Table 2 Characteristics of studies: observational longitudinal studies

Study

Participant

Years FU

Social activity defined/measure

Cognitive outcome measures

Results

Aartsen (2002)

[30]

Netherlands (LAS)

N = 2076

Age: 55–85

M Age = 68.7

6 years

Social activity: activities, social status, and service, incl. Visiting church, visiting neighbourhood associations, attending organisation for helping older adults, neighbours, or disabled persons.

Cognitive function (MMSE)

Episodic Memory (Immediate Recall, AVLT: 15 Words Test)

Reasoning (Raven’s Coloured Progressive Matrices)

Processing speed (Coding Task)

No association between indicators of social activity and cognitive functioning.

Andrew (2010)

[78]

Canada (CSHA)

N = 2468

Age: 70+

M Age = 79.1

5 years

Social support: self-report potential social deficits - living situation, social support, family relationships, friendships, etc.

Cognitive function (Modified MMSE: 3MS) ^c

Each added baseline social deficit was associated with increased odds of cognitive decline (OR, 1.03; 95% CI, 1.00 to 1.06; p = .02, N = 2391).

Compared to low social vulnerability, high social vulnerability led to 36% increased odds of cognitive decline (OR, 1.36; 95% CI, 1.06 to 1.74).

Barnes (2004)

[17]

USA (CHAP)

N = 6102

Age: 65+

5.2 years

Social networks: number of children, relatives, and friends seen at least once a month.

Social activity: 4 items related to social and productive activity.

Cognitive function (composite including measures of episodic memory, perceptual speed, and MMSE) ^c

Greater social networks and social activity positively correlated with initial level of cognitive function (both p < 0.001).

Rate of global cognitive decline reduced by 39% for 16 social ties versus 1 social tie; and reduced by 91% for a max score of 8 versus a score of 0 on the social engagement (activity) scale.

Beland (2005) [61]; Zunzunegui (2003) [74]

Spain

(Ageing in Legane’s)

N = 1571

Age: 65–100

M Age = 75.6

7 years

Social activity: frequency of attending community associations, religious services, recreational activities.

Social network: no. relatives seen monthly, presence of friends

Social support: how often helped children, family, and friends; or felt useful or important to them.

Cognitive function (Leganes’ Cognitive Function Test; PCL: composite of orientation and memory items from Short Portable Mental Status Questionnaire, the Barcelona Test and EPESE short story recall) ^b

Social activities, network, and support were associated with rate of change in cognitive function (coefficient equal to or greater than twice the standard error, no p-values provided). Those with greater social networks and support (compared to those with less) maintained better cognitive functioning up to 80 years. (Beland, 2005).

Social network (visual contact with family, p = 0.028) and activity (p = 0.040) were significant predictors of cognitive decline for both sexes at follow-up. Engagement with friends predicted lower probability of decline for women (p = 0.013) (Zunzunegui, 2003).

Bielak (2007) [20]

Canada (VLS)

N = 530

Age: 55–94

M Age = 68.5

2 years

Social activity: frequency of engagement in social activities incl. Attending concerts or visiting friends.

VLS Activity Lifestyle Questionnaire

Executive functioning

(Lexical decision making ^b ; semantic decision making; simple and choice reaction time)

Social activities associated with lexical decision making only (p < 0.05).

Bourassa (2017)

[72]

EU (SHARE)

N = 19,832

Age: 50+

M Age = 64.4

6 years

Social activity: frequency of participation in volunteer work, social club, religious organisation, community/political group.

Episodic Memory (Ten Word Delayed Recall Test; immediate and delayed recall) ^c

Verbal fluency (Category Fluency Task) ^c

Higher levels of social activity associated with higher memory and fluency scores at BL (p < 0.01), time 1 (p < 0.01) and time 3 (p < 0.05); and social activity at baseline predicted change in memory and fluency over time (p < 0.01). Replicated in a second subsample.

Bosma (2002) [62]

Netherlands

(MAAS)

N = 830

Age: 49–81

3 years

Social activity: frequency per week of engagement in organisational memberships, e.g. clubs.

Episodic Memory (Verbal Learning Test - immediate and delayed recall ^b , total recall ^b )

Attention (Stroop Test)

Processing speed (Letter-Digit Coding Test)

Verbal fluency (Word Fluency Test)

Cognitive Function (MMSE)

Compared to no social activities, socially active participants scores decreased 0.94 points less on total recall (p < 0.05) and 0.30 points less on delayed recall (p < 0.05).

Chen (2016) [58]

Taiwan (TLSA)

N = 2300

Age: 60+

M Age = 70.9

14 years

Social activity: playing games and socialising with friends, neighbours, relatives.

Social support (emotional support): being cared for, being listened to by friends, relatives.

Cognitive function (SPMSQ; short version; 5 questions) ^c

Activity and support related to progression of decline in cognitive function. Low/declining, high/declining or high/stable groups differed in ratings of activity and support (both p < 0.001); lower ratings of activity and support for low/declining. A 1-point increase in emotional support decreased the odds in being in the low/declining group: OR, 0.77; 95% CI, 0.60 to 0.99.

De Frias (2014) [73]

Canada

(VLS)

N = 501

Age: 53–90

M Age = 68.2

4.5 years

Social activity: activities sub-score of VLS Activity Lifestyle Questionnaire

Processing speed (WAIS-R DSS)

Inductive reasoning (Letter Series Test)

Episodic memory (Immediate recall)

Verbal fluency (ETS-CA; ETS-Recognition Vocabulary)

Executive function (composite of Hayling Sentence Completion Test; Stroop; Brixton Test; Colour Trails; Computational span; Reading span)

Social activity not associated with maintenance of cognitive status (high, normal or low) over time.

Social activity moderated the relationship between cognitive status and executive functioning over time but not significantly.

Ellwardt (2013) [22]

Amsterdam

(LASA)

N = 2255

Age: 55–85

M Age = 63

6 years

Social support: 9 most frequent regular socially active contacts;

Emotional support: self-reported;

Instrumental support: self-reported.

Mediating variable: Loneliness

Processing speed (Coding Task) ^a

Reasoning (Raven’s Coloured Progressive Matrices) ^a

Cognitive function (MMSE) ^a

Composite of cognitive function (Coding task, Raven’s, MMSE)^b

Emotional support correlated with improved MMSE, coding (p < 0.001) and Raven’s (p < 0.01) at baseline.

Emotional support directly (p = 0.06) and indirectly (p < 0.05) associated with improved cognitive function.

Instrumental support only indirectly associated with cognitive function (p < 0.05).

Ertel (2008) [27]

USA

(HRS Study)

N = 16,638

Age: 50+

M Age = 64.5

6 years

CMSR: marital status, volunteer activities, and contact with parents, children, neighbours.

Episodic Memory (immediate recall, delayed recall) ^b

Cognitive Status (TICS)

Higher baseline level of social relationships associated with slower rates of memory decline over 6 years (p < .01). Memory decline among the most socially integrated was less than half the rate of change of the least socially integrated (p < .01).

Glei (2005) [63]; Hsu (2007) [64]

Taiwan

(The SHLSE)

N = 2387

Age = 60+

7 years

Social network: marital status, close relatives, other relatives, friends and neighbours (all weekly contact).

Social activities: games, socialising, organised groups, political group; unpaid work, religious groups, business associations, political groups, clan associations, elderly association (score 0/1–2/ 3+).

Cognitive function (SPMSQ) ^b

No social network measure was related to cognitive function. Compared to those with 0 social activities; those with 1–2 social activities failed 13% fewer cognitive tasks (p < 0.001); and those with 3+ failed 33% fewer cognitive tasks (p < 0.001) at FU (Glei, 2005).

Men in political groups at baseline were less likely to show impaired cognitive function (OR = 0.536, p < 0.05)

Compared to no work, women with unpaid work at baseline were more likely to have impaired cognitive function at follow-up (OR = 1.652; p < 0.05) (Hsu, 2007)

Ho (2001) [75]

Hong Kong

N = 988

Age: 70+

M Age = 77.4

3 years

Social network: (composite) contact with friends, family, neighbours; participation in community religious activities (questions adopted from the Lubben 1998 Social Network Scale).

Marital status

Residence (community/institution)

Cognitive function (CAPE) ^b

Male participants; being divorced (p < 0.01), having poor social support (p < 0.001) and institutional living (p < 0.001) was associated with a greater risk of cognitive impairment (CI) at FU.

Men and woman residing in institutions had 4.4 and 2.5-times increased risk of having CI compared with those living in the community.

Holtzman (2004) [21]

Baltimore

(ECA; Waves 1–3)

N = 354

Age: 50+

M Age = 61.3

12 years

Social network: no. of friends, relatives, neighbours

Social activity: frequency of contact with plus no. of friends, neighbours, relatives

Social support: Emotional support

Cognitive function (MMSE) ^b

Maintenance of MMSE scores at FU were associated with baseline social network size (p = 0.02), and network size change (p = 0.02).

Significant predictors of MMSE scores included network size (p = 0.004), social activity (p < 0.006), emotional support (p < 0.005).

Frequency of contact alone did not significantly predict MMSE scores (p > 0.051)¹

Hughes (2008) [76]

USA

(Charlotte Co Healthy

Ageing Study)

N = 217

M Age = 72.4

5 years

Social network: family and friends Social support: emotional support, informational support; instrumental support; satisfaction with support; and negative social interactions.

Attention (Stroop Test) ^a

Processing speed (TMT A and B) ^a

Episodic memory (Hopkins Verbal Learning Test) ^b

Cognitive function (Modified MMSE) ^a

More negative social interactions (p = 0.03) and greater satisfaction with support (p = 0.02) were associated with better MMSE scores at baseline; better performance on speed and attention was associated with greater satisfaction with support (p = 0.01) at baseline.

Over 5 years, less satisfaction with support was marginally associated with decline in episodic memory performance (p = 0.06).

Iwasa (2012) [65]

Japan

N = 567

Age: 70–84

M Age = 75.8

5 years

Social activity: composite score including volunteering and group activities for the elderly.

Cognitive function (MMSE)

No significant association between social activity and cognitive decline (p = .14).

The proportion of those with decline who did not engage in social activities was similar to the proportion of those with decline who did (20.5% vs. 17.1%, p = 0.31).

James (2011) [66]

USA

(Rush Memory and Ageing Project)

N = 1138

M Age = 79.6

5.2 years

Social activity: composite score including frequency of engagement in six common activities that involve social interaction within the last year (e.g. going out, day trips, volunteer work, visit friends, participate in groups, attend church).

Cognitive function (composite score from 19 tests) ^c.

Episodic memory (Logical Memory, East Boston Story, Word List Memory, Word List Recall, Word List Recognition) ^c

Semantic memory (Boston Naming, Verbal Fluency, reading test) ^b

Working Memory (Digit Span Forward and Backward, Digit Ordering) ^b

Processing speed (SDMT, Number Comparison, Stroop Test) ^b

Visuospatial abilities (Line Orientation, Standard Progressive Matrices) ^b

Social activity was associated with higher baseline levels of global cognition (p = 0.002) and with a reduced rate of cognitive decline (p = 0.001) (adjusted model).

A one-point increase in social activity score was associated with a 47% decrease in the annual rate of decline in global cognitive function.

Social activity was associated with episodic memory at baseline (p < 0.001); and with reduced cognitive decline across all five domains at FU (episodic memory, p < 0.001; semantic memory, p < 0.001; working memory, p < 0.001; perceptual speed, p < 0.001; visuospatial, p = 0.003).

Kimura (2016)

Japan (Taketoyo Project)

N = 100

M Age = 73.9

3 years

Social activity: Frequency of going out and contact with friends

Social network: number of companions and number of friends to engage in activities with

Cognitive function (Brief Cognitive Function Examination) ^b

Those with a lower frequency of going out (OR: 0.56, 95% CI, 0.80–6.38, p < 0.05) and fewer friends to engage in activities with (OR: 0.34, 95% CI, 0.13–0.93, p < 0.05) were more likely to experience cognitive decline at follow-up.

Lee (2009) [68]

Korea

(SLAS)

N = 977

Age: 65+

2 years

Social activity: frequency of meeting friends, neighbours, relatives; attending church; going to the movies, sports or cultural exhibits.

Cognitive function (MMSE - Korean version) ^b

Social activity was significantly associated with improved cognitive scores and predicted change in cognitive function over time (p < 0.01).

Lee (2016) [69]

Korea (KLoSA)

N = 1568

Age:65+

M Age = 71.1

4 years

Social activity: frequency of participation in church, senior citizen’s clubs, societies; frequency of phone/face-to-face contact with children, friends, family.

Cognitive function (MMSE - Korean version) ^b

Those who participated in senior citizen’s clubs had a lower risk of cognitive decline at follow-up (p = 0.012). For those >75 years, more frequent contact with children by phone (p = 0.038) or face-to-face (p = 0.001) was associated with reduced cognitive decline.

Lövdén (2005)

[59]

Germany

(BASE)

N = 516

Age: 70–103

M Age = 85

2 years

Social activity: Yesterday Interview (incl. Leisure, instrumental, and social activities and work).

Activity List (engagement in 12 activities, incl. Restaurants, day trips, hobbies, volunteer work, travelling, etc).

Processing speed (Digit Letter and Identical Pictures) ^b

Higher levels of social participation predicted a 2-year positive deviation from the average linear population decline in processing speed (p < 0.05); prior scores of processing speed did not modify socialisation scores.

Marioni (2014)

UK (Paquid)

N = 3653

Age: 65+

20 years

Social activity: membership of a group or association, visits from family and friends, membership of a golden age club, and membership of another club.

Cognitive function (MMSE)^a

The odds of being in the high baseline cognition group (MMSE >27) compared to the immediate decliners (steep linear decline) or slow decliners (>27 MMSE up to age 75 then steep decline) or low baseline cognition (<26) groups were around 10, 3, and 5 times greater for those with high social engagement.

Mousavi-Nasab (2014) [71]

Sweden (Betula Project)

N = 794

Age: 65–85

10 years

Social activity: Visiting family/friends, travelling, going to a restaurant, movies, concerts, or theatre.

Memory (composite of free and cued recall tasks, and recognition tasks) ^b

Verbal fluency (composite of three verbal fluency tasks and a vocabulary test)

Social activity at time 1 and time 2 significantly predicted change in memory performance at time 2 (p < 0.001) and time 3 (p < 0.001).

Nelson (2013) [81]

USA

(HRS-ss)

N = 203

Age: 50+

12 years

CMSR: marital status, volunteer activities, and contact with parents, children, neighbours.

Cognitive function (TICS)

Episodic Memory (immediate and delayed recall of word list)

No significant association (adjusted analysis) between CMSR scores and cognitive function (p = 0.06) or memory (p = 0.43).

Plehn (2004) [60]

Virginia, USA

N = 133

Age: 55+

M age = 75.6

4 years

Social activity: Social activities sub-score of SELF-scale

Cognitive function (MMSE, MDRS – Initiation and Perseveration Subtest) ^c

Memory (composite, Fuld Object Memory Evaluation) ^a

Working memory (Clock Drawing Test) ^a

Verbal fluency (WAIS-R Vocabulary) ^a

At baseline all measures were significantly correlated with social activity (p < 0.05 for clock drawing; p < 0.01 for vocab; p < 0.001 for all other measures).

Participants with cognitive decline at FU reported significantly fewer baseline social activities compared to those who did not decline (p < 0.001).

Seeman (2001) [32]

USA (MSSA)

N = 1189

Age: 70–79

7.5 years

Social networks: no. close social ties; no. of groups; marital status.

Social support: frequency of emotional support; instrumental support; conflicts/ demands; support provided to others (MSSA battery).

Cognitive function (composite score: Boston Naming Test, WAIS-R Similarities, Copying Task, delayed spatial recognition, incidental recall, delayed recall)^c

At baseline, marital status (p = 0.03), emotional support (p = 0.002), and conflicts/demands (p = 0.03) were significantly related to better cognitive functioning.

Only frequency of emotional support was independently related to change in cognitive performance over 7.5 years (p = 0.05).

Shankar (2013) [80]

UK (ELSA)

N = 6034

Age: 50+

M age = 65.6

4 years

CMSR: married/cohabiting, frequency of contact with children/immediate family/ friends and participation in organisations, religious groups, gyms, committees.

UCLA Loneliness Scale (short form)

Episodic Memory (Immediate recall of word lists) ^c

Episodic Memory (Delayed recall of word lists) ^c

Verbal fluency (animals) ^c

At baseline, less integration (isolation) was associated with poorer verbal fluency (p < 0.001), immediate recall (p < 0.001), delayed recall (p < 0.001); and loneliness was associated with poorer immediate recall (p < 0.001) and delayed recall (p = 0.02). After 4 years, increasing isolation was associated with lower scores on verbal fluency (p < 0.05), immediate recall (p < 0.001), and delayed recall (p < 0.001).

Small (2012) [38]

Canada (VLS)

N = 952

Age: 55–85

12 years

Social activity: Frequency of engagement in social activities over the past 2 years.

VLS Activity Lifestyle Questionnaire

Verbal fluency (lexical and semantic decision task) ^b

Episodic memory (story recall)

Semantic memory (fast recall)

Higher social activity was related to greater losses in verbal fluency (p < 0.05).

Lower episodic memory scores were related to greater declines in social activities (p < .01)

Lower semantic memory scores led to greater declines in social activities (p = 0.002).

Tomioka (2016)

Japan

N = 6093

Age: 65+

M Age=

3 years

Social activity: participation in neighbourhood associations, hobby groups, event groups , senior citizen’s clubs, volunteer activities.

Cognitive function (Cognitive Performance Scale) ^b

Relationship between more participation in social groups and lower decline for women (p = 0.026) but not men. 3+ activities associated with prevention of decline (OR = 0.67). Neighbourhood associations (OR = 0.81); event (OR = 0.79), hobby (OR = 0.70) and volunteer (OR = 0.66) groups, all significantly associated with lower risk of cognitive decline.

Wang (2013) [24]

China

N = 1463

Age: 65+

M Age = 71

2.4 years

Social activity: Frequency of engagement in social activities from predefined list (visiting family or friends, receiving visitors at home, giving advice).

Cognitive function (CSID) ^b

Episodic memory (Word List Learning; Word List Recall; IU Story Recall)

Verbal fluency (Animal Fluency Test)

Executive function (IU Token Test)

High level of social activity was associated with less decline in global cognition (p < .05), but not with episodic memory, language or executive function.

Wilson (2015) [77]

USA (Rush Memory and Ageing Project)

N = 529

Age: 50+

4.8 years

Social support - negative social interactions: frequency of negative interactions (NIs), 12 items incl. Neglect/rejection, unwanted intrusion, failure by others to help, insensitive behaviour by others

Cognitive function (composite score on 19 tests) ^c.

Episodic memory ^b

Semantic memory ^b

Working memory ^c

Processing speed ^b

Visuospatial abilities ^c

See James et al. (2011) above for full list of tests.

Higher frequency of NIs at baseline was associated with greater risk of MCI (HR = 1.53, 95% CI = 1.13–2.07). Higher NI score at BL associated with lower BL cognitive test score (p = 0.005), but not rate of decline. Higher mean NI score related to lower global cognition score (p < 0.001) and faster decline (p = 0.002).

Higher mean NI score associated with lower level of function in all domains and faster decline in episodic, semantic and working memory.

Windsor (2014) [79]

Australia (PATH)

N = 2551

Age: 60–64

8 years

Social support: positive/negative exchanges with family, friends and spouses.

Episodic memory (CVLT immediate recall) ^b

Working memory (WMS-DSB) ^b

Processing speed (SDMT) ^b

More (compared to fewer) positive exchanges with friends associated with better memory (p < 0.05); less decline in speed (p < 0.05).

More (compared to less) frequent positive exchanges with family associated with a slower rate of decline in speed (p < 0.05)

Better working memory associated with increased positive exchanges with spouse (p < 0.05).