Social Connectivity is Related to Mild Cognitive Impairment and Dementia

Abstract

Background:

Evidence supports a relationship between loneliness, social isolation, and dementia, but less is known about whether social connections confer protection against cognitive decline in disadvantaged neighborhoods.

Objective:

This longitudinal population-based study examines the relationship between social connectivity and cognitive impairment in a multi-ethnic cohort with low socioeconomic status and high vascular disease risk.

Methods:

Northern Manhattan Study participants self-reported frequency of social visits, phone calls, satisfaction with social visits, number of friends, and loneliness at baseline, and were followed prospectively with a series of neuropsychological assessments. Social connectivity was examined in relation to incident mild cognitive impairment (MCI)/dementia using logistic regression adjusting for demographics and vascular risk factors.

Results:

Among 952 participants (mean age at first neuropsychological assessment = 69±8 years, 62% women, 17% Black, 13% white, 68% Hispanic), 24% developed MCI/dementia. Participants who had phone contact with friends/family 2 + times/week (91%) had a lower odds of MCI/dementia (OR = 0.52, 95% CI = 0.31–0.89), with no association for frequency of in-person visits. Compared to those who were neither socially isolated (≥3 friends) nor lonely (reference, 73%), those who were socially isolated and lonely (3%) had an increased odds of MCI/dementia (OR = 2.89, 95% CI = 1.19–7.02), but differences were not observed for those who were socially isolated but not lonely (10%, OR = 1.05, 95% CI = 0.60–1.84), nor those who were lonely but not isolated (11%, OR = 1.58, 95% CI = 0.97–2.59).

Conclusion:

This study raises the possibility that social connections confer some protection for cognitive health in the face of adversity and supports potential opportunities for community social interventions for improving cognition in disadvantaged populations.

Keywords

Dementia depression epidemiology mild cognitive impairment social isolation

INTRODUCTION

Social isolation and loneliness, markers of poor social health, are associated with negative psychological outcomes, including depressive symptoms, sleep fragmentation, increased anxiety, vigilance, and decreased impulse control [1 –4]. Loneliness is also a risk factor for cognitive decline and the progression of Alzheimer’s disease [5, 6], recurrent stroke [7], high blood pressure [8], altered cortisol levels, and reduced cellular immunocompetency [9]. A 2016 meta-analysis showed that poor social health, as defined by social relationships, predicted cogni-tive decline across 43 studies, but there were many methodological weaknesses across these studies, including limited follow-up periods (most less than eight years), infrequent assessment of social relationships in middle age, infrequent inclusion of incident cognitive impairment as an outcomes, and inadequate control of confounding factors [10].

In contrast, individuals with strong social connectivity show distinct health advantages over the life course. Although social connectivity has been conceptualized in different ways, the classic study by Lee and Robbins in 1995 defined it in the context of belonging to a social relationship [11]. Social connectivity has been shown to reduce the risk of developing dementia [12] and mitigate cognitive deterioration over time [13, 14]. It has been suggested that social connectivity may increase cognitive reserve, improve resiliency, and promote neuronal networks [15, 16].

While these findings are compelling, most research in this area is carried out in neighborhoods characterized as largely white, educated, and advantaged. Less is known regarding the degree to which social connectivity plays a role conferring protection in the face of adversity among those with fewer resources. While we assume that social connectivity is a buffer against stress for everyone [17], there is little research examining individuals residing in neighborhoods characterized by disadvantage. A critical question is whether social connectivity can mitigate the risk of cognitive decline among individuals with vascular disease and other medical conditions residing in communities where most residents are living below the poverty line and where resources are scarce. To address this important issue, we focused on social connectivity, as defined by the number of social contacts endorsed by participants, either in person or by telephone, and examined prospectively the relationship between loneliness, social isolation, and cognition in a cohort of racially/ethnically diverse elderly individuals with limited education, a high prevalence of vascular risk factors, and economic disadvantage.

METHODS

Summary and timeline

This is an ongoing longitudinal population-based urban cohort study in which questionnaires relating to subjective self-reported loneliness, social isolation, and frequency of in-person and phone contacts were included at study baseline (1993–2001). A series of neuropsychological evaluations began in 2003, with the mean time from study baseline to the first neuropsychological evaluation 7±2 years, and the mean time from the first to the second neuropsychological evaluation 6±2 years. Adjudicated mild cognitive impairment (MCI) and dementia diagnoses during follow-up were based on data collected from these two neuropsychological evaluations. This study examines the associations between self-reported subjective loneliness and social interactions at study baseline and risk of dementia/MCI during follow-up, adjusting for sociodemographic, health behaviors and vascular risk factors.

Study population

The Northern Manhattan Study (NOMAS) is a longitudinal population-based cohort study examining the incidence and risk factors for stroke, cognitive decline, MCI, and dementia in a multi-ethnic, largely Hispanic urban population. Northern Manhattan is a well-defined, diverse, and predominantly low socioeconomic status (SES) area of New York City. Study details have been published previously [18]. From 1993 to 2001 participants were recruited who a) had never been diagnosed with a stroke; b) were 40 years old or older; and c) resided in Northern Manhattan for ≥3 months, in a household with a telephone. Eligible participants were identified by random-digit dialing (91% telephone response rate) and recruited from the telephone sample to have an in-person baseline interview and assessment (75% enrollment response rate). Between 2003 and 2008 an MRI subcohort was recruited during annual follow-up of NOMAS participants age ≥50 who were clinically stroke-free with no contraindications to MRI (N = 1,091 out of 3,298). To reach recruitment targets, 199 stroke-free household members, but not first-degree relatives, of existing NOMAS participants were enrolled for a total sample of 1,290. Following an initial evaluation with a standardized brain MRI and neuropsychological (NP) battery, this cohort has been prospectively followed with annual telephone contacts and two additional in-person NP assessments roughly six years apart. NOMAS participants are predominantly minority, lower socioeconomic (half are on Medicaid or uninsured) with low educational attainment, and a high prevalence of vascular risk factors. The study was approved by the IRBs of Columbia University and the University of Miami, and all subjects provided written informed consent.

Extensive data were collected at baseline (1993–2001) using structured interviews with trained bilingual research assistants in English or Spanish, and physical examinations with study physicians. Race and ethnicity were based upon self-identification through a series of questions modeled after the US census and conforming to standard definitions outlined by Directive 15 [19]. Vascular risk factors, including smoking, medical history, and medication use, were ascertained using standardized questions adapted from the Behavioral Risk Factor Surveillance System by the Centers for Disease Control [20]. Smoking was based on self-reported use of cigarettes, cigars, and pipes, and categorized as never, former, and current (within the past year). Leisure-time physical activity was measured with a questionnaire based on the National Health Interview Survey regarding participation in one or more selected rigorous physical activities in a typical 14-day period and dichotomized as moderate-heavy versus none-light, as previously described [21]. Moderate alcohol use was defined as current drinking of at least one drink per month and≤2 drinks per day. Baseline blood pressure was averaged from two measurements, before and after the physical examination, and fasting blood specimens were collected to determine glucose and lipid profiles. Diabetes, hypertension, and hypercholesterolemia have been defined previously [21, 22].

The Center for Epidemiologic Studies Depression Scale (CESD) was administered at the time of initial NP assessment following MRI subcohort recruitment to ascertain depressive symptoms [23]. A CESD score of 16 or higher was defined as consistent with depression.

Social connectivity

Questions related to social connectivity were included in the NOMAS baseline interview (1993–2001), including information about who participants lived with. Social isolation was defined by self-report as knowing fewer than three people well enough to visit in their home, consistent with our previous publication [24]. Loneliness was assessed by a direct single question, as participants were asked to report the frequency with which they felt lonely (quite often, sometimes, almost never), and loneliness was defined as “feeling lonely quite often” [25]. Participants were categorized into four mutually exclusive groups based on both social isolation and loneliness: socially isolated and lonely, socially isolated and not lonely, not socially isolated and lonely, and not socially isolated and not lonely (reference group).

Participants were also asked to report the number of times in the past week they talked on the telephone to someone such as a friend or relative (range 0 to≥3 times, dichotomized as < 2 versus≥2); the number of times in the past week they visited or went out with someone not living with them (not at all, 1 time, 2–6 times, once a day, dichotomized as < 2 versus≥2); and whether or not they saw friends and family as often as they wanted.

Outcome

The primary outcome of interest was adjudicated MCI or dementia at the second neuropsychological (NP) visit, as previously described [26]. The study population received comprehensive cognitive assessments over the course of follow-up, including two rounds of NP testing, annual modified Telephone Interview for Cognitive Status (TICS-m), functional assessments, and structured interviews of the participant and informant [23 , 27–40]. A Mini-Mental Status Examination (MMSE) was performed at the NOMAS study baseline assessment [41]. The first complete neuropsychological assessment was conducted on average 7±2 years after the baseline assessment, starting in 2003, and the second complete neuropsychological assessment was conducted on average 6±2 years after the first assessment. Testing was administered in English or Spanish by trained research assistants in a quiet room. Age and education-adjusted specific norms for each NP test were created, with standardized test scores and four cognitive domain specific scores (memory, language, processing speed, and executive function), as previously described [42]. Since established norms do not exist for Hispanics, data from our battery at the initial visit were used to create language, age, and education specific norms. At the initial visit we estimated premorbid intelligence with the Peabody Picture Vocabulary Test (Test de Vocabulario en Imagenes Peabody for Spanish speakers), and literacy with the Wide Range Achievement Test for English speakers and the Word Accentuation Test for Spanish speakers. We defined our algorithm to conservatively segregate those with no cognitive impairment (NCI) from those with possible cognitive impairment who then required adjudication. Adjudicators reviewed all data blindly. Data from NP visits 1 and 2 were used to determine cognitive status at the second visit. Each case was reviewed by a neurologist and a neuropsychologist and assigned a provisional diagnosis of MCI (MCI subtype)/mild neurocognitive disorder (NCD) or all-cause dementia/major NCD based on the performance on cognitive domain specific scores, depressive symptoms, medical history, and medications. Cognitive-related disability to define probable dementia and functional abilities were based on the Informant Questionnaire of Cognitive Decline in the Elderly (IQCODE) and Functional Activities Questionnaire (FAQ), respectively. Additional data, vision, and hearing status at the time of the visit, NOMAS stroke clinical and imaging data forms (if applicable), and the cognitive failures questionnaire were used to classify cognitive status using established criteria. Disagreements were resolved by consensus of the pair, with reconciliations tracked. For those cases in which consensus was not achieved, our interdisciplinary Dementia Consensus Committee (DCC) reviewed the case to reach a resolution.

Statistical analysis

Study population characteristics were compared across categories of social isolation and loneliness. A sequence of multivariable logistic regression models were constructed to examine the associations between the social connectivity variables (social isolation and loneliness, phone contacts, in-person visits) and MCI/dementia, as follows: 1) adjusted for the time from baseline to neuropsychological assessment, age, sex, education, insurance status, race/ethnicity; 2) additionally adjusted for behavioral health factors (moderate alcohol use, moderate-heavy physical activity, smoking) and living partners (alone, with spouse, other family/friend); 3) additionally adjusted for traditional vascular risk factors (BMI, hypertension, diabetes, hypercholesterolemia).

Sensitivity analyses were conducted among those without depression symptoms defined as a CESD score < 16 at the first NP assessment (N = 161 excluded), controlling for the variables in model 2. Lastly, to address the possibility of reverse causality, sensitivity analyses were conducted restricted to those without cognitive impairment at NOMAS baseline, as defined by previously published criteria of a MMSE score of 17 or higher for those with less than eight years of education, and a MMSE score of 24 or higher among those with eight or more years of education (N = 229 excluded) [43]. However, it is important to note that all participants were free of obvious dementia at study baseline.

The analyses of social isolation and loneliness as a combined construct did not adjust for the phone and in-person contact variables. The analyses of phone contact frequency were all adjusted for in-person visit frequency, and the analyses of in-person visit frequency and satisfaction were all adjusted for phone contact frequency.

Data availability

Anonymized data will be shared by request from any qualified investigator.

RESULTS

Characteristics of the study population, stratified by social isolation/loneliness, are presented in Table 1. Inclusion in this sub-cohort was unrelated to self-reported social isolation, loneliness, and frequency of social connections at baseline, suggesting that selection bias was not present.

Table 1

Characteristics of the study population

	Overall (N = 952)	Not socially isolated, not lonely (N = 731)	Socially isolated, not lonely (N = 96)	Not socially isolated, lonely (N = 100)	Socially isolated, lonely (N = 25)
Age at baseline, Mean±SD	63±8	63±8	63±8	63±7	62±9
Age at first NP assessment, Mean±SD	69±8	69±8	68±9	70±8	68±10
Age at second NP assessment, Mean±SD	75±8	75±8	74±9	75±8	73±10
Male, %	38	39	44	31	28
High school completion, %	69	66	81	78	84
Medicaid and no insurance, %	49	46	55	60	80
White, %	13	15	7	11	4
Black, %	17	18	16	7	12
Hispanic, %	68	64	74	82	84
Moderate alcohol use, %	43	46	41	31	20
Moderate-heavy physical activity, %	10	11	3	8	8
Never smoker, %	47	47	46	53	40
Former smoker, %	37	39	31	35	32
Current smoker, %	15	15	23	12	28
Normal weight, %	25	25	27	22	28
Overweight, %	45	46	31	49	36
Obese, %	30	29	42	28	36
Hypertension, %	67	67	69	62	80
Diabetes, %	19	18	21	22	32
Hypercholesterolemia, %	65	64	65	70	68
Live alone, %	25	23	27	34	20
Live with a spouse, %	43	45	46	27	32
Live with other family/friend, %	32	32	27	39	48
MCI, %	20	18	22	27	40
Dementia, %	4	4	5	4	8

Dementia adjudication was completed for 989 NOMAS participants, and social characteristics were available for 952 of these participants (96%), of whom 187 had MCI and 42 had probable dementia at visit 2. In this study population, the mean age at baseline was 63±8 years, the mean age at the first NP assessment was 69±8 years, and the mean age at the second NP assessment was 75±8 years. 62% were women, 17% were non-Hispanic Black, 13% non-Hispanic White, and 68% Hispanic. Social isolation and loneliness was self-reported by 25 participants (3% among Hispanic, 2% among Black, 1% among White participants), social isolation without loneliness by 96 participants (11% among Hispanic, 9% among Black, 5% among White participants), loneliness without social isolation by 100 participants (13% among Hispanic, 4% among Black, 9% among White participants), and neither social isolation nor loneliness by 731 participants (73% among Hispanic, 84% among Black, 85% among White participants), χ² = 22.07, degrees of freedom (DF)=9, p = 0.01. Further, 577 participants visited friends and family as frequently as desired (60% among Hispanic, 65% among Black, and 59% among White participants, χ² = 1.41, DF = 3 p = 0.70), 544 visited with people two or more times in a week (57% among Hispanic, 56% among Black, and 58% among White participants, χ² = 0.07, DF = 3, p = 1.00), and 870 had phone contact two or more times in a week (90% among Hispanic, 94% among Black, and 95% among White participants, χ² = 5.27, DF = 3, p = 0.15).

Table 2 shows the associations of the social variables with incident MCI/dementia across the sequence of multivariable adjusted logistic regression models. Frequency of in-person visits with friends or family was not associated with MCI/dementia risk in any of the models, nor was the satisfaction with the frequency of visits. However, participants who reported two or more phone conversations in a week had a significantly lower risk of MCI/dementia compared to those who reported 0–1 phone conversations, and this association remained significant across all models. In model 2 that adjusted for demographics and behavioral risk factors, participants who reported 2 or more phone conversations had a lower odds of incident MCI/dementia (OR = 0.52, 95% CI 0.31–0.89). The association between frequent phone conversations and MCI/dementia risk persisted among those who did not have depression symptoms at the time of first NP assessment (OR = 0.41, 95% CI 0.23–0.75).

Table 2

Associations between social connectivity and MCI/dementia

	Odds Ratio (95% CI) for MCI/dementia
	Model 1	Model 2	Model 3	Model 2, among those without depressive symptoms	Model 2, among those without cognitive impairment at study baseline
Not socially isolated, not lonely	REF	REF	REF	REF	REF
Socially isolated, not lonely versus Not socially isolated, not lonely	1.18 (0.70–2.00)	1.05 (0.60–1.84)	1.05 (0.60–1.83)	1.29 (0.71–2.36)	1.10 (0.54–2.22)
Not socially isolated, lonely versus not socially isolated, not lonely	1.59 (0.98–2.58)	1.58 (0.97–2.59)	1.58 (0.91–2.48)	1.21 (0.63–2.34)	1.57 (0.91–2.71)
Socially isolated, lonely versus not socially isolated, not lonely	2.97 (1.23–7.16)	2.89 (1.19–7.02)	2.87 (1.18–6.98)	2.96 (1.01–8.65)	2.99 (1.00–8.94)
See friends and relatives as often as you want (yes versus no)^a	0.89 (0.64–1.24)	0.91 (0.65–1.27)	0.89 (0.65–1.23)	1.01 (0.69–1.48)	0.85 (0.58–1.25)
Visit with friends or relatives 2 + times/week (yes versus no)^a	0.91 (0.65–1.28)	0.91 (0.65–1.28)	0.91 (0.66–1.27)	1.04 (0.71–1.54)	0.86 (0.58–1.27)
Talk on phone with friends or relatives 2 + times/week (yes versus no)^b	0.52 (0.30–0.88)	0.52 (0.31–0.89)	0.49 (0.29–0.83)	0.41 (0.23–0.75)	0.53 (0.29–0.97)

Model 1: adjusted for the time from baseline to neuropsychological assessment, age, sex, education, insurance status, race/ethnicity. Model 2: adjusted for the variables in model 1 and moderate alcohol use, moderate-heavy physical activity, smoking, living partners. Model 3: adjusted for the variables in model 2 and BMI, hypertension, diabetes, hypercholesterolemia. ^aAdditionally adjusted for frequency of phone contacts. ^bAdditionally adjusted for frequency of visits with friends or relatives.

Figure 1 shows the percentage of participants in each category of social isolation/loneliness that had MCI or dementia. Participants who were socially isolated and reported loneliness had higher odds of MCI/dementia compared to those who were not socially isolated and not lonely in all models (model 2 OR = 2.89, 95% CI 1.19–7.02), and in analyses restricted to those without depression (CESD score < 16) at the first NP assessment (OR = 2.96, 95% CI 1.01–8.65). Participants who were socially isolated but did not report loneliness did not have higher odds of MCI/dementia compared to those who were not socially isolated and not lonely (Table 2). The increased odds of MCI/dementia among participants who were not socially isolated but lonely compared to those who were neither socially isolated nor lonely also did not reach statistical significance (model 2 OR = 1.58, 95% CI 0.97–2.59, p = 0.07). The associations between the covariates in model 3 with MCI/dementia are shown in Supplementary Table 1.

Fig. 1

Percentage of participants in each category of social isolation/loneliness that had MCI or dementia.

The results remained consistent in sensitivity analyses restricted to participants without cognitive impairment at NOMAS baseline based on performance on the MMSE, controlling for the variables in model 2 (Table 2).

There was no significant relationship between living partnership and odds of MCI/dementia. Controlling for the variables in model 1 and compared to those who lived alone, participants who lived with a spouse did not have a decreased odds of MCI/dementia (OR = 0.91, 95% CI 0.58–1.42), and the increased odds of MCI/dementia among participants who lived with friends or family other than a spouse did not reach statistical significance (OR = 1.52, 95% CI 0.99–2.34).

DISCUSSION

Although social health is considered to be important in the epidemiology of dementia [44], it remains understudied, in particular among those from disadvantaged settings. This study supports the importance of including social health in prediction models for dementia. The results of the current study suggest that being both socially isolated and lonely is associated with an increased odds of developing MCI or dementia, and that frequent phone conversations with friends and family are associated with a lower odds of developing MCI or dementia, adding to the growing body of evidence showing social connectivity is important for cognitive health. Most importantly, our findings suggest that social connectivity is associated with decreased risk of incident MCI and dementia, independent of established vascular risk factors for cognitive decline, among those residing in neighborhoods characterized by disadvantage [45]. These data raise the possibility that social connections may confer some degree of protection in the face of adversity and offer potential opportunities for interventions at the community level for improving cognitive health among those with limited resources. This study also supports previous work showing an association between social networks and cognitive health among rural and urban populations in other countries including England, South Africa, and China [46 –48].

Lee and Robbin’s definition of social connectedness is defined in the context of social relationships but includes the experience of belongingness with others [11]. Therefore, social relationships and social connectedness are not interchangeable as its possible to have numerous social relationships without feeling connected. Similarly, individuals differ in how they experience isolation and loneliness. Certainly, people can feel lonely even when they have an extensive social circle and multiple connections. They can also function well in isolation without a experiencing a sense of loneliness. This study examined both of these constructs. We found that participants who did not report feeling lonely often and had at least three people that they knew well enough to visit in their home at baseline were significantly less likely to develop MCI or dementia during follow-up compared to those who reported that they felt lonely often and had a smaller social network.

An unexpected finding was the stronger association between frequency of phone contact with friends and family compared to in-person visits in relation to MCI/dementia protection. Though the data on frequency of social interactions was collected between 1993 and 2001, this analysis was conducted during the time of the COVID19 pandemic, when in-person visits are being significantly restricted and, at the same time, raising concerns about the potential health implications of recommended social isolation. The results of our analysis are encouraging as they suggest that frequent phone conversations may confer protection for cognitive health, but future studies are needed to better understand the relative importance of phone and in-person social interactions for cognitive health.

We conducted sensitivity analyses restricted to those without depression symptoms assessed by the CESD at the first neuropsychological assessment (CESD score < 16). This was included as social isolation can be both a symptom and risk factor for depression, which in turn has been shown to be an important risk factor for dementia. The associations between frequent phone contact with friends and relatives and social connectivity/lack of loneliness with MCI/dementia risk remained strong and significant even in this restricted sample. The relationship between depression, social isolation and MCI is complex: Depression can be both a risk factor and prodromal marker of dementia/MCI that can lead to social withdrawal. In addition, early dementia or MCI can result in depression and subsequent withdrawal. Social isolation can also contribute to cognitive decline, which in turn, can produce mood changes. Further analysis on these complex inter-relationships is needed in studies with longitudinal data. The results of our sensitivity analysis suggest that social isolation is a relevant and modifiable risk factor for MCI/dementia even among those who are not depressed.

Although there is a lack of consensus with regard to which specific criteria most accurately characterizes a disadvantaged neighborhood, our study cohort mirrors the base population, a region in which the majority of neighborhoods have over a quarter of residents living below the federal poverty line [46]. In addition, 31% did not complete high school and 51% lack medical insurance. The results of the current study suggest that social connectivity might represent an impactful and relatively low cost intervention focus in disadvantaged communities at high risk for dementia. The findings support the need for trials to evaluate interventions focused on increasing social engagement in high risk communities to reduce the risk of cognitive impairment. They also underscore the importance of addressing feelings of loneliness and social isolation by primary care physicians and community health workers as part of a routine visit. We acknowledge that studying the number of social connections does not speak directly to the quality of social support. Dementia has a major impact on public health, affecting 14% of the population older than 71, with care estimated to be $305 billion annually, reaching $511 billion by 2040 [49 –51]. Therefore, efforts targeting modifiable risk factors to promote cognitive health remain a high priority, with specific emphasis on promoting deeper social support systems.

The current study includes several strengths lacking in the existing literature, including over 25 years of longitudinal follow-up, formal committee adjudication of both MCI and dementia status, and adjustment for a broad range of important potential confounding variables. Dementia adjudication was completed using a second NP assessment for 85% of study participants who remained alive. The fact that we were able to include participants who had been formally diagnosed with MCI, a group in which there is little information pertaining to the relationship between mild cognitive impairment, social isolation, and loneliness, is a major methodological advantage, as most study samples do not have a formal adjudication process for classification of cognitive impairment. Additional strengths of the current study include the comprehensive neuropsychological assessments and low attrition follow-up from the time of baseline to the present. Another important and unique strength includes the large number of diverse participants who are “aging in place” and characterized as carrying the highest disease burden risk, allowing us to prospectively study incident dementia and MCI within a racially and ethnically diverse urban disadvantaged neighborhood. The prospective nature of the current study is important to note as the social variables were collected at NOMAS baseline when all participants were free of apparent dementia, several years prior to the first neuropsychological assessment, though we are unable to infer causality from this observational study. The consistent conclusions in sensitivity analyses restricted to those without any cognitive impairment as measured by the MMSE at baseline suggests that reverse causality is an unlikely source of bias in the current study. Limitations include insufficient power to examine dementia alone as an outcome, and the collection of data on loneliness and social connectivity at a single time point limiting our ability to examine any changes in social behaviors and loneliness over follow-up in relation to cognitive health. The questions about the frequency of phone and in-person visits referred to the week prior to the baseline interview. It is possible that the social variables self-reported at study baseline were not representative of long-term behavior. In addition, social isolation and loneliness were based on single questions rather than standardized and validated instruments. Participants who were included in the analysis were not significantly different from those who were not included, suggesting that selection bias is an unlikely threat to the validity of the current analysis.

In conclusion, this study highlights the importance of social connectivity in promoting cognitive health among those residing in disadvantaged neighborhoods. Further, our results suggest that individual and community interventions aimed at improving social cohesion in communities could have a positive cognitive health impact, but further research is needed to better understand which interventions may be impactful. Future studies are needed to identify the most sensitive time points during which social isolation may be most etiologically relevant, and the complex interrelationships with other lifestyle factors including physical activity, diet, and sleep.

Footnotes

ACKNOWLEDGMENTS

This work was supported by the National Institutes of Health/National Institute of Neurological Disorders and Stroke (R01 NS 29993).

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Cacioppo

, Hughes

, Waite

, Hawkley

, Thisted

(2006) Loneliness as a specific risk factor for depressive symptoms: Cross-sectional and longitudinal analyses. Psychol Aging 21, 140–151.

Kurina

, Knutson

, Hawkley

, Cacioppo

, Lauderdale

, Ober

(2011) Loneliness is associated with sleep fragmentation in a communal society. Sleep 34, 1519–1526.

Yanguas

, Pinazo-Henandis

, Tarazona-Santabalbina

(2018) The complexity of loneliness. Acta Biomed 89, 302–314.

Hawkley

, Cacioppo

(2010) Loneliness matters: A theoretical and empirical review of consequences and mechanisms. Ann Behav Med 40, 218–227.

Lara

, Caballero

, Rico-Uribe

, Olaya

, Haro

, Ayuso-Mateos

, Miret

(2019) Are loneliness and social isolation associated with cognitive decline? Int J Geriatr Psychiatry 34, 1613–1622.

Wilson

, Krueger

, Arnold

, Schneider

, Kelly

, Barnes

, Tang

, Bennett

(2007) Loneliness and risk of Alzheimer disease. Arch Gen Psychiatry 64, 234–240.

Cacioppo

, Capitanio

, Cacioppo

(2014) Toward a neurology of loneliness. Psychol Bull 140, 1464–1504.

Xia

, Li

(2018) Loneliness, social isolation, and cardiovascular health. Antioxid Redox Signal 28, 837–851.

Kiecolt-Glaser

, Ricker

, George

, Messick

, Speicher

, Garner

, Glaser

(1984) Urinary cortisol levels, cellular immunocompetency, and loneliness in psychiatric inpatients. Psychosom Med 46, 15–23.

10.

Kuiper

, Zuidersma

, Zuidema

, Burgerhof

JGM

, Stolk

, Oude Voshaar

, Smidt

(2016) Social relationships and cognitive decline: A systematic review and meta-analysis of longitudinal cohort studies. Int J Epidemiol 45, 1169–1206.

11.

Lee

, Robbins

(1995) Measuring belongingness: The Social Connectedness and the Social Assurance scales. J Counsel Psychol, 42, 232–241.

12.

Umberson

, Montez

(2010) Social relationships and health: A flashpoint for health policy. J Health Soc Behav 51, S54–S66.

13.

Fratiglioni

, Paillard-Borg

, Winblad

(2004) An active and socially integrated lifestyle in late life might protect against dementia. Lancet Neurol 3, 343–353.

14.

Atti

, Forlani

, De Ronchi

, Palmer

, Casadio

, Dalmonte

, Fratiglioni

(2010) Cognitive impairment after age 60: Clinical and social correlates in the Faenza Project. J Alzheimers Dis 21, 1325–1334.

15.

, Hasselgren

, Zettergren

, Zetterberg

, Blennow

, Skoog

, Halleröd

(2020) The impact of social networks and APOE ɛ4 on dementia among older adults: Tests of possible interactions. Aging Ment Health 24, 395–404.

16.

Bassuk

, Glass

, Berkman

(1999) Social disengagement and incident cognitive decline in community-dwelling elderly persons. Ann Intern Med 131, 165–173.

17.

Cohen

, Wills

(1985) Stress, social support, and the buffering hypothesis. Psychol Bull 98, 310–357.

18.

Sacco

, Anand

, Lee

, Boden-Albala

, Stabler

, Allen

, Paik

(2004) Homocysteine and the risk of ischemic stroke in a triethnic cohort: The NOrthern MAnhattan Study. Stroke 35, 2263–2269.

19.

Wallman

, Hodgdon

(1977) Race and ethnic standards for Federal statistics and administrative reporting. Stat Report, pp. 450-454.

20.

Kargman

, Sacco

, Boden-Albala

, Paik

, Hauser

, Shea

(1999) Validity of telephone interview data for vascular disease risk factors in a racially mixed urban community: The Northern Manhattan Stroke Study. Neuroepidemiology 18, 174–184.

21.

Willey

, Moon

, Paik

, Boden-Albala

, Sacco

, Elkind

(2009) Physical activity and risk of ischemic stroke in the Northern Manhattan Study. Neurology 73, 1774–1779.

22.

Willey

, Gardener

, Cespedes

, Cheung

, Sacco

, Elkind

MSV

(2017) Dietary sodium to potassium ratio and risk of stroke in a multiethnic urban population: The Northern Manhattan Study. Stroke 48, 2979–2983.

23.

Radloff

(1977) Center for Epidemiological Studies Depression Scale. Appl Psychol Meas 1, 385–401.

24.

Rodriguez

, Elkind

, Clemow

, Jin

, Di Tullio

, Sacco

, Homma

, Boden-Albala

(2011) Association between social isolation and left ventricular mass. Am J Med 124, 164–170.

25.

Shiovitz-Ezra

, Ayalon

(2012) Use of direct versus indirect approaches to measure loneliness in later life. Res Aging 34, 572–591.

26.

Wright

, DeRosa

, Moon

, Strobino

, DeCarli

, Cheung

, Assuras

, Levin

, Stern

, Sun

, Rundek

, Elkind

MSV

, Sacco

(2021) Race/ethnic disparities in mild cognitive impairment and dementia: The Northern Manhattan Study. J Alzheimers Dis 80, 1129–1138.

27.

Broadbent

, Cooper

, FitzGerald

, Parkes

(1982) The Cognitive Failures Questionnaire (Cfq) and Its Correlates. Br J Clin Psychol 21, 1–16.

28.

D’Elia

, Satz

, Uchiyama

, White

(1994) Color Trails Test Professional Manual. Psychological Assessment Resources, Odessa.

29.

Beery

(1989) The VMI, Developmental Test of Visual Motor Integration, Modern Curriculum Press, Cleveland, OH.

30.

Ruff

, Parker

(1993) Gender- and age-specific changes in motor speed and eye-hand coordination in adults: Normative values for the Finger Tapping and Grooved Pegboard Tests. Percept Mot Skills 76, 1219–1230.

31.

Weschler

(1997) Wechsler Adult Intelligence Scale - Third Edition, Psychological Corporation.

32.

Hoppe

, Muller

, Werheid

, Thone

, von Cramon

(2000) Digit Ordering Test: Clinical, psychometric, and experimental evaluation of a verbal working memory test. Neuropsychol 14, 38–55.

33.

Goodglass

, Kaplan

(1983) The Assessment of Aphasia and Related Disorders, Lea & Febiger, Philadelphia.

34.

Benton

A HK

, Sivan

(1994) Manual for the Multiligual Aphasia Examination, 3rd Ed., AJA Associates, Inc., Iowa City.

35.

Smith

(1982) Symbol Digit Modalities Test: Manual. Western Psychological Services.

36.

Dunn

L DD

1997) Peabody Picture Vocabulary Test. American Guidance Service, Circle Pines.

37.

Jastak

M, S

, Wilkinson

(1995) Wide Range Achievement Test. Rasch Measure Trans 8, 403.

38.

Del Ser

, Gonzalez-Montalvo

, Martinez-Espinosa

, Delgado-Villapalos

, Bermejo

(1997) Estimation of premorbid intelligence in spanish people with the Word Accentuation Test and its application to the diagnosis of dementia. Brain Cogn 33, 343–356.

39.

Morris

(1993) The Clinical Dementia Rating (CDR): Current version and scoring rules. Neurology 43, 2412–2414.

40.

Jorm

(1994) A short form of the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE): Development and cross-validation. Psychol Med 24, 145–153.

41.

Folstein

, Folstein

, McHugh

(1975) “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

42.

Gardener

, Caunca

, Dong

, Cheung

, Rundek

, Elkind

MSV

, Wright

, Sacco

(2020) Obesity measures in relation to cognition in the Northern Manhattan Study. J Alzheimers Dis 78, 1653–1660.

43.

Gambhir

, Khurana

, Kishore

, Sinha

, Mohapatra

(2014) A clinico-epidemiological study of cognitive function status of community-dwelling elderly. Indian J Psychiatry 56, 365–370.

44.

Vernooij-Dassen

, Jeon

(2016) Social health and dementia: The power of human capabilities. Int Psychogeriatr 28, 701–703.

45.

Lavery

(2020) Socioeconomic status (NSES index) by census tract, 2011-2015. https://www.arcgis.com/home/item.html?id=2a98d90305364e71866443af2c9b5d06, Last updated June 26, 2021, Accessed on July 16, 2021.

46.

Harling

, Kobayashi

, Farrell

, Wagner

, Tollman

, Berkman

(2020) Social contact, social support, and cognitive health in a population-based study of middle-aged and older men and women in rural South Africa. Soc Sci Med 260, 113167.

47.

Kelly

, Duff

, Kelly

, McHugh Power

, Brennan

, Lawlor

, Loughrey

(2017) 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 6, 259.

48.

Read

, Comas-Herrera

, Grundy

(2020) Social isolation and memory decline in later-life. J Gerontol B Psychol Sci Soc Sci 75, 367–376.

49.

(2020) 2020 Alzheimer’s disease facts and figures. Alzheimers Dement 16, 391–460.

50.

Plassman

, Langa

, Fisher

, Heeringa

, Weir

, Ofstedal

, Burke

, Hurd

, Potter

, Rodgers

, Steffens

, Willis

, Wallace

(2007) Prevalence of dementia in the United States: The aging, demographics, and memory study. Neuroepidemiology 29, 125–132.

51.

Hurd

, Martorell

, Langa

(2013) Monetary costs of dementia in the United States. N Eng J Med 369, 489–490.