Background

Maintaining cognitive function is crucial for healthy aging [1,2,3]. Therefore, identifying and exploring modifiable risk or protective factors for cognitive function are key foci of aging research [4]. Social support is an important modifiable protective factor for cognitive function [5,6,7,8].

Structural social support is a quantifiable measure of social relationships, such as the number of people in one’s social network or the degree of participation in social events. Functional social support is the extent to which an individual perceives their needs can be met by members of their social network, such as the availability of someone to drive them to the doctor or help with grocery shopping, if required [9, 10].

Multiple reviews reported that large social networks and frequent engagement with these networks promote cognitive stimulation and protect against cognitive decline [11,12,13,14]. However, the literature has devoted less attention to functional social support and cognitive function, even though functional support more accurately represents the depth and quality of social support experienced by individuals than structural support [9].

Kelly et al. reviewed the association between functional social support and cognitive function in nine longitudinal studies of healthy older adults [15]. They reported variability in the direction and magnitude of the association, depending on the measures of functional support and cognitive function. Since Kelly et al.’s review [15], additional literature [6, 7, 16, 17] has emerged on the topic, underlining the need for an updated review.

We conducted this systematic review to investigate the association between functional social support and cognitive function across multiple cognitive domains (i.e., memory, executive function) and cognitive disease states (i.e., mild neurocognitive disorder, major neurocognitive disorder) in middle-aged and older adults. Our review focused exclusively on functional social support, reflecting Menec et al.’s conceptual distinction between objective (structural) and subjective (functional) social relationships: one may report many social contacts yet believe most will not help in times of need, or vice versa [18]. Importantly, this review differs from Costa-Cordella et al.’s recently published review [19], which included articles on structural and functional social support without age restrictions and excluded articles on neurological conditions characterized by cognitive deficits (e.g., mild or major neurocognitive disorder).

Methods

Our review followed the 2020 Preferred Reporting Items for Systematic Reviews and Meta Analyses (PRISMA) guidelines [20] (Additional file 1). We departed slightly from our published protocol [4] and did not conduct a meta-analysis or formally assess publication bias, nor did we narratively synthesize the extracted data by sex, setting, or risk of bias level. These proposed undertakings were precluded by heterogeneity in definitions and measures of functional social support and cognitive function, as well as by multiple different means of reporting quantitative results in the included articles.

Data sources and searches

We searched PubMed, PsycINFO, Sociological Abstracts, CINAHL and Scopus from inception to September 2021. Google Scholar was searched to retrieve grey literature. A medical librarian generated the syntax for PubMed (Additional file 2), which was adapted for the other databases.

Eligibility criteria

The review included any study with a comparison group (e.g., cohort, cross-sectional, case–control) enrolling adults aged ≥ 40 years, regardless of residential setting (e.g., community, long-term care facility). Articles had to be published in English or French and report distinct results for persons in the age range of interest. The exposure was functional social support, sometimes called ‘perceived social support’ or ‘social support availability’, and the outcome was cognitive function. Included articles could assess global/overall functional social support or a subtype, such as emotional/informational support, tangible support, affectionate support, positive social interaction, using any tool or questionnaire. Similarly, the articles could measure cognitive function globally or by domain (e.g., memory, executive function) with any instrument or combination of tools (neuropsychological battery). We also included studies of neurological conditions characterized by cognitive deficits (e.g., mild or major neurocognitive disorder).

In line with the PICOS (population, intervention, comparator, outcome, and setting) framework, we present the inclusion criteria as follows:

  • P = Adults aged 40 years or over from any residential setting, including those residing in the community or independent-living older age homes, or persons residing in institutionalized settings such as long-term care facilities;

  • I = Any level of exposure to functional social support, defined broadly as one’s perception of the amount of help they would expect to receive from members of their social network in times of need;

  • C = A different level of functional social support relative to ‘I’ above, e.g., comparing persons with lower scores on a social support scale (C) to persons with higher (better) scores on the scale (I);

  • O = Any measure of differences between I and C, such as differences in cognition scale score or differences in the incidence or prevalence of a neurological condition; and

  • S = Study conducted anywhere in the world and in any setting.

We excluded articles that did not assess any form of functional social support, cognitive function, or neurological condition with cognitive deficits. We also excluded articles that did not include comparison groups or articles published in languages other than English or French.

Study selection, data extraction and risk of bias assessment

Following removal of duplicates, two reviewers used the eligibility criteria and Covidence software (Veritas Health Innovation, Melbourne, Australia) to independently screen the titles/abstracts and full texts of identified citations. Two reviewers independently extracted the following data from included articles into a prepared Excel spreadsheet: first author, year of publication, country of data collection, proportion female, setting, length of follow-up, type and measure of social support, type and measure of cognitive function, and outcomes. Reviewers extracted outcome data in the form reported by authors. Where possible, extracted data came from fully adjusted regression models. Two independent reviewers assessed risk of bias using the Newcastle–Ottawa Scale (NOS) [21]. In all cases, discrepancies between reviewers were resolved by consensus or a third reviewer.

Synthesis methods

The extracted data were narratively synthesized in groups based on cognitive outcome, study design, and functional social support subtype. Studies of visuospatial skills or reasoning were classified under executive function; those of verbal memory, non-verbal memory, working memory, or episodic memory were classified under memory; and those of attention or processing speed were placed in their own unique category. We followed the Synthesis Without Meta-Analysis (SWiM) guidelines to conduct a narrative synthesis [22] and reported the effect measures contained in the included articles.

Results

Study characteristics

Our search yielded 2,976 articles and 85 of these articles, published between 1986 and 2021, were included in the review (Fig. 1). Of these 85 articles, 44 were cross-sectional and 41 were cohort studies, with sample sizes ranging from 20 to 30,029 (Table 1). Most samples included community-dwelling persons, but four studies exclusively enrolled persons in institutionalized settings [23,24,25,26]. Nineteen articles examined dementia due to Alzheimer's disease (AD) or all-cause dementia, 38 examined global cognitive functioning or general cognitive impairment or decline, and 20 examined specific cognitive domains. Sixty-two articles reported multiple subtypes of functional social support. Common control variables were age, sex, race, education, income, social network, marital status, activities of daily living (ADLs), depression, and chronic conditions such as diabetes, cardiovascular disease, and hypertension. Most articles had low risk of bias (Table 2; Fig. 2). Overall, functional social support was protective against cognitive outcomes (Fig. 3).

Fig. 1
figure 1

PRISMA Flow Diagram

Full size image
Table 1 Study characteristics
Full size table
Table 2 Overall risk of bias ratings
Full size table
Fig. 2
figure 2

Risk of Bias

Full size image
Fig. 3
figure 3

Count of Reported Associations between Functional Social Support and Cognition-related Outcomes in the Narrative Synthesis

Full size image

Narrative synthesis

Alzheimer’s disease or all-cause dementia

Cross-sectional studies

Four of the five cross-sectional studies reported on dementia, while the remaining study reported results for AD and non-AD dementia (Table 3). Four studies focused on functional social support, two of which reported no association with dementia. One found greater functional social support to be significantly associated with lower severity of dementia. One reported this support as being a moderate protective factor against AD, but a small risk factor for non-AD dementia. One study found that all-cause dementia was associated with lower satisfaction with diffuse social relationships, but not with close social relationships [29, 37, 42, 52, 57].

Table 3 Studies reporting outcome of Alzheimer’s Disease or dementia
Full size table

Cohort studies

Nine of 14 cohort studies reported an outcome of all-cause dementia, four studies reported outcomes of AD and non-AD dementia independently, and one study reported an outcome of only AD (Table 3). Eight studies explored the effects of emotional social support, six of which found small to moderate protective effects against dementia (one reached statistical significance). One observed a small protective effect in both male and female strata. Two studies reported small positive, but not statistically significant, associations between emotional support and all-cause dementia. Two of the eight studies found moderate protective effects for emotional social support against AD [16, 25, 71, 77, 81, 85, 87, 92,93,94,95].

Four studies assessed instrumental social support, one of which reported a large positive association with both AD and non-AD dementia (statistically significant in the case of AD). Another study found small protective effects against dementia in both male and female participants. One study found that individuals identified as having increasing dementia were more likely to fall within the low instrumental support group. One study found no association [16, 77, 92, 93].

Two studies found satisfaction with social support to have moderate protective effects against dementia, with one being statistically significant. One of these also found satisfaction to have a moderate and nonsignificant protective effect against AD [25, 67]. Khondoker et al. reported positive social support had small protective effects against dementia [81]. Andel et al. showed workplace social support was protective against AD and non-AD dementia (statistically significant for non-AD) [68].

Global cognitive functioning

Cross-sectional studies

Three cross-sectional studies examined participant satisfaction with functional social support and global cognitive function (Table 4). Two reported positive yet statistically non-significant associations, and one found no association [27, 34, 40].

Table 4 Studies reporting outcome of global cognitive functioning
Full size table

Twelve cross-sectional studies explored the association between perceived or subjective functional social support and global cognitive function, with 11 reporting positive associations (10 statistically significant), and one reporting a negative association (Table 4). One study observed significant positive effects among females only. One reported that support from a wife was positively associated with a husband’s cognitive function, but not vice versa. One observed a positive association for spouse-provided support, but not support from children, friends, and extended family. One found links between greater subjective cognitive decline and greater levels of perceived social support [6, 23, 24, 29, 43, 45, 46, 54, 58, 61, 63, 66].

Eight studies assessed the association between emotional social support and global cognitive function; authors reported positive associations in all eight, with seven reaching statistical significance. Six studies explored the effect of instrumental social support on cognitive function and two found statistically significant positive associations, one found a non-significant positive association, one found no association, one reported a small (non-significant) negative association, and one found positive associations in male (significant) and female (non-significant) strata. Three studies assessed the combined effects of emotional and instrumental social support on global cognitive function and found significant positive associations [30, 32, 33, 36, 38, 41, 47,48,49,50,51, 56].

Rashid et al. assessed general functional social support and observed that individuals with lower reported levels of support were at an increased risk of cognitive impairment [53]. Jang et al. used family solidarity as a measure of functional social support and found no association between this variable and cognitive function [39].

Cohort studies

One study found a positive association between functional social support and global cognitive function. Nine other studies assessed the association between perceived / subjective social support and global cognitive function, with six reporting positive associations, four of which were significant. One reported a negative association for Black people and a positive association for White people, although neither was significant. One showed a negative association for support from the family and a positive association for support from friends, with neither being statistically significant. One found perceived social support to be significant positively associated with cognitive function in persons whose cognition test scores were rapidly declining but found no association when scores were slowly declining or stable [26, 70, 73, 75, 78,79,80, 86, 88, 91].

Nine other cohort studies assessed the impact of emotional social support on global cognitive function. Three reported positive associations, one of which was significant. Two studies reported negative associations, neither of which was significant. In one study, emotional social support received from participants’ children was inversely associated with cognitive function. Similarly, inverse associations were found in male and female strata, though neither was statistically significant. One study identified significant protective effects for emotional support in persons whose baseline cognition was low and declining over time, and non-significant protective effects in those with high and declining cognition, compared to individuals with high and stable cognition [17, 38, 69, 72, 76, 79, 89, 96, 98].

Eight cohort studies explored instrumental social support and global cognitive function. Six studies reported positive associations, one of which was statistically significant. Three found non-significant negative associations. One study assessed the combined effects of emotional and instrumental social support, stratified by the source of support (co-residing family, non-residing family and relatives, neighbours and friends), and reported significant positive associations in the neighbours and friends stratum; the associations in the other two strata were inverse and non-significant [17, 69, 74, 76, 79, 89, 96, 100].

Studies reporting outcomes by cognitive domain

Twenty-seven studies examined the effects of functional social support on one or more specific cognitive domains (Table 5). Most studies assessed multiple domains, with 17 studies examining memory, 13 executive function, 3 attention and processing speed, 4 language ability, and 3 mild cognitive impairment (MCI).

Table 5 Studies Reporting other cognitive outcomes
Full size table

Memory

Cross-Sectional Studies. Ten cross-sectional studies explored the association between functional social support and memory. One found a positive, non-significant association for satisfaction with available support. Two of five studies reported positive and statistically significant associations between perceived social support and memory. Two reported positive associations between perceived support and verbal memory, with the only statistically significant association involving memory measured longitudinally. They also found negative and non-significant associations between perceived support and working memory at both time periods, and a positive and significant association between perceived support and visual memory measured longitudinally. One found a significant association between lower perceived social support and greater problems with memory or forgetfulness [34, 35, 43, 64, 65, 90].

Four studies examining emotional social support and memory reported positive associations, with results in three achieving statistical significance. One found the association between emotional support and verbal memory to be mediated by hippocampal volume, one reported similar strengths of association for immediate and delayed recall memory, and one found positive associations of the same magnitude for working and episodic memory [7, 41, 44, 59].

Three studies assessed the effects of instrumental social support on memory: one reported a statistically significant negative association with general memory [55]; one found a small and non-significant negative association with overall memory [44]; and one identified a small positive and non-significant association with working memory and a small negative and non-significant association with episodic memory [59]. Finally, Oremus et al. found positive social interactions and affectionate support to be independently and positively associated with immediate and delayed recall memory (statistically significant for affectionate support) [7].

Cohort Studies. Two studies of perceived support and memory found either no association [26] or statistically significant and positive associations with both working and episodic memory [60]. Liao and Scholes found a positive and statistically significant association between positive social support and global memory [84]. Hughes et al. found a negative association in the case of emotional support, and positive associations for instrumental support and satisfaction with social support [79]. Zahodne et al. found positive and negative associations, respectively, between emotional and instrumental support, and working memory; they also observed negative associations between emotional support and episodic memory, and no association between instrumental support and episodic memory [102].

Executive function

Cross-Sectional Studies. Gow et al. reported a positive and non-statistically significant association between participant satisfaction with functional social support and executive function, although they did not provide any numerical findings [34]. Hamalainen et al. and Krueger et al. reported positive and statistically significant associations between perceived social support and executive function [35, 43].

Three of four cross-sectional studies found positive associations between emotional social support and executive function, two of which were statistically significant. One study stratified by individual sources of emotional support and only spousal support remained statistically significantly associated with executive function. One study observed a statistically significant negative association [28, 31, 44, 59].

Three cross-sectional studies assessed the independent effect of instrumental social support on executive function: La Fleur and Salthouse found a small yet non-significant positive association, Zahodne et al. observed no association, and Bourne et al. reported a statistically significant negative association [28, 44, 59]. Ge et al. evaluated combined emotional and instrumental support on executive function and reported a statistically significant positive association [32].

Cohort Studies. Five cohort studies evaluated the effect of functional social support on executive function. Dickinson et al. and Liao & Scholes found positive and statistically significant associations for instrumental and positive support [74, 84]. Zahodne et al. showed a positive, but non-significant, association for emotional support and a negative, non-significant association for instrumental support [102]. Liao found no association for confiding support, and Hudetz et al. showed no significant association between perceived social support and post-operative executive function [26, 83].

Other cognitive domains

(Table 5). La Fleur and Salthouse’s cross-sectional study found a positive association between instrumental support and language ability, and a stronger and statistically significant association between emotional support and language ability [44]. Three cohort studies reported mixed results of no [26], positive [60], or negative associations (the latter being non-statistically significant) with language ability [102].

Two cross-sectional studies and one cohort study measured attention or processing speed. The cross-sectional studies reported positive associations for perceived social support [64, 65], with the former reporting a statistically significant result. The cohort study found no association for instrumental support, a positive association for emotional support, and a larger positive association with satisfaction with social support [79].

Three cross-sectional studies found slight protective effects between perceived/overall support and conversion to MCI [42, 52, 62]. One cohort study observed that negative social interaction was a risk for MCI [99].

Discussion

Overall, functional social support was positively associated with cognitive function in middle- and older-aged adults (Fig. 3). However, the results were not uniform across the 85 included studies.

Overall functional social support

Individual perceptions of functional social support did not appear to be associated with a diagnosis of AD or all-cause dementia. Conversely, perceived support was most often positively associated with improved cognitive function, although these associations did not always reach statistical significance. Negative associations, or a lack of association, were sometimes observed in the context of male participants or family members as the only sources of perceived social support [45, 70]. The negative association observed for male participants could suggest that males and females experience social support differently and emphasizes distinct aspects of the quality of social relationships. Social support from family members may be inversely associated with cognition because tumultuous intra-family relations could lead to psychosocial stress.

Emotional social support

Most studies involving a clinical diagnosis of AD or all-cause dementia reported non-significant negative associations between emotional social support and these outcomes. Most of these studies also found significant and positive associations with both global and domain-specific cognitive function. However, negative associations or absence of any association were sometimes observed when considering emotional support provided by family members [79, 89]. Individuals in need of strong emotional support from their co-residing family members might concomitantly be experiencing some form of family-based physical or psychological stressors that negatively affect cognition.

Instrumental social support

In contrast to the findings with perceived or emotional support, an equal number of studies observed positive and negative associations between instrumental support and AD or all-cause dementia. Most studies reported non-significant positive associations between instrumental support and domain-specific cognitive outcomes, although several studies in this group found an inverse association. For global cognitive function, an approximately equal number of studies reported positive and negative associations. The number of studies with negative associations was larger in the case of instrumental support compared to perceived and emotional support. Perhaps these findings merely reflect the increased need for functional support in day-to-day life among people with dementia, which can be partially provided by instrumental social support.

Emotional-instrumental social support, satisfaction with social support

Most studies that assessed the combined effects of emotional and instrumental support reported positive associations with global and domain-specific cognitive function. All studies that assessed participant satisfaction with functional social support found protective effects against both AD and global dementia. All articles that measured domain-specific cognitive outcomes found satisfaction with social support to be non-significantly positively associated with cognition. Reported satisfaction with social support was also positively associated with global cognition in most cases.

Positive, affectionate, confiding social support

Five studies examined positive, affectionate or confiding types of support [7, 51, 81, 83, 84]. Receiving positive social support was associated with a decreased risk of dementia, as well as improved global cognition and memory. Similarly, affectionate social support was associated with decreased risk of dementia and improved memory. One study explored the effects of confiding support on executive function and reported no association between the two variables.

Domain-specific cognitive outcomes

Memory was the most frequently assessed, domain-specific cognitive outcome. In most cases, functional social support was positively associated with memory. The same results were found with executive function. Turning to the domains of language and attention/processing speed, all studies reported either no association or a positive association. Some studies used a clinical diagnosis of MCI as the cognitive outcome and found functional social support acted as a protective factor, whereas negative social interaction served as a risk factor.

Strengths and limitations

A self-assessment with AMSTAR2 (Additional file 3) showed the quality of our systematic review was strong [103]. Our comprehensive search strategy captured many articles across a spectrum of functional social support exposures and cognitive outcomes. The nature of the exposure prevented us from looking at randomized controlled trials. One of the included articles was at high risk of bias and the narrative synthesis was facilitated by the similarity of covariate sets in the included articles.

Our review is unique from Kelly et al. [15] and Costa-Cordella et al. [19] because it focused exclusively on functional social support. Further, our review contained the most up-to-date synthesis of the literature on the topic. The adverse impact of the COVID-19 pandemic on social engagement, especially among older adults, provides a renewed impetus to understand how functional social support affects the cognitive health and well-being of aging populations.

Conclusions

The findings of this review show that functional social support may act as a protective factor against dementia and cognitive decline. This association appears to be stronger in the case of overall and emotional support, relative to instrumental support. Policy makers may wish to allocate public funds for community-based programs centered on fostering quality social relationships high in emotional support among middle-aged and older adults.