson 5, Natalie L Denburg 6
PMCID: PMC8743299  NIHMSID: NIHMS1725861  PMID: 34252006

Abstract

Social support is associated prospectively with cognitive decline and dementia among the elderly; however, little is known about the impact of social support on healthy neurological aging. The current study investigates whether perceived social support has an influence on neurological health among a large sample of healthy postmenopausal women. Social support and neuropsychological outcomes were measured annually for six years through the Women’s Health Initiative Study of Cognitive Aging. In postmenopausal women, higher perceived social support was associated with significantly better overall neuropsychological functioning at baseline, especially in the domains of short-delay figural memory, short-delay verbal memory, and semantic fluency. No significant associations were found between social support and longitudinal changes in neuropsychological function over a median follow-up period of six years. Additionally, there was no significant relationship between social support and regional brain volumes. These findings suggest that social support is related to performance in a subset of neuropsychological domains and contributes to the existing literature that points to the importance of social support as a modifiable lifestyle factor that has the potential to help protect against the decline of cognitive aging, specifically among older adult women.

Keywords: cognition, perceived social support, neuropsychology, aging, cognitive health

For a healthy 65-year-old woman in the United States, the subsequent lifetime risk of developing dementia is approximately 20% 1,2. As the United States population ages, efforts to identify modifiable risk factors for cognitive decline and dementia have taken on a newfound urgency. Social and emotional support has previously been linked to general health and well-being 3–5. Similarly, social support may also be a protective factor against cognitive decline and dementia in older adults 6–12. Much of the previous work investigating this relationship focuses on general cognitive functioning, providing little insight on how individual domains of cognition may be impacted. Given the complexity of cognitive and neuropsychological functioning, it is important to better understand the relationship between social support and the various domains of cognition (e.g., working memory, verbal learning). The goal of the present study is to use a more comprehensive approach to further study the relationship between perceived social support and cognitive health (i.e., performance on various types of neuropsychological testing, brain volumes) in older adult women.

Social Support as a Construct

According to Gottlieb and colleagues 13, three main terms have emerged that best characterize the study of social resources available to individuals—social support, social network, and social integration. These dimensions of support often overlap but are not interchangeable. Social support is defined by the social resources—both perceived and actual—that are available to an individual 13. The social network, in turn, is the unit of social structure that makes up an individual’s social connections. Social integration then refers to the extent to which an individual physically engages in social interactions with individuals from their social network 13.

Social support can be further assessed in varying contexts. For example, functional support is defined by the various resources made available to an individual through their social connections, and can encompass informational support (e.g., knowledge about a specific topic), instrumental support (e.g., financial assistance), and emotional support (e.g., nurturing). Structural support, on the other hand, refers to the number of social relationships that an individual has and the structure of social connections within those relationships14.

Social Support and General Health

Higher levels of social support have been linked to lower all-cause mortality and lower incidence of and mortality from heart disease, cancer, and stroke 15–20. Potential mechanisms by which social support is protective for health include increased medical adherence 21, improved cardiovascular and neuroendocrine function22, and inflammation and immune functioning 23,24. In addition to their associations with social support, vascular disease and elevated inflammatory biomarkers have also been implicated as risk factors for dementia 25,26.

Two main hypotheses related to the impact of social support on well-being emerged in early investigations: the direct effects hypothesis and the buffering hypothesis. The direct effects hypothesis posits that social support is generally beneficial for health and well-being, during both stressful and non-stressful situations 14,27. The buffering hypothesis argues that the beneficial effects of social support on health only become evident during periods of high stress, as the presence of social support is protective against the pathogenic effects of stress 14,27. Both hypotheses are well established in the literature, with some indication that evidence for the buffering hypothesis is more likely to be found if functional support is assessed, whereas evidence for the direct effects hypothesis is more likely to be found when social network integration is assessed 14,27,28.

Social Support and Cognitive Health

Several longitudinal studies support the notion that social support is protective against cognitive decline. A large prospective study by Wilson, Krueger, Arnold, Schneider, Kelly, Barnes, Tang, Bennett 29 demonstrated that baseline loneliness in healthy older adults was predictive of cognitive decline and Alzheimer’s onset. Reductions in social network complexity longitudinally are associated with declines in cognitive function 30. Similarly, subjective loneliness was related to cognitive decline over a ten-year follow-up 31. Although the precise mechanisms by which social support impacts cognition are not known, hypotheses such as stress buffering, synaptic complexity and efficiency, and fronto-striatal connectivity have been suggested 32–34.

Research also suggests a correlation between social support and brain structure and volume; however these findings are mixed and at times difficult to interpret 35–37. The volumes of the prefrontal cortex and anterior insular cortex have been positively associated with social network size among older adults 38,39. Conflicting evidence exists, however, for the association between amygdala volume and social network size, such that amygdala volume has been both positively and negatively correlated with network size 37,39. White matter tract integrity in the right cingulum, right extreme capsule, and right arcuate fasciculus has also been correlated with larger social networks 40. Cognitive decline has been most strongly linked with global grey matter loss 35,41 although cognitive decline has also been associated more specifically with atrophy in the hippocampus and amygdala, although findings are mixed 42,43. A randomized controlled trial by Ehlers, Daugherty, Burzynska, Fanning, Awick, Chaddock-Heyman, Kramer, McAuley 44 found that brain volume in the amygdala and prefrontal cortex moderated the effectiveness of a loneliness reduction training, suggesting that the relationships between brain volume and social support may have clinical implications as well. Although these relationships are evident, better quantifying the magnitude of the protective effects of social support and elucidating possible mechanisms remain important research objectives.

Social support has also been linked to aspects of cognition among healthy, normal aging populations. For example, increased levels of social support are protective against declines in memory performance 6,8,45,46. Social isolation and loneliness have been shown to be related to decreased performance in memory and executive functioning 8,45. Work by La Fleur, Salthouse 47 suggests that specific aspects of social support (e.g., contact with family and friends, emotional and informational support, anticipated support, negative interactions) are related primarily to measures of overall cognition.

Notably, many of the previous studies on social support and cognitive functioning focused on broad measures of cognitive functioning typically used to screen for cognitive impairment such as the Mini-Mental State Exam (MMSE)48, the Repeated Battery for the Assessment of Neuropsychological Status (RBANS)49, or the Telephone Interview for Cognitive Status (TICS)50. Through the MacArthur Studies of Successful Aging, Seeman and colleagues 51 provide perhaps the most comprehensive battery of cognitive functioning to date, assessing the relationship between social support and cognitive performance over a 7.5 year period in the following cognitive domains: memory, spatial recognition, similarities, story recall, figure copying, and recall naming. However, this study primarily reports on the relationship between a total cognitive performance score that was calculated as a composite of the various cognitive domains reported. That is, although a relationship between social environment factors and overall cognitive performance was reported by Seeman and colleagues 51, the study did not conclusively report any relationships between social support and the specific cognitive or neuropsychological domains. Moreover, this work did not concurrently examine whether there was a relationship between social support and brain structure and function. Given the complexity of cognitive and neuropsychological functioning, it is important to better understand the relationship between the various domains of cognition (e.g., working memory, verbal learning) and social support.

A Consideration of Gender

It is also important to consider existing gender differences observed in relation to social support and health. Historically, research in social sciences was conducted on white men. In a review by Shumaker and Hill 52, the relationship between social support and health was reported to be more equivocal among women as compared to men, and this relationship was mediated by factors such as age, marital status, and caregiving roles. Investigations into the role of social support on health broadly tend to support a moderating effect of gender on health, such that specific social support factors impact health differentially for men and women 18,20. Moreover, there are gender-based differences in the dimensionality and structure of perceived social support 53. Given that social support may vary by gender, a focused report on women is critical.

Of the previous studies on social support and cognitive functioning, very few studies have investigated the relationship between social support and cognitive functioning specifically among women. For example, Crooks and colleagues 50 found a link between smaller social networks and cognitive functioning among woman with dementia. More broadly, there are significant gender effects in dementia risk and incidence; women are nearly twice as likely to develop Alzheimer’s dementia than men, whereas vascular, Lewy Body, and Parkinson disease dementias are more prevalent in men 54. As age is a major risk factor for dementia, and as women tend to outlive men, the adverse effects of vascular and other risk factors on dementia may impact women more than they do men 54. Given the differential effects of social support on health in men and women, and differences in dementia incidence and risk factors in men and women, a mixed-gender sample may obscure important gender differences, making it imperative to investigate the relationship between social support and cognitive health independently among a cohort of women.

The Present Study

In summary, previous research on social support and cognitive health has primarily focused on composite measures of cognition or tasks used to screen for cognitive impairment. The relationship between social support and specific domains of neuropsychological and cognitive function remains to be fully explored. Moreover, given the existing gender differences related to social support, it is important to consider the relationship between social support and cognitive health among women specifically. In the present study, the relationship between social support and cognitive health (i.e., neuropsychological functioning, brain volumes) was examined in a cohort of postmenopausal women who participated in the Women’s Health Initiative (WHI) Study of Cognitive Aging (WHISCA). More specifically, using linear mixed modeling, the aim of the current study was to investigate the relationship between perceived social support and nine independent domains of neuropsychological functioning: overall cognitive function, short-delay figural memory, visuospatial judgment, long-delay verbal memory, short-delay verbal memory, semantic fluency, phonemic fluency, verbal knowledge, and working memory/attention. It was hypothesized that among postmenopausal women higher levels of perceived social support would be predictive of better neuropsychological functioning. It was also hypothesized that social support would be predictive of the yearly rate of change in neuropsychological functioning. Additionally, it was hypothesized that among postmenopausal women, higher levels of social support would be related to larger brain volumes, as measured by magnetic resonance imaging (MRI).

Methods

Study Population

This study was a secondary analysis of longitudinal data collected for the WHI Memory Study (WHIMS) and the WHI Study of Cognitive Aging (WHISCA). The WHIMS and WHISCA are ancillary studies of women who participated in the WHI Hormone Therapy (HT) trials. In the WHI HT trials, 27,347 postmenopausal women were randomly assigned to receive hormone therapy (estrogen alone or estrogen and progestin based on prior hysterectomy status) or placebo 55,56. Specifics on participant enrollment and recruitment for the WHI HT study have been previously described in detail 57–59. Briefly, all women enrolled in the WHI HT trial were between 50 and 79 years old and postmenopausal at the time of enrollment and were recruited from 40 WHI clinical centers throughout the United States, with enrollment of racial and ethnic minority groups proportionate to the total minority population of women between the ages of 50 and 79 years of age at the time 58. All informed consent and procedures were reviewed by The Data and Safety Monitoring Board (DSMB)60 of the WHI and local WHI clinical sites. Major exclusion criteria related to competing risks with HT (e.g., invasive cancer in the past 10 years; acute myocardial infarction, stroke). The estrogen plus progestin (E+P) and estrogen-alone (E-only) HT trials were terminated early, in 2002 and 2004 respectively, when it was established that the risk-benefit profile of HT was unfavorable 55,56.

Of the 27,347 postmenopausal women in the WHI HT trials, 7,479 participants were recruited to the WHIMS cohort at 39 of the 40 WHI clinical centers. The WHIMS was a randomized, double-blind, placebo-controlled clinical trial that investigated the effects of HT on the risk of dementia and changes in global cognitive function. To be included as a participant in the WHIMS study, participants needed to be at least 65 years old and free of dementia at baseline. The WHIMS clinical trial had two arms—estrogen and progestin or estrogen alone. Of the 7,479 participants recruited to the WHIMS clinical trials, 4,532 were randomly assigned to either the combined estrogen and progestin hormone therapy or placebo, and the other 2,947 participants were randomly assigned to receive estrogen alone or placebo.

The WHISCA ancillary study evaluated the effect of HT on domain-specific neuropsychological functioning 61. As described by Resnick, Coker, Maki, Rapp, Espeland, and Shumaker 62, WHISCA participants were enrolled from 14 of the 39 WHIMS clinical sites, with priority given to sites that allowed for the retention of participants and selected to increase the diversity in recruitment with respect to geographic, ethic, and racial identity. Participants were recruited to WHISCA on an average of three years after their initial entry into the HT trials. From the 14 select sites, WHISCA enrolled 2,302 participants. Median length of follow-up for cognitive outcomes in the WHISCA cohort was approximately six years, including both on-trial and post-trial follow-up, with start of follow up starting a median length of 3 years after HT trial entry. The WHIMS and WHISCA data indicated that HT also had mild-to-moderate negative effects on cognitive function, with estrogen plus progestin being somewhat more deleterious than estrogen-only 61. As with the WHI HT trials, the WHIMS trials were terminated early when it was established that the risk-benefit profile of HT was unfavorable 55,56.

The WHIMS-MRI study was an ancillary study of the WHIMS, and participants were recruited from 14 of the 39 WHIMS clinical centers. Clinical centers were selected based on participation in the WHISCA study and based on the experience with multicenter MRI studies at the clinical site. WHIMS-MRI recruited a total of 1,527 participants and successfully collected MRI scans on 1,403 of those participants. Median length of follow-up for cognitive outcomes in the WHIMS-MRI cohort was approximately eight years after the start of the WHIMS.

For the current study, of the WHIMS cohort of 7,479, participants were excluded if they were missing data on social support at baseline (N = 246) or if participants self-reported degenerative neurological disease at baseline. Therefore, for the current study, complete WHISCA data was available on 2,242 participants and complete WHIMS-MRI data was available for 1,337 participants. See Figure 1 for details.

Figure 1.

Figure 1.

Open in a new tab

Flow diagram showing selection of eligible participants from the Women’s Health Initiative Memory Study (WHIMS) and Women’s Health Initiative Study of Cognitive Aging (WHISCA).

Perceived Social Support Measurement

Perceived social support was assessed at trial enrollment with nine survey questions adapted from the Medical Outcome Study (MOS) Social Support Survey 62,63. Each question asked how often the respondent felt that social support was available to them in a particular domain or situation, e.g., “[Do you have] someone you can count on to listen to you when you need to talk?” Possible responses for each item ranged from 1 (none of the time) to 5 (all of the time). Numeric values associated with the responses to all questions were averaged to create a single composite measure of perceived social support, with scores ranging from 1 (low) to 5 (high). See Appendix Table 1 for details.

Neuropsychological Assessment

Details of the WHISCA neuropsychological study have been published previously62. Briefly, based on the recommendation of the National Institute on Aging (NIA) investigators with longitudinal study experience on age-related changes in cognition (e.g., memory) and with studies related to the hormonal effects on cognition, WHISCA participants completed annually a comprehensive battery of neuropsychological tests. The neuropsychological instruments selected for the current study were: 1) Primary Mental Abilities (PMA) Vocabulary Test, to assess verbal knowledge; 2) Category Fluency Test, to assess semantic fluency; 3) Letter Fluency Test, to assess phonemic fluency; 4) Benton Visual Retention Test (BVRT), to assess short-delay figural memory; 5) California Verbal Learning Test (CVLT) short free recall task, to assess short-delay verbal memory; 6) CVLT delayed free recall task, to assess long-delay verbal memory; 7) Digit Span Test, to assess attention and working memory; and 8) Card Rotation Test, to assess visuospatial judgment.

To simplify the reporting and interpretation of these results, WHISCA test scores were standardized to Z-scores based on the sample mean and standard deviation for each instrument at baseline. (A standardized score of 0.10 would imply neuropsychological performance 0.10 standard deviations above the sample mean.) The Z-scores for all eight neuropsychological tests were averaged to obtain a measure of overall neuropsychological function, which was also standardized to have a mean of zero and standard deviation of one. We designated the composite measure of overall neuropsychological function as the primary endpoint, and the domain-specific neuropsychological outcomes as secondary endpoints. Summary statistics for the WHISCA participants’ raw neuropsychological test scores are provided in Appendix Table 2.

We also evaluated associations between social support and regional brain volumes as measured by magnetic resonance imaging (MRI). Magnetic resonance imaging scans were performed on a subset of WHIMS participants at study closeout, an average of eight years following the WHI HT trial randomization. Details of the WHIMS-MRI study have been published previously 64,65. Briefly, MRI scans were performed in accordance with a standardized protocol, and validated computer algorithms were used to process the images and estimate regional brain volumes. In the present study, we focused on possible associations between social support and the volumes of the whole brain, frontal lobe, temporal lobe, hippocampus, and ventricles.

Covariates

A number of trial, demographic, and health status variables were considered as potential confounders of the relationship between social support and neuropsychological function. As with social support, all covariates were assessed at trial entry. Covariate selection was determined a priori. In model 1 we included social support, age, and HT trial arm assignment. In model 2, we further adjusted for education, income, race, region, and primary job classification. Model 3 included all model 2 covariates plus a count of major medical comorbidities adapted from a prior study of morbidity among WHI participants 66.We also assessed the relationship between social support and regional brain volumes as estimated by MRI. Following Resnick and colleagues 64, we additionally adjusted for intracranial volume in these analyses. Covariates were otherwise the same as those described above for the neuropsychological measures.

Statistical Methods

Linear mixed models with increasing levels of covariate adjustment were used to estimate the cross-sectional and longitudinal associations of social support on neuropsychological function. Random intercept and time effects were used to account for within-person correlations in test scores over time. All fixed effects were included as both main effects and interactions with time, so that influences on neuropsychological performance at baseline and the rate of change in neuropsychological function over the study period could be evaluated. Restricted cubic spline transformations of all continuous and ordinal covariates (social support, time, education, income, number of major comorbidities) were considered to allow for flexible modeling of possible non-linear relationships with neuropsychological performance 67.

In the linear regression models of neuropsychological function, all available data points were used to estimate mean baseline neuropsychological performance and changes in neuropsychological performance over time. Social support was treated as a continuous predictor variable. Since there was no evidence for a non-linear relationship between social support and neuropsychological function, effect estimates are reported as the expected mean difference in standardized neuropsychological performance associated with better performance by a single-unit increase in social support on a 5-point scale.

Linear mixed models were also used to evaluate the relationship between social support and regional brain volumes. For these analyses, only one MRI measurement time-point was available for each participant. To account for the significant within-study center correlations that were observed in the MRI measurements, study center was included as a random effect in the MRI outcomes regression models.

Of the 17 trial, demographic, and health status variables identified as potential confounding factors, 9.5% of participants had at a missing value for at least one variable. A total of 0.7% of all covariate values were missing, with household income having the highest rate of missing data (5.5%). A fully conditional specification approach 68 was used to impute these missing values. All missing values were for categorical variables, and discriminant functions were used to predict likely values for these missing observations.

To quantify the uncertainty introduced by imputation, we repeated the imputation process 10 times with different random seeds to create 10 complete datasets. For each endpoint, the 10 datasets were analyzed separately and the parameter estimates were pooled using Rubin’s method 69. Compared to a complete-case analysis, where participants with any missing data are excluded entirely, multiple imputation of missing values is more statistically efficient and generally more robust against potential sources of bias 70. Statistical analyses were performed with the SAS 9.4 software package (SAS Institute, Cary, NC).

Results

Two thousand two hundred and forty-two WHISCA participants were eligible for inclusion in this study (See Figure 1). Their median age was 73 years (range: 65–83) and median baseline 3MSE (Modified Mini-Mental State Examination)71 score was 97 (range: 68–100). All were free of dementia at enrollment. Seventy-four percent of participants had some post-secondary education; 90% were white. Additional cohort characteristics are described in Table 1. Median length of follow-up was six years for the WHISCA cohort.

Table 1.

Baseline WHISCA participant characteristics by level of perceived social support.

Social Support (5-point scale)
Variables Total N=2242(%) 1 to 3 (None, little or some time) N=581(%) 4 (Most of the time) N=939(%) 5 (All of the time) N=722(%)
HRT Arm
 E-alone intervention 432 (19.3%) 135 (23.2%) 165 (17.6%) 132 (18.3%)
 E-alone control 444 (19.8%) 124 (21.3%) 190 (20.2%) 130 (18.0%)
 E+P intervention 668 (29.8%) 158 (27.2%) 295 (31.4%) 215 (29.8%)
 E+P control 698 (31.1%) 164 (28.2%) 289 (30.8%) 245 (33.9%)
Age at WHISCA start (years) 72.5 (65.9 – 83.5) 73.0 (65.9 – 83.1) 72.5 (66.0 – 83.5) 72.1 (65.9 – 82.1)
Baseline 3MSE score 97 (68 – 100) 96 (79 – 100) 97 (68 – 100) 97 (73 – 100)
Education
 College or more 725 (32.3) 171 (29.4) 303 (32.3) 251 (34.8)
 Some post-HS education 923 (41.2) 252 (43.4) 385 (41.0) 286 (39.6)
 HS or less 588 (26.2) 157 (27.0) 248 (26.4) 183 (25.3)
 Missing 6 1 3 2
Income
 $50K+ 456 (20.3) 97 (16.7) 173 (18.4) 186 (25.8)
 $20K to <50K 1165 (52.0) 283 (48.7) 495 (52.7) 387 (53.6)
 <$20K 497 (22.2) 175 (30.1) 221 (23.5) 101 (14.0)
 Missing 124 26 50 48
Occupation
 Managerial / Professional 794 (35.4) 190 (32.7) 338 (36.0) 266 (36.8)
 Technical / Sales / Admin 675 (30.1) 194 (33.4) 281 (29.9) 200 (27.7)
 Service / Labor 484 (21.6) 139 (23.9) 189 (20.1) 156 (21.6)
 Homemaker only 248 (11.1) 46 (7.9) 114 (12.1) 88 (12.2)
 Missing 41 12 17 12
Race
 Black 139 (6.2) 49 (8.4) 55 (5.9) 35 (4.8)
 White 2020 (90.1) 506 (87.1) 850 (90.5) 664 (92.0)
 Other 79 (3.5) 25 (4.3) 32 (3.4) 22 (3.0)
 Missing 4 1 2 1
U. S. Region
 Northeast 464 (20.7) 151 (26.0) 180 (19.2) 133 (18.4)
 South 338 (15.1) 80 (13.8) 144 (15.3) 114 (15.8)
 Midwest 861 (38.4) 205 (35.3) 369 (39.3) 287 (39.8)
 West 579 (25.8) 145 (25.0) 246 (26.2) 188 (26.0)
 Missing 0 0 0 0
Comorbidity count
 0 861 (38.4) 206 (35.5) 366 (39.0) 289 (40.0)
 1 951 (42.4) 246 (42.3) 405 (43.1) 300 (41.6)
 2 300 (13.4) 85 (14.6) 116 (12.4) 99 (13.7)
 3+ 77 (3.4) 28 (4.8) 29 (3.1) 20 (2.8)
 Missing 53 16 23 14
Open in a new tab

For categorical variables, counts and column percentages are shown. For continuous variables (age and baseline 3MSE score), the median, minimum and maximum by group are shown.

Comorbidity count includes the following health conditions: myocardial infarction, heart failure, peripheral arterial disease, cerebrovascular disease, diabetes, hypertension, chronic obstructive pulmonary disease, cancer, mild liver disease, and colitis.

The mean level of social support reported by WHISCA participants was 4 (i.e., support was available “most of the time” across the evaluated domains). An average social support score of 5 (highest) was observed in 32% of participants, 4 in 42%, 3 in 19%, 2 in 6% and 1 (lowest) in 1%. On average, women with higher levels of social support were slightly younger and more likely to have a higher household income, to be white, and to have fewer chronic health conditions (Table 1).

For the primary endpoint of overall neuropsychological function, after partial covariate adjustment (model 1), a single-unit higher level of social support was associated with a mean of 0.11 (95% CI: 0.07, 0.16) standard deviations (SD) better performance at baseline (Table 2). After full covariate adjustment (model 3), this cross-sectional effect estimate was reduced to 0.07 SD (95% CI: 0.02, 0.11) (Table 2). There was no significant association between social support and the yearly rate of change in overall neuropsychological function (Table 2, Figure 2).

Table 2.

Estimated cross-sectional and longitudinal effects of 1-unit better social support on cognitive outcomes with increasing levels of covariate adjustment.

Outcome Model Cross-sectional effect estimate (95% CI) p value Yearly change effect estimate (95% CI) p value
1 Overall cognitive function 1 0.114 (0.069, 0.160) < .001 0.001 (−0.005, 0.007) 0.79
1 Overall cognitive function 2 0.068 (0.025, 0.111) < .001 −0.001 (−0.007, 0.006) 0.88
1 Overall cognitive function 3 0.066 (0.024, 0.109) < .001 −0.001 (−0.007, 0.006) 0.79
2 Short-delay figural memory 1 0.102 (0.060, 0.145) < .001 −0.006 (−0.013, 0.001) 0.12
2 Short-delay figural memory 2 0.067 (0.026, 0.108) < .001 −0.007 (−0.014, 0.000) 0.06
2 Short-delay figural memory 3 0.065 (0.024, 0.106) < .001 −0.007 (−0.014, 0.000) 0.06
3 Visuospatial judgment 1 0.060 (0.015, 0.104) 0.01 −0.001 (−0.008, 0.005) 0.69
3 Visuospatial judgment 2 0.041 (−0.004, 0.087) 0.08 −0.003 (−0.009, 0.004) 0.38
3 Visuospatial judgment 3 0.039 (−0.006, 0.085) 0.09 −0.003 (−0.010, 0.003) 0.34
4 Long-delay verbal memory 1 0.063 (0.021, 0.106) < .001 0.009 (0.001, 0.017) 0.04
4 Long-delay verbal memory 2 0.038 (−0.006, 0.081) 0.09 0.007 (−0.002, 0.015) 0.12
4 Long-delay verbal memory 3 0.037 (−0.007, 0.080) 0.10 0.007 (−0.002, 0.015) 0.13
5 Short-delay verbal memory 1 0.077 (0.035, 0.119) < .001 0.003 (−0.005, 0.011) 0.43
5 Short-delay verbal memory 2 0.052 (0.010, 0.094) 0.02 0.001 (−0.007, 0.009) 0.78
5 Short-delay verbal memory 3 0.050 (0.008, 0.093) 0.02 0.001 (−0.007, 0.009) 0.82
6 Semantic fluency 1 0.107 (0.063, 0.150) < .001 −0.002 (−0.009, 0.005) 0.63
6 Semantic fluency 2 0.077 (0.034, 0.121) < .001 −0.002 (−0.009, 0.005) 0.61
6 Semantic fluency 3 0.076 (0.032, 0.120) < .001 −0.002 (−0.009, 0.005) 0.57
7 Phonemic fluency 1 0.040 (−0.006, 0.086) 0.10 0.004 (−0.002, 0.009) 0.23
7 Phonemic fluency 2 0.015 (−0.031, 0.061) 0.52 0.001 (−0.005, 0.007) 0.66
7 Phonemic fluency 3 0.014 (−0.032, 0.060) 0.55 0.001 (−0.005, 0.007) 0.70
8 Verbal knowledge 1 0.089 (0.043, 0.134) < .001 0.003 (−0.002, 0.008) 0.21
8 Verbal knowledge 2 0.040 (−0.002, 0.083) 0.06 0.004 (−0.002, 0.009) 0.17
8 Verbal knowledge 3 0.039 (−0.003, 0.081) 0.07 0.003 (−0.002, 0.008) 0.20
9 Working memory/attention 1 0.044 (−0.001, 0.088) 0.06 0.001 (−0.005, 0.006) 0.86
9 Working memory/attention 2 0.017 (−0.028, 0.062) 0.46 0.000 (−0.006, 0.006) 1.00
9 Working memory/attention 3 0.017 (−0.028, 0.062) 0.45 −0.000 (−0.006, 0.006) 0.96
Open in a new tab

Covariates included in each model:

Model 1: social support, hormone therapy trial assignment, age.

Model 2: model 1 covariates + education, income, occupation, race and region.

Model 3: model 2 covariates + medical comorbidity count (myocardial infarction, heart failure, peripheral arterial disease, cerebrovascular disease, diabetes, hypertension, chronic obstructive pulmonary disease, cancer, mild liver disease, colitis).

Figure 2.

Figure 2.

Open in a new tab

Predicted mean overall cognitive function over time (with 95% confidence bands) by degree of perceived social support for a typical cohort member after full covariate adjustment. Caption: Predicted mean standardized score for a typical cohort member with social support of 4.5 (red line) vs. 3.5 (blue line) on a 5-point scale. Shaded areas represent 95% confidence bands. Predicted trajectories are based on model 3 parameter estimates, adjusted for hormone therapy trial assignment, age, education, income, occupation, race, region, and medical comorbidity count (myocardial infarction, heart failure, peripheral arterial disease, cerebrovascular disease, diabetes, hypertension, chronic obstructive pulmonary disease, cancer, mild liver disease, and colitis).

For the secondary endpoints of domain-specific neuropsychological outcomes similar patterns were observed for most of the individual domains (allowing for sampling variability). Of significance, after full covariate adjustment (model 3), a single-unit higher level of social support was associated with better performance by 0.07 SD (95% CI: 0.02, 0.11) in short-delay figural memory, 0.05 SD (95% CI: 0.01, 0.09) in short-delay verbal memory, and 0.08 SD (95% CI: 0.03, 0.12) in semantic fluency. There were no significant associations between social support and yearly rate of change in any of the domain-specific neuropsychological functions (Table 2).

MRI measurements and social support data were available for 1,337 WHIMS participants, of whom 1,030 were members of the WHISCA sub-cohort (Figure 1). We assessed associations between social support and regional brain volumes estimated from MRI among these 1,337 women. In all models, no significant association was found between social support and total brain volume, frontal lobe volume, temporal lobe volume, hippocampal volume, or ventricular volume (Appendix Table 3, Appendix Figure 1).

Discussion

Summary of Results

Using data collected through the WHI, the current set of analyses investigated the relationship between social support and cognitive health in a large sample of healthy older women. In support of the hypothesis that higher levels of social support would predict cognitive health, cross-sectional analyses demonstrated that more social support at baseline was predictive of significantly better overall neuropsychological functioning, even after adjusting for a host of relevant trial, demographic, and health status variables. Moreover, when assessing distinct neuropsychological domains, significant associations were observed between social support and short-delay figural memory, short-delay verbal memory, and semantic fluency. In terms of effect magnitude, a one-unit higher social support score (e.g., “4-most of the time” versus “3-some of the time”) was associated with a 0.05–0.10 SD increase in overall neuropsychological function in these domains.

The data did not support the hypothesis that longitudinal changes would be observed in the relationship between social support and neuropsychological functioning. That is, no significant associations were found between social support and longitudinal changes in neuropsychological function over a median follow-up period of six years. Additionally, in the current study we did not observe a relationship between social support and regional brain volumes. We discuss the implications of these findings for understanding the relationship between social support and cognitive health, specifically among postmenopausal women.

The Role of Social Support in Cognitive Health Among Postmenopausal Women

Although these results suggest that the impact of social support on cognitive aging is likely to be modest, the current research contributes to a large existing literature on the relationship between social support and a variety of emotional, physical, and cognitive health outcomes. In the literature it is clear that social support plays an important role in promoting emotional wellness and reducing the risk of depression 72,73. For these reasons, identifying community models and interventions that promote social interaction and social support in older adult populations remains an important public health priority. Moreover, given the relationship observed between social support and specific domains of neuropsychological functioning, women may be particularly vulnerable to the effects of social support on cognitive health.

An interesting finding in the current study is that the effects of social support were particularly robust in the domains of short-delay figural memory, short-delay verbal memory, and semantic fluency. One can speculate as to what makes these neuropsychological tasks unique as compared to the other neuropsychological domains of interest. For instance, perhaps the mechanisms involved in these processes are more susceptible to the protective effects of social support factors than other cognitive processes. Interestingly, short-delay figural memory and short-delay verbal memory both typically are associated with more frontal memory systems. Moreover, semantic fluency has been shown to decline at a greater rate with age as compared to phonemic fluency 74 and discrete neural correlates of semantic fluency, primarily in relation to phonemic fluency, have been reported in the literature, particularly in frontal brain regions 75,76. Future work should use alternative neuroimaging methods (e.g., EEG, fMRI) to further explore the neural mechanisms that potentially could be affected by social support.

It is important to consider the current findings in the context of previous work that has been reported on the relationship between social support and cognition. Social support has been found to have a protective effect on rates of memory decline 6, and has been linked to better performance across more general cognitive performance and measures of cognitive decline 8,45,51,77. The current study differs from this previous work in that the neuropsychological testing in the WHISCA was more comprehensive, using a series of tasks designed to specifically tap into each of the reported neuropsychological domains.

It is also important to consider the null findings that social support was not predictive of yearly rate of change in neuropsychological functioning. These null findings are not entirely surprising given that in longitudinal studies of age-related changes in cognition, retest effects have been found to lead to discrepancies in age trends 78. More specifically, cross-sectional comparisons consistently reveal a relationship between age and cognitive performance; however, these findings often do not hold in longitudinal comparisons. Also of note, similar findings were observed in a study by Brown and colleagues 79 investigating the relationship between cognition and social activity (i.e., number of hours an individual engaged in social activity). That is, Brown and colleagues (2012) observed a relationship between baseline social activity and cognition, however, these results were not consistent and did not hold longitudinally. Interestingly, consistent with this observation of retest effects in longitudinal neuropsychological testing, in the current study there is a slight, although insignificant, increase in trajectory of the mean overall cognitive function over time, which over the years levels off (Figure 2).

Additionally, in the current study we did not observe a relationship between social support and regional brain volumes, so we are unable make definitive inferences as to what is occurring at the neural level. It is important to note that in general it has been shown that the relationship between brain structure and cognitive variables is not always robust, with very weak evidence showing that age-related changes in cognition are attributed to age-related changes in brain structure (for a review see Salthouse 80). One could argue that the current null MRI findings could partially be explained by volumetric MRI data not being a strong indicator of function. That is, perhaps the neural functioning driving this relationship is functional, rather than structural. Alternatively, it is possible that social support has a protective effect on cognition, leading to an effect similar to what is reported as cognitive reserve, the notion that there is a disconnect between changes in the brain and pathology and the clinical manifestation of these changes due to individual differences (e.g., education, literacy) that allow some people to cope better with changes in the brain 81.

One could also interpret the null findings of the longitudinal data as evidence that relative deficits in social support observed at baseline are maintained over time. That is, the lack of relationship between social support and yearly rate of change in neuropsychological functioning suggests that the relative deficits observed at baseline in neuropsychological functioning are maintained and enduring. This would suggest that a lack of social support has long term consequences on neuropsychological functioning. Conversely, this would suggest that strong social support has a protective effect on neuropsychological functioning.

Limitations

This work is not without limitations. First, the WHISCA and WHIMS-MRI cohorts from which we draw data were predominately (90%) white. This, and the fact that these women were volunteers who were eligible for a randomized clinical trial of postmenopausal hormone therapy, may limit the generalizability of these findings to other cohorts (e.g., males, other racial/ethnic groups). Another limitation is that only one measure of social support was collected, and therefore we are unable to consider how other aspects of social resources, such as social network, social integration, or social activity, or how the specific source of the social support (e.g., family, romantic partner, friends) may impact cognitive functioning. Another potential limitation is that it is difficult to rule out is that of reverse causality, i.e. that poorer cognitive function may lead to diminished social support.

Conclusion

The current study investigated whether perceived social support had an influence on neurological health among a large sample of healthy postmenopausal women. Social support and neuropsychological outcomes were measured annually for six years through the Women’s Health Initiative Study of Cognitive Aging. Additionally, in a subset of these women structural MRI data was collected. In this cohort of postmenopausal women, higher perceived social support was associated with significantly better overall neuropsychological functioning at baseline, especially in the domains of short-delay figural memory, short-delay verbal memory, and semantic fluency. These finding held even after adjusting for a host of relevant trial, demographic, and health status variables. Interestingly, no significant longitudinal associations were observed between social support and cognitive health among women. Moreover, no significant associations were observed between structural brain volumes and social support. Although these findings are limited to only a subset of neuropsychological domains, the current research contributes to the existing literature that points to the importance of social support as a modifiable lifestyle factor that has the potential to help protect against the decline of cognitive aging, specifically among women.

Supplementary Material

Online Appendices

Acknowledgments

Source of Funding

The WHI program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, U.S. Department of Health and Human Services through contracts HHSN268201600018C, HHSN268201600001C, HHSN268201600002C, HHSN268201600003C, and HHSN268201600004C.

Footnotes

Conflicts of Interest

No conflicts of interest to report.

Data Availability Statement

Data is available through the WHI online resource, https://www.whi.org/researchers/data/Pages/Home.aspx while the WHI remains funded (currently through 2020) and indefinitely through BioLINCC https://biolincc.nhlbi.nih.gov/studies/whi_ctos/. Eligible researchers (See https://www.whi.org/researchers/data/Pages/Home.aspx for eligibility) may download the data directly at the WHI online resource. Other researchers may download the publicly available data through BioLINCC, in accordance with NHLBI’s BioLINCC guidelines (https://biolincc.nhlbi.nih.gov/media/guidelines/handbook.pdf?link_time=2019-03-07_12:38:37.619479).

References

  • 1.Brookmeyer R, Abdalla N. Estimation of lifetime risks of Alzheimer’s disease dementia using biomarkers for preclinical disease. Alzheimers Dement. 2018;14(8):981–988. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Licher S, Darweesh SKL, Wolters FJ, et al. Lifetime risk of common neurological diseases in the elderly population. J Neurol Neurosurg Psychiatry. 2019;90(2):148–156. [DOI] [PubMed] [Google Scholar]
  • 3.Heinze JE, Kruger DJ, Reischl TM, Cupal S, Zimmerman MA. Relationships Among Disease, Social Support, and Perceived Health: A Lifespan Approach. Am J Community Psychol. 2015;56(3–4):268–279. [DOI] [PubMed] [Google Scholar]
  • 4.Nguyen AW, Chatters LM, Taylor RJ, Mouzon DM. Social Support from Family and Friends and Subjective Well-Being of Older African Americans. J Happiness Stud. 2016;17(3):959–979. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Reblin M, Uchino BN. Social and emotional support and its implication for health. Curr Opin Psychiatry. 2008;21(2):201–205. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Ertel KA, Glymour MM, Berkman LF. Effects of social integration on preserving memory function in a nationally representative US elderly population. Am J Public Health. 2008;98(7):1215–1220. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Andel R, Crowe M, Hahn Ea, et al. Work-related stress may increase the risk of vascular dementia. Journal of the American Geriatrics Society. 2012;60:60–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Hughes TF, Andel R, Small BJ, Borenstein AR, Mortimer JA. The association between social resources and cognitive change in older adults: Evidence from the Charlotte County Healthy Aging Study. Journals of Gerontology Series B-Psychological Sciences and Social Sciences. 2008;63(4):P241–P244. [DOI] [PubMed] [Google Scholar]
  • 9.Kuiper JS, Zuidersma M, Oude Voshaar RC, et al. Social relationships and risk of dementia: A systematic review and meta-analysis of longitudinal cohort studies. Ageing Res Rev. 2015;22:39–57. [DOI] [PubMed] [Google Scholar]
  • 10.Sims RC, Levy SA, Mwendwa DT, Callender CO, Campbell AL Jr. The influence of functional social support on executive functioning in middle-aged African Americans. Neuropsychol Dev Cogn B Aging Neuropsychol Cogn. 2011;18(4):414–431. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Yeh SJ, Liu Y. Influence of social support on cognitive function in the elderly. BMC Health Services Research. 2003;3(1):9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Almenar L, Zunzunegui JL, Baron G, et al. Update on heart failure, heart transplant, congenital heart disease, and clinical cardiology. Rev Esp Cardiol (Engl Ed). 2013;66(4):290–297. [DOI] [PubMed] [Google Scholar]
  • 13.Gottlieb BH, Bergen AE. Social support concepts and measures. Journal of psychosomatic research. 2010;69(5):511–520. [DOI] [PubMed] [Google Scholar]
  • 14.Taylor SE. Social support: A review. In: Friedman HS, ed. Oxford Handbook of Health Psychology. New York, NY: Oxford University Press; 2011:192–217. [Google Scholar]
  • 15.Barth J, Schneider S, Von Känel R. Lack of social support in the etiology and the prognosis of coronary heart disease: a systematic review and meta-analysis. Psychosomatic medicine. 2010;72(3):229–238. [DOI] [PubMed] [Google Scholar]
  • 16.Brummett BH, Barefoot JC, Siegler IC, et al. Characteristics of socially isolated patients with coronary artery disease who are at elevated risk for mortality. Psychosomatic medicine. 2001;63(2):267–272. [DOI] [PubMed] [Google Scholar]
  • 17.Glass TA, Matchar DB, Belyea M, Feussner J. Impact of social support on outcome in first stroke. Stroke. 1993;24(1):64–70. [DOI] [PubMed] [Google Scholar]
  • 18.Hibbard JH, Pope CR. The quality of social roles as predictors of morbidity and mortality. Social Science & Medicine. 1993;36(3):217–225. [DOI] [PubMed] [Google Scholar]
  • 19.Ikeda A, Kawachi I, Iso H, Iwasaki M, Inoue M, Tsugane S. Social support and cancer incidence and mortality: the JPHC study cohort II. Cancer Causes & Control. 2013;24(5):847–860. [DOI] [PubMed] [Google Scholar]
  • 20.Lyyra T-M, Heikkinen R-L. Perceived social support and mortality in older people. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences. 2006;61(3):S147–S152. [DOI] [PubMed] [Google Scholar]
  • 21.DiMatteo MR. Social support and patient adherence to medical treatment: a meta-analysis. Health psychology. 2004;23(2):207. [DOI] [PubMed] [Google Scholar]
  • 22.Uchino BN. Social support and health: a review of physiological processes potentially underlying links to disease outcomes. Journal of behavioral medicine. 2006;29(4):377–387. [DOI] [PubMed] [Google Scholar]
  • 23.Kiecolt-Glaser JK, Gouin J-P, Hantsoo L. Close relationships, inflammation, and health. Neuroscience & biobehavioral reviews. 2010;35(1):33–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Yang YC, Schorpp K, Harris KM. Social support, social strain and inflammation: Evidence from a national longitudinal study of US adults. Social science & medicine. 2014;107:124–135. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Schmidt R, Schmidt H, Curb JD, Masaki K, White LR, Launer LJ. Early inflammation and dementia: a 25‐year follow‐up of the Honolulu‐Asia Aging Study. Annals of neurology. 2002;52(2):168–174. [DOI] [PubMed] [Google Scholar]
  • 26.Whitmer RA, Sidney S, Selby J, Johnston SC, Yaffe K. Midlife cardiovascular risk factors and risk of dementia in late life. Neurology. 2005;64(2):277–281. [DOI] [PubMed] [Google Scholar]
  • 27.Cohen S, Wills TA. Stress, social support, and the buffering hypothesis. Psychological Bulletin. 1985;98(2):310–357. [PubMed] [Google Scholar]
  • 28.Cohen S, Syme SL. Issues in the study and application of social support. Social support and health. 1985;3:3–22. [Google Scholar]
  • 29.Wilson RS, Krueger KR, Arnold SE, et al. Loneliness and Risk of Alzheimer Disease. Arch Gen Psychiatry. 2007;64:234–240. [DOI] [PubMed] [Google Scholar]
  • 30.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;125:107–115. [DOI] [PubMed] [Google Scholar]
  • 31.Tilvis RS, Kähönen-Väre MH, Jolkkonen J, Valvanne J, Pitkala KH, Strandberg TE. Predictors of cognitive decline and mortality of aged people over a 10-year period. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences. 2004;59(3):M268–M274. [DOI] [PubMed] [Google Scholar]
  • 32.Cacioppo JT, Hawkley LC. Perceived social isolation and cognition. Trends Cogn Sci. 2009;13(10):447–454. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Fratiglioni L, Paillard-Borg S, Winblad B. An active and socially integrated lifestyle in late life might protect against dementia. Lancet Neurology. 2004;3:343–353. [DOI] [PubMed] [Google Scholar]
  • 34.Pillai JA, Verghese J. Social networks and their role in preventing dementia. Indian J Psychiatry. 2009;51:S22. [PMC free article] [PubMed] [Google Scholar]
  • 35.Fletcher E, Gavett B, Harvey D, et al. Brain volume change and cognitive trajectories in aging. Neuropsychology. 2018;32(4):436–449. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Kwak S, Joo WT, Youm Y, Chey J. Social brain volume is associated with in-degree social network size among older adults. Proc Biol Sci. 2018;285(1871). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Spagna A, Dufford AJ, Wu Q, et al. Gray matter volume of the anterior insular cortex and social networking. Journal of Comparative Neurology. 2017;526(7):1183–1194. [DOI] [PubMed] [Google Scholar]
  • 38.Kwang T, Crockett EE, Sanchez DT, Swann WB Jr. Men seek social standing, women seek companionship: sex differences in deriving self-worth from relationships. Psychol Sci. 2013;24(7):1142–1150. [DOI] [PubMed] [Google Scholar]
  • 39.Sherman SM, Cheng YP, Fingerman KL, Schnyer DM. Social support, stress and the aging brain. Soc Cogn Affect Neurosci. 2016;11(7):1050–1058. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Noonan MP, Mars RB, Sallet J, Dunbar RIM, Fellows LK. The structural and functional brain networks that support human social networks. Behav Brain Res. 2018;355:12–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Karas GB, Scheltens P, Rombouts SA, et al. Global and local gray matter loss in mild cognitive impairment and Alzheimer’s disease. Neuroimage. 2004;23(2):708–716. [DOI] [PubMed] [Google Scholar]
  • 42.Zanchi D, Giannakopoulos P, Borgwardt S, Rodriguez C, Haller S. Hippocampal and Amygdala Gray Matter Loss in Elderly Controls with Subtle Cognitive Decline. Front Aging Neurosci. 2017;9:50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Bickart KC, Wright CI, Dautoff RJ, Dickerson BC, Barrett LF. Amygdala volume and social network size in humans. Nature neuroscience. 2011;14(2):163–164. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Ehlers DK, Daugherty AM, Burzynska AZ, et al. Regional Brain Volumes Moderate, but Do Not Mediate, the Effects of Group-Based Exercise Training on Reductions in Loneliness in Older Adults. Front Aging Neurosci. 2017;9:110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Giles LC, Anstey KJ, Walker RB, Luszcz MA. Social Networks and Memory over 15 Years of Followup in a Cohort of Older Australians: Results from the Australian Longitudinal Study of Ageing. J Aging Res. 2012;2012:856048. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Fu C, Li Z, Mao Z. Association between social activities and cognitive function among the elderly in China: a cross-sectional study. International journal of environmental research and public health. 2018;15(2):231. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.La Fleur CG, Salthouse TA. Which aspects of social support are associated with which cognitive abilities for which people? Journals of Gerontology Series B: Psychological Sciences and Social Sciences. 2017;72(6):1006–1016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.McDowell I, Kristjansson B, Hill GB, Hebert R. Community Screening for Dementia: The Mini Mental State Exam (MMSE) and Modified MinieMental State Exam (3MS) Compared Journal of Clinical Epidemiology 1997;50(4):377–383. [DOI] [PubMed] [Google Scholar]
  • 49.Pillemer SC, Holtzer R. The differential relationships of dimensions of perceived social support with cognitive function among older adults. Aging Ment Health. 2015:1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.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):1221–1227. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Seeman TE, Lusignolo TM, Albert M, Berkman L. Social relationships, social support and patterns of cognitive aging in healthy, high functioning older adults: MacArthur studies of successful aging. Health psychology. 2001;20:243–255. [DOI] [PubMed] [Google Scholar]
  • 52.Shumaker Sa, Hill DR. Gender differences in social support and physical health. Health Psychology. 1991;10(2):102–111. [DOI] [PubMed] [Google Scholar]
  • 53.Matud MaP, Ibañez I, Bethencourt JM, Marrero R, Carballeira M. Structural gender differences in perceived social support. Personality and Individual Differences. 2003;35(8):1919–1929. [Google Scholar]
  • 54.Kim M-Y, Kim K, Hong CH, Lee SY, Jung Y-S. Sex differences in cardiovascular risk factors for dementia. Biomolecules & therapeutics. 2018;26(6):521. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Rossouw JE, Anderson GL, Prentice RL, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results From the Women’s Health Initiative randomized controlled trial. JAMA. 2002;288(3):321–333. [DOI] [PubMed] [Google Scholar]
  • 56.Anderson GL, Limacher M, Assaf AR, et al. Effects of conjugated equine estrogen in postmenopausal women with hysterectomy: the Women’s Health Initiative randomized controlled trial. JAMA. 2004;291(14):1701–1712. [DOI] [PubMed] [Google Scholar]
  • 57.Shumaker SA, Legault C, Rapp SR, et al. Estrogen plus progestin and the incidence of dementia and mild cognitive impairment in postmenopausal women: the Women’s Health Initiative Memory Study: a randomized controlled trial. Jama. 2003;289(20):2651–2662. [DOI] [PubMed] [Google Scholar]
  • 58.Hays J, Hunt JR, Hubbell FA, et al. The Women’s Health Initiative recruitment methods and results. Annals of epidemiology. 2003;13(9):S18–S77. [DOI] [PubMed] [Google Scholar]
  • 59.Anderson GL, Manson J, Wallace R, et al. Implementation of the Women’s Health Initiative study design. Annals of epidemiology. 2003;13(9):S5–S17. [DOI] [PubMed] [Google Scholar]
  • 60.Wittes J, Barrett-Connor E, Braunwald E, et al. Monitoring the randomized trials of the Women’s Health Initiative: the experience of the Data and Safety Monitoring Board. Clinical Trials. 2007;4(3):218–234. [DOI] [PubMed] [Google Scholar]
  • 61.Coker LH, Espeland MA, Rapp SR, et al. Postmenopausal hormone therapy and cognitive outcomes: the Women’s Health Initiative Memory Study (WHIMS). J Steroid Biochem Mol Biol. 2010;118(4–5):304–310. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Resnick SM, Coker LH, Maki PM, Rapp SR, Espeland Ma, Shumaker Sa. The Women’s Health Initiative Study of Cognitive Aging (WHISCA): a randomized clinical trial of the effects of hormone therapy on age-associated cognitive decline. Clinical Trials. 2004;1:440–450. [DOI] [PubMed] [Google Scholar]
  • 63.Sherbourne CD, Stewart AL. The MOS social support survey. Soc Sci Med. 1991;32(6):705–714. [DOI] [PubMed] [Google Scholar]
  • 64.Resnick SM, Espeland MA, Jaramillo SA, et al. Postmenopausal hormone therapy and regional brain volumes: the WHIMS-MRI Study. Neurology. 2009;72(2):135–142. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 65.Coker LH, Hogan PE, Bryan NR, et al. Postmenopausal hormone therapy and subclinical cerebrovascular disease: the WHIMS-MRI Study. Neurology. 2009;72(2):125–134. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 66.Gold R, Michael YL, Whitlock EP, et al. Race/ethnicity, socioeconomic status, and lifetime morbidity burden in the women’s health initiative: a cross-sectional analysis. J Womens Health (Larchmt). 2006;15(10):1161–1173. [DOI] [PubMed] [Google Scholar]
  • 67.Desquilbet L, Mariotti F. Dose-response analyses using restricted cubic spline functions in public health research. Stat Med. 2010;29(9):1037–1057. [DOI] [PubMed] [Google Scholar]
  • 68.van Buuren S Multiple imputation of discrete and continuous data by fully conditional specification. Stat Methods Med Res. 2007;16(3):219–242. [DOI] [PubMed] [Google Scholar]
  • 69.Rubin DB. Multiple imputation after 18+ years. Journal of the American Statistical Association. 1996;91(434):473–489. [Google Scholar]
  • 70.White IR, Carlin JB. Bias and efficiency of multiple imputation compared with complete-case analysis for missing covariate values. Stat Med. 2010;29(28):2920–2931. [DOI] [PubMed] [Google Scholar]
  • 71.Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12(3):189–198. [DOI] [PubMed] [Google Scholar]
  • 72.Paykel ES. Life events, social support and depression. Acta Psychiatr Scand Suppl. 1994;377:50–58. [DOI] [PubMed] [Google Scholar]
  • 73.Thoits PA. Mechanisms linking social ties and support to physical and mental health. J Health Soc Behav. 2011;52(2):145–161. [DOI] [PubMed] [Google Scholar]
  • 74.Brickman AM, Paul RH, Cohen RA, et al. Category and letter verbal fluency across the adult lifespan: relationship to EEG theta power. Arch Clin Neuropsychol. 2005;20(5):561–573. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 75.Meinzer M, Flaisch T, Wilser L, et al. Neural signatures of semantic and phonemic fluency in young and old adults. J Cogn Neurosci. 2009;21(10):2007–2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 76.Stuss DT, Alexander MP, Hamer L, et al. The effects of focal anterior and posterior brain lesions on verbal fluency. J Int Neuropsychol Soc. 1998;4(3):265–278. [PubMed] [Google Scholar]
  • 77.Shankar A, Hamer M, McMunn A, Steptoe A. Social Isolation and Loneliness: Relationships With Cognitive Function During 4 Years of Follow-up in the English Longitudinal Study of Ageing. Psychosomatic Medicine. 2013;75:161–170. [DOI] [PubMed] [Google Scholar]
  • 78.Salthouse TA. When does age-related cognitive decline begin? Neurobiol Aging. 2009;30(4):507–514. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 79.Brown CL, Gibbons LE, Kennison RF, et al. Social activity and cognitive functioning over time: a coordinated analysis of four longitudinal studies. J Aging Res. 2012;2012:287438. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 80.Salthouse TA. Neuroanatomical substrates of age-related cognitive decline. Psychological bulletin. 2011;137:753–784. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 81.Steffener J, Stern Y. Exploring the neural basis of cognitive reserve in aging. Biochim Biophys Acta. 2012;1822(3):467–473. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Online Appendices
NIHMS1725861-supplement-Online_Appendices.docx (84.8KB, docx)

Data Availability Statement

Data is available through the WHI online resource, https://www.whi.org/researchers/data/Pages/Home.aspx while the WHI remains funded (currently through 2020) and indefinitely through BioLINCC https://biolincc.nhlbi.nih.gov/studies/whi_ctos/. Eligible researchers (See https://www.whi.org/researchers/data/Pages/Home.aspx for eligibility) may download the data directly at the WHI online resource. Other researchers may download the publicly available data through BioLINCC, in accordance with NHLBI’s BioLINCC guidelines (https://biolincc.nhlbi.nih.gov/media/guidelines/handbook.pdf?link_time=2019-03-07_12:38:37.619479).

RESOURCES