Social Relationships and Mortality Risk a Meta-analytic Review
Abstract
The size of a person's social network is linked to health and longevity, but it is unclear whether the number of strong social ties or the number of weak social ties is almost influential for wellness. We examined social network characteristics as predictors of bloodshed in the Finnish Public Sector Study (due north = 7,617) and the Health and Social Back up Study (n = 20,816). Social network characteristics were surveyed at baseline in 1998. Data near mortality was obtained from the Finnish National Death Registry. During a hateful follow-upwards period of 16 years, participants with a small social network (≤ten members) were more than probable to die than those with a large social network (≥21 members) (adjusted take chances ratio (60 minutes) = one.23, 95% confidence interval (CI): ane.04, 1.46). Mortality chance was increased among participants with both a small number of strong ties (≤2 members) and a pocket-size number of weak ties (≤5 members) (HR = 1.55, 95% CI: 1.26, i.79) and among participants with both a large number of strong ties and a pocket-sized number of weak ties (60 minutes = 1.28, 95% CI: i.08, one.52), but non among those with a small number of stiff ties and a large number of weak ties (Hour = 1.04, 95% CI: 0.87, one.25). These findings suggest that in terms of mortality hazard, the number of weak ties may be an important component of social networks.
Observational studies take shown that people with larger social networks live healthier and longer lives than those who are lonely and socially isolated (1–four). The mechanisms underlying these associations are likely to involve behavioral, psychosocial, and physiological pathways (five). For example, interpersonal relationships may bear upon an private'due south health habits via social influence and behavior regulation (e.g., normative disapproval of smoking), offer social support that reduces psychosocial stress, and have physiological implications such as enhanced immune, endocrine, and cardiovascular office (half-dozen, 7).
Previous studies of the associations of social networks with wellness have been bars to an examination of structural aspects, such every bit overall size (e.g., number of friends and acquaintances) and frequency of social interactions in different domains of life. Less is understood about which specific aspects of social networks are virtually influential (8–11) and what role the closeness of social relationships might play in health risk and longevity. Due to their proximity, people most often rely on their closest relationships for emotional support (12). These potent ties are oftentimes assumed to be more influential for health than weaker, more distant ties (13). Still, recent inquiry has suggested that more peripheral members of social networks may also contribute importantly to wellness and well-beingness (xiv, fifteen). Beneficial associations between weak ties and well-being are plausible because they are less time-consuming and less emotionally taxing than strong ties but are also more probable to exclude negative aspects of social ties, such equally conflicts and social pressure. Weak ties may also provide an individual with access to people who could exist useful in an emergency (due east.g., medical professionals, financial advisors)—that is, those providing critical instrumental/advisory back up (12).
The social convoy model developed past Antonucci (16) depicts social relationships in terms of 3 concentric circles according to distance from the ego: inner, middle, and outer circles. Although the social convoy model was publicized more than 3 decades agone, few studies to date have sought to distinguish whether weak or stiff ties are more important for health. In this study, we used the social convoy model equally a conceptual framework to examine whether the number of strong or weak ties in a person'due south social network or the overall size of the social network is more influential for wellness, as indicated by adventure of total bloodshed in 2 prospective cohort studies.
METHODS
Written report population
We used information from an occupational cohort study, the Finnish Public Sector Study (FPS), and a population-based accomplice study, the Wellness and Social Support Report (HeSSup), both of which take been described in detail elsewhere (17, eighteen). In brief, in the FPS, data were fatigued from personnel working in four hospital districts in Finland; a total of 7,617 participants (mean age = 42.9 years) provided information about their social network size at the time of the baseline survey in 1998 and had no missing data on baseline covariates (86% of eligible baseline respondents). In HeSSup, 20,816 participants (mean historic period = 36.7 years) met the same inclusion criteria at the baseline survey in 1998 (80% of eligible baseline respondents). The FPS was approved by the ethics committee of the Finnish Institute of Occupational Wellness and the Helsinki University Central Hospital, and HeSSup was approved past the Turku University Hospital Ethics Committee.
Assessment of social networks
Social network size was assessed in both cohort studies at baseline using the social convoy model described by Antonucci (16). The model is based on a set of 3 concentric circles, each of which is considered to represent unlike levels of closeness to the respondent (Figure i). Study participants were asked to identify the initials of people to whom they felt so shut that it was hard to imagine life without them in the innermost circumvolve. The middle circle referred to those persons who felt not quite that shut but still important, and in the outer circumvolve the respondent entered the initials of those persons who had non already been mentioned but were close and important plenty to belong to the individual's personal network. Network members in the inner circumvolve were referred to equally strong ties, while those in the middle and outer circles constituted weak ties. In Web Table i (available at https://bookish.oup.com/aje), we show that any differences in associations with mortality betwixt the numbers of middle- and outer-circle social relationships were small, providing an empirical justification for merging these 2 circumvolve categories.
Figure 1.
Fictitious example of a response to an inquiry regarding the number of persons in the inner circle (stiff ties) and the intermediate and outer circles (weak ties) of the respondent'southward social network. In the current study, social network size was assessed at baseline using the concentric model developed by Antonucci (16) (adapted).
Figure 1.
Fictitious example of a response to an research regarding the number of persons in the inner circle (strong ties) and the intermediate and outer circles (weak ties) of the respondent's social network. In the electric current report, social network size was assessed at baseline using the concentric model developed past Antonucci (16) (adapted).
Overall network size was adamant past summing the numbers of network members in all 3 circles and was categorized equally low (0–10 members), intermediate (11–20 members), or high (≥21 members) (nineteen). A more granulated categorization, used in supplementary analysis, divided participants into groups with 0–two, 3–5, 6–ten, 11–20, or 21 or more members in the social network. We also divided members in the inner circle into groups with pocket-size numbers of strong ties (0–two members) and large numbers of stiff ties (≥three members), respective to the threshold at the lowest quartile. On the same ground, the numbers of members in the eye and outer circles were combined and categorized into small (0–v members) versus large (≥vi members) numbers of weak ties. We constructed a variable combining these categories into four groups: 1) a small number of potent ties and a small number of weak ties (pocket-size/minor), 2) a small-scale number of strong ties and a large number of weak ties (small/large), three) a large number of potent ties and a small number of weak ties (big/minor), and 4) a big number of stiff ties and a large number of weak ties (large/large).
Assessment of covariate information
Baseline covariates included education (basic, intermediate, or high); diagnosed chronic weather condition (diabetes, rheumatoid arthritis, asthma, coronary heart affliction, or cancer), obtained using linkage to the records of the National Drug Reimbursement Annals and the Finnish Cancer Registry (the total number of these conditions was calculated and classified into 2 categories: "none" and "at least one"); and history of depression, assessed by means of the question, "Accept y'all ever been diagnosed with depression past a medico?" (yes/no).
Baseline covariates also included obesity, heavy alcohol consumption, smoking, and low concrete activity, all drawn from standard questionnaires. Body mass alphabetize (weight (kg)/height (m)2) was calculated on the ground of cocky-reported summit and weight, and participants were classified as nonobese (body mass index <xxx) or obese (body mass index ≥30). Alcohol intake, expressed as absolute amount of ethanol consumed in grams per week, was estimated on the basis of reported average consumption of beer, wine, and/or spirits. As was done previously (twenty), the threshold for heavy booze utilize was 288 g/week in men (equivalent of 24 units per calendar week) and 192 thou/week in women (equivalent of 16 units per week). Participants were categorized by smoking condition as nonsmokers (including one-time smokers) or current smokers. Information about average amounts of time spent in physical activity of different intensities was used to estimate metabolic equivalents (METs), a validated mensurate of concrete activity level (21). A MET measurement is obtained past multiplying the amount of fourth dimension spent in each activity by its typical energy expenditure. Nosotros used the following MET values (activity metabolic rate divided by resting metabolic rate): iii.v for an activity intensity corresponding to walking, 5 for an intensity corresponding to vigorous walking, 8 for an intensity corresponding to jogging, and 11 for an intensity corresponding to running. Activity level was and then expressed as the summary score of MET-hours/week (21). Equally in a previous written report, participants whose physical activeness level was less than xiv MET-hours/week were regarded as physically inactive (22).
Ascertainment of bloodshed
Data about mortality was collected past linking the participants to the records from the National Death Registry maintained by Statistics Republic of finland, using the unique personal identification code assigned to all residents of Finland. This database includes exact dates of death and provides near consummate data on population bloodshed (23).
Statistical analyses
Differences in baseline characteristics co-ordinate to social networks were assessed using the t test for continuous variables and the χ2 test for categorical variables. No violation of the proportional hazards assumption was apparent in either of the cohort studies (24). Therefore, Cox proportional hazards models were used to separately examine the associations of the size of a person's overall social network and the person'southward numbers of strong and weak ties with mortality during the follow-upward catamenia. Follow-up started in 1998 and continued until the date of decease or the end of 2013 (FPS) or 2015 (HeSSup), whichever came first. Take a chance ratios were adjusted for historic period and sex in model i and additionally for education, chronic weather condition, lifestyle, and low in model 2. In model 3, we too carried out common aligning for the numbers of potent and weak social ties.
We then examined how the combinations of pocket-sized and large numbers of strong and weak ties were associated with mortality risk. Gamble ratios for the categories "small-scale numbers of both stiff and weak ties" (small-scale/small-scale), "modest number of strong ties but large number of weak ties" (small-scale/large), and "large number of potent ties but pocket-sized number of weak ties" (large/pocket-size) were estimated, with "large numbers of both strong and weak ties" (large/large) as the reference group. To examination the robustness of the associations according to different contexts, we stratified the assay by age, sex, and marital status. To examine reverse causation, we further performed sensitivity analyses after excluding the outset 5 years of follow-upwardly.
The study-specific results were pooled as summary estimates past means of fixed-effects meta-analysis (25). In meta-analysis, data from individual studies are weighted outset and so combined, which avoids some of the problems of simple pooling, such as the ecological fallacy (26). Statistical analyses of study-specific data were performed using SAS software, version 9.4 (SAS Establish, Inc., Cary, Northward Carolina), and the meta-analysis was carried out using the R statistical parcel (version three.2.3; R Foundation for Statistical Calculating, Vienna, Austria).
RESULTS
The hateful age of participants at baseline in the combined data set was 38.3 years (range, 19–63 years), and 67% were women. In both cohort studies, persons who were aged 50 years or over, were male, were single, had a bones didactics, had a low physical action level, or had a history of depression were more likely to have smaller numbers of potent and weak ties (small/small) than younger persons, women, persons who were married/cohabiting, those with an intermediate or high level of pedagogy, those who were physically active, and those without a history of depression ( Web Table 2). In addition, in HeSSup, persons who were obese, heavy booze users, or smokers were more than likely to take small numbers of strong and weak ties than those without these behavioral risk factors.
A total of 461,429 person-years at risk (hateful duration of follow-upward = 16 years) gave rise to ane,080 (3.eight%) deaths in the total study population. Figure 2 shows the distributions of network characteristics and mortality rates. For total social network size and number of weak ties, an increased number of deaths was apparent at the lower end of the distribution. This was not the case for number of strong ties.
Effigy 2.
Number of persons in the respondent's total social network (A), number of stiff social ties (inner circle) (B), and number of weak social ties (outer circles) (C) and corresponding numbers of deaths, Finland, 1998–2013/2015. The figure shows summary estimates pooled from report-specific (Finnish Public Sector Report/Health and Social Back up Study) results.
Figure 2.
Number of persons in the respondent's full social network (A), number of strong social ties (inner circle) (B), and number of weak social ties (outer circles) (C) and corresponding numbers of deaths, Finland, 1998–2013/2015. The figure shows summary estimates pooled from written report-specific (Finnish Public Sector Study/Wellness and Social Back up Study) results.
Tabular array one shows minimally and multivariably adjusted results for the social network–mortality associations. After adjustment for age and sexual practice, participants with a small-scale social network (≤10 members) were one.48 times more than likely to dice during follow-upwards than those with a large social network (≥21 members) (hazard ratio (Hour) = 1.48, 95% conviction interval (CI): 1.25, 1.75). After further adjustment for teaching, chronic atmospheric condition, lifestyle factors, and depression, this clan was markedly adulterate (60 minutes = 1.23, 95% CI: 1.04, ane.46). The pattern of results was like in analyses using a more granulated categorization for overall network size ( Web Tabular array iii). In improver, analyses treating overall network size as a continuous variable (log-transformed) showed a significant association with mortality, both before (Hour = 0.77, 95% CI: 0.71, 0.83; model ane) and subsequently (Hour = 0.87, 95% CI: 0.80, 0.95; model 2) adjustment for covariates ( Web Table 1).
Table 1.
Hazard Ratiosa for All-Cause Bloodshed Co-ordinate to Different Categorizations of Social Network Size, Republic of finland, 1998–2013/2015b
| Social Network Size | No. of Deaths (n = i,080) | Total No. of Participants (n = 28,433) | Model 1c | Model twod | Model 3east | |||
|---|---|---|---|---|---|---|---|---|
| 60 minutes | 95% CI | Hr | 95% CI | Hour | 95% CI | |||
| No. of members in full social network | ||||||||
| ≥21 | 214 | 7,394 | 1.00 | Referent | 1.00 | Referent | ||
| 11–twenty | 398 | 12,955 | 1.02 | 0.86, 1.21 | 0.96 | 0.81, 1.thirteen | ||
| 0–ten | 468 | viii,084 | i.48 | 1.25, 1.75 | ane.23 | 1.04, 1.46 | ||
| No. of strong ties | ||||||||
| ≥3 | 747 | 22,167 | 1.00 | Referent | 1.00 | Referent | 1.00 | Referent |
| 0–2 | 333 | 6,266 | ane.28 | ane.thirteen, i.46 | 1.17 | 1.03, 1.33 | 1.10 | 0.96, ane.26 |
| No. of weak ties | ||||||||
| ≥6 | 713 | 23,007 | 1.00 | Referent | 1.00 | Referent | ane.00 | Referent |
| 0–5 | 367 | five,426 | 1.63 | 1.43, 1.86 | 1.37 | i.20, 1.56 | 1.34 | 1.17, 1.54 |
| Social Network Size | No. of Deaths (due north = 1,080) | Full No. of Participants (due north = 28,433) | Model ic | Model 2d | Model 3e | |||
|---|---|---|---|---|---|---|---|---|
| Hr | 95% CI | HR | 95% CI | HR | 95% CI | |||
| No. of members in full social network | ||||||||
| ≥21 | 214 | 7,394 | 1.00 | Referent | 1.00 | Referent | ||
| 11–20 | 398 | 12,955 | 1.02 | 0.86, ane.21 | 0.96 | 0.81, i.thirteen | ||
| 0–ten | 468 | 8,084 | 1.48 | 1.25, ane.75 | one.23 | one.04, 1.46 | ||
| No. of strong ties | ||||||||
| ≥3 | 747 | 22,167 | 1.00 | Referent | 1.00 | Referent | i.00 | Referent |
| 0–two | 333 | half-dozen,266 | i.28 | 1.13, ane.46 | 1.17 | 1.03, 1.33 | 1.10 | 0.96, 1.26 |
| No. of weak ties | ||||||||
| ≥half dozen | 713 | 23,007 | one.00 | Referent | 1.00 | Referent | 1.00 | Referent |
| 0–5 | 367 | five,426 | 1.63 | 1.43, 1.86 | 1.37 | 1.20, 1.56 | i.34 | i.17, one.54 |
Abbreviations: CI, confidence interval; FPS, Finnish Public Sector Written report; HeSSup, Health and Social Support Study; HR, hazard ratio.
a Summary estimates pooled from study-specific (FPS or HeSSup) results.
b The follow-upwardly period started in 1998 and connected until the date of expiry or the end of 2013 (FPS) or 2015 (HeSSup), whichever came kickoff.
c Model 1: HRs were adjusted for historic period and sex activity.
d Model two: HRs were adjusted for age, sex, education, chronic conditions, lifestyle (obesity, smoking, booze use, and physical activity), and low.
e Model iii: HRs were adjusted for age, sex activity, education, chronic weather condition, lifestyle (obesity, smoking, alcohol utilize, and physical action), and depression. In addition, results were mutually adjusted for numbers of stiff ties and weak ties.
Tabular array one.
Run a risk Ratiosa for All-Cause Mortality According to Dissimilar Categorizations of Social Network Size, Finland, 1998–2013/2015b
| Social Network Size | No. of Deaths (due north = ane,080) | Total No. of Participants (n = 28,433) | Model onec | Model iid | Model 3east | |||
|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | Hour | 95% CI | HR | 95% CI | |||
| No. of members in full social network | ||||||||
| ≥21 | 214 | seven,394 | 1.00 | Referent | i.00 | Referent | ||
| eleven–twenty | 398 | 12,955 | 1.02 | 0.86, one.21 | 0.96 | 0.81, 1.13 | ||
| 0–10 | 468 | 8,084 | 1.48 | ane.25, 1.75 | 1.23 | 1.04, 1.46 | ||
| No. of strong ties | ||||||||
| ≥3 | 747 | 22,167 | ane.00 | Referent | 1.00 | Referent | one.00 | Referent |
| 0–2 | 333 | 6,266 | one.28 | i.13, 1.46 | 1.17 | i.03, one.33 | 1.x | 0.96, 1.26 |
| No. of weak ties | ||||||||
| ≥vi | 713 | 23,007 | i.00 | Referent | 1.00 | Referent | 1.00 | Referent |
| 0–v | 367 | 5,426 | 1.63 | 1.43, 1.86 | 1.37 | i.20, 1.56 | 1.34 | one.17, 1.54 |
| Social Network Size | No. of Deaths (n = i,080) | Total No. of Participants (n = 28,433) | Model anec | Model iid | Model 3e | |||
|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | Hour | 95% CI | HR | 95% CI | |||
| No. of members in total social network | ||||||||
| ≥21 | 214 | seven,394 | i.00 | Referent | i.00 | Referent | ||
| xi–20 | 398 | 12,955 | ane.02 | 0.86, one.21 | 0.96 | 0.81, 1.13 | ||
| 0–10 | 468 | 8,084 | 1.48 | ane.25, 1.75 | 1.23 | one.04, i.46 | ||
| No. of strong ties | ||||||||
| ≥3 | 747 | 22,167 | 1.00 | Referent | one.00 | Referent | ane.00 | Referent |
| 0–two | 333 | 6,266 | 1.28 | one.xiii, one.46 | 1.17 | one.03, 1.33 | 1.10 | 0.96, i.26 |
| No. of weak ties | ||||||||
| ≥6 | 713 | 23,007 | one.00 | Referent | 1.00 | Referent | 1.00 | Referent |
| 0–5 | 367 | 5,426 | ane.63 | 1.43, 1.86 | 1.37 | 1.xx, one.56 | one.34 | 1.17, 1.54 |
Abbreviations: CI, confidence interval; FPS, Finnish Public Sector Study; HeSSup, Health and Social Support Study; Hr, take a chance ratio.
a Summary estimates pooled from report-specific (FPS or HeSSup) results.
b The follow-upwardly menstruum started in 1998 and continued until the date of death or the cease of 2013 (FPS) or 2015 (HeSSup), whichever came get-go.
c Model 1: HRs were adapted for age and sex.
d Model 2: HRs were adjusted for historic period, sex activity, education, chronic atmospheric condition, lifestyle (obesity, smoking, alcohol use, and concrete activity), and depression.
e Model iii: HRs were adjusted for age, sex, education, chronic conditions, lifestyle (obesity, smoking, booze use, and concrete activity), and low. In addition, results were mutually adjusted for numbers of potent ties and weak ties.
Table 1 also shows that in comparison with participants with a large number of weak ties (≥half-dozen), mortality run a risk was significantly increased amidst those who had a small number of weak ties (0–v), both before (model 1; 60 minutes = 1.63, 95% CI (CI): 1.43, 1.86) and after (model two; HR = 1.37, 95% CI: 1.20, 1.56) adjustment for covariates. Furthermore, the association remained afterward additional aligning for the number of potent ties (model 3; HR = 1.34, 95% CI: 1.17, one.54). In dissimilarity, the clan between the number of strong ties and mortality was weaker, and after aligning for the number of weak ties it was not statistically meaning. Similarly, analyses treating the numbers of potent and weak ties as continuous variables showed a stronger association with bloodshed for weak ties than for strong ties ( Web Table 1).
Assay in which covariates were added individually showed that the virtually important contributors to the associations between network variables and mortality were education, smoking, and having a history of depression ( Web Tabular array iv). Aligning for these variables attenuated the age- and sex-adapted association betwixt network size and mortality by 47.ix% (from a hazard ratio of 1.48 to a hazard ratio of 1.25 (95% CI: 1.06, 1.49)), the clan between the number of stiff ties and mortality by 35.seven% (from a run a risk ratio of i.28 to a hazard ratio of 1.18 (95% CI: 1.03, 1.35)), and the association betwixt the number of weak ties and bloodshed by 36.v% (from a take a chance ratio of i.63 to a hazard ratio of 1.40 (95% CI: 1.23, ane.60)).
Effigy three presents findings from the analysis of a four-category combination variable for the numbers of strong and weak ties. Subsequently adjustment for baseline covariates, the risk of mortality was i.55 times higher amid persons with small numbers of strong and weak ties (HR = 1.55, 95% CI: i.26, 1.79) and 1.28 times higher among those who had a large number of strong ties and a modest number of weak ties (HR = one.28, 95% CI: 1.08, i.52), when compared with participants with large numbers of both strong and weak ties. In contrast, no increment in mortality risk was observed amongst persons with a small number of strong ties but a large number of weak ties (Hr = 1.04, 95% CI: 0.87, ane.25). The gamble ratios inverse little after additional adjustment for overall network size (a continuous variable): The hazard ratio was one.53 (95% CI: 1.24, 1.88) for pocket-size numbers of strong and weak ties, 1.thirty (95% CI: 1.07, 1.57) for a large number of potent ties and a pocket-size number of weak ties, and 1.05 (95% CI: 0.87, 1.27) for a small number of strong ties and a large number of weak ties. Minimally adjusted hazard ratios, which did not materially differ from those presented in a higher place, are bachelor in Spider web Figure 1. Furthermore, repeating the main analyses with torso mass alphabetize and alcohol apply modeled as continuous covariates did non change these findings ( Spider web Effigy two).
Figure 3.
Hazard ratios (HRs) for all-cause mortality according to numbers of potent ("small" refers to 0–2 members and "large" refers to ≥3 members) and weak ("small" refers to 0–v members and "large" refers to ≥half-dozen members) ties in the respondent's social network, Finland, 1998–2013/2015. The figure shows summary estimates pooled from study-specific (Finnish Public Sector Study/Health and Social Support Written report) results. HRs were adjusted for age, sex, education, chronic conditions, lifestyle, and depression. Confined, 95% confidence intervals (CIs).
Figure iii.
Hazard ratios (HRs) for all-crusade mortality according to numbers of strong ("minor" refers to 0–two members and "large" refers to ≥3 members) and weak ("pocket-size" refers to 0–5 members and "large" refers to ≥6 members) ties in the respondent's social network, Republic of finland, 1998–2013/2015. The figure shows summary estimates pooled from written report-specific (Finnish Public Sector Report/Health and Social Support Study) results. HRs were adapted for age, sex activity, pedagogy, chronic conditions, lifestyle, and depression. Confined, 95% confidence intervals (CIs).
Figure iv shows that these results were also apparent in subgroup analyses, with the associations existence like in both men and women, younger and older individuals, and single too as married/cohabitating individuals. Formal tests of statistical interaction did not bear witness significant differences between subgroups. In improver, findings of sensitivity analyses showed substantially no change in the master results subsequently exclusion of the commencement 5 years of follow-upwardly from the analyses, a standard approach used to reduce reverse-causation bias ( Spider web Figure three). Similarly, any differences in these ( Web Table v) or other ( Spider web Table six) associations betwixt the two accomplice studies (FPS and HeSSup) were pocket-size. Thus, use of an alternative analytical arroyo, such as pooling of the individual data from the ii studies, yielded findings very like to those of our main analysis based on fixed-effects meta-analysis ( Web Tabular array vii).
Figure 4.
Hazard ratios (HRs) for all-cause bloodshed according to numbers of strong ("small" number refers to 0–2 members and "large" number to ≥3 members) and weak ("pocket-size" number refers to 0–v members and "big" number to ≥6 members) ties in the respondent's social network, by sex, age, and marital status, Republic of finland, 1998–2013/2015. The figure shows summary estimates pooled from study-specific (Finnish Public Sector Report/Wellness and Social Support Written report) results. HRs were adjusted for historic period, sex, instruction, chronic conditions, lifestyle, and depression, as appropriate. Bars, 95% confidence intervals (CIs).
Figure 4.
Hazard ratios (HRs) for all-crusade mortality co-ordinate to numbers of strong ("modest" number refers to 0–2 members and "large" number to ≥3 members) and weak ("small" number refers to 0–5 members and "large" number to ≥6 members) ties in the respondent's social network, past sex, age, and marital status, Republic of finland, 1998–2013/2015. The figure shows summary estimates pooled from study-specific (Finnish Public Sector Report/Health and Social Support Written report) results. HRs were adjusted for age, sex, pedagogy, chronic conditions, lifestyle, and depression, as appropriate. Bars, 95% confidence intervals (CIs).
Word
In this pooled assay of data from 2 contained cohort studies of centre-anile adults followed up for a hateful of 16 years, excess bloodshed risk was observed among persons who had a small number of weak social ties, irrespective of the number of strong social ties. This association was apparent in the full cohort every bit well as in subgroups of study participants, including men and women, younger and older persons, and single and married/cohabitating persons. Furthermore, the association between the number of weak ties and mortality was non attributable to differences in instruction, health condition, lifestyle, or low measured at baseline. The associations of overall network size and number of strong ties with mortality were weaker, and the primacy of weak ties over other characteristics of social networks was observed in both studies, ane based on a general population cohort and the other on an occupational cohort.
We are not aware of whatsoever previous studies on the relationship between numbers of stiff and weak social ties and mortality chance. Withal, our results on overall social network size accord with those of studies showing small social network size and social isolation to be associated with poorer health (2, 27, 28). In our analyses adjusting for age and sexual activity, participants with an overall social network size of only 2 people or less had well-nigh 2 times' greater take a chance of death than those with big social networks including 21 or more members. Nonetheless, results from multivariable aligning suggested that much of this association was attributable to major risk factors, such as low educational level, depression, and unhealthy lifestyle, in the group of people with small social networks. The contained association with having no social network could be particularly hazardous, but in the present study the depression number of participants reporting zero friends (n = six) precluded examination of this consequence.
Some studies have examined unlike types of social networks, such as friend-focused, family unit-focused, neighbor-focused, and restricted networks (29, thirty). They have shown that amongst older adults, friend-focused and diverse networks are associated with lower bloodshed adventure than are restricted networks or having a lower number of friends in i's social network. These studies did not assess the closeness of the members in these particular networks, but it might be assumed that relationships with family members are likely to be the closest, followed by relationships with friends and neighbors (8). Furthermore, stiff social ties are characterized past similarity between the members of the network, whereas weaker social ties, including those with friends or betwixt "friends of friends," are likely to show greater network diversity (12). Thus, our results indicating that it is the number of weak ties that accounts virtually for mortality risk are consistent with previous studies suggesting that diverse social networks are benign for health (31, 32). In the nowadays study, multivariable adjustment showed that nearly half of the association between social network variables and bloodshed was owing to 3 factors: education, smoking, and depression. This suggests that these factors may partially underlie the associations between social network and health.
The results regarding the primacy of weak social ties may be seen every bit unexpected, because emotional support is typically received from stiff ties, oftentimes including spouses and relatives. Notwithstanding, in add-on to emotional support, there are at least 2 other factors explaining the association between social networks and health: informational/instrumental back up, which enables a person to make healthy choices, and negative aspects of interpersonal relationships, such as conflicts and group pressures, which may cause stress and discourage healthy behaviors (33). Having a larger number of weak ties increases network diversity, potentially allowing access to people who are useful in an emergency—that is, those providing critical instrumental/advisory support (e.g., physicians, lawyers, higher admission officials, and depository financial institution loan officers) (12). Interpersonal relationships in weak ties that are, by definition, less time-consuming, emotionally intense, and intimate are also more probable to help people avoid a serious burden of negative social influences (12).
The strength of this investigation was that it was based on 2 large cohort studies, including both occupational and population-based data. In addition, the follow-upward menses was long, extending up to 17 years. Furthermore, information about mortality was obtained from the National Death Registry, providing near complete mortality information on the Finnish population. The fact that the principal finding was replicable beyond 2 different cohort studies supports the generalizability of our results.
Some limitations are noteworthy. First, this was an observational study and therefore cannot bear witness causality. 2d, participants' social networks were assessed only at baseline, and no information about changes in the size of a person'southward social network or in numbers of strong and weak ties was bachelor during follow-upwards. Some previous studies have shown social networks to exist relatively stable over time with respect to full size (34), only network turnover—that is, change in the limerick of the network—likewise occurs and has been shown to exist associated with wellness (35, 36). This should be taken into account in future studies. Third, reverse causation between baseline social network size and health-related factors, such as chronic weather and depression, is an important source of biased results. We addressed this doubtfulness by adjusting the results of the final analyses for chronic conditions and depression. In addition, we conducted sensitivity analyses excluding the first 5 years of follow-up in order to deal with potential confounding by occult diseases. The results of these analyses remained practically unchanged, suggesting that reverse-causation bias is an unlikely explanation for our findings.
In conclusion, our findings support the hypothesis that social networks with large numbers of weak ties protect against premature mortality. This evidence is consistent with policies increasing opportunities to form the types of interpersonal relationships that exercise not need to be highly time-consuming, emotionally intense, or intimate in order to benefit wellness. Examples of such relationships could include patient support groups, support networks for maintaining healthy lifestyles, guild memberships, and other resources for strengthening interpersonal relationships in the community. Clinical trials and natural experiments are now needed to determine the extent to which increases in social networks may reduce risk of morbidity and bloodshed in both younger and older people.
ACKNOWLEDGMENTS
Writer affiliations: Finnish Institute of Occupational Health, Turku and Helsinki, Finland (Maarit Kauppi, Tuula Oksanen, Ville Aalto, Marianna Virtanen); Department of Social and Behavioral Sciences, Harvard T. H. Chan Schoolhouse of Public Health, Boston, Massachusetts (Ichiro Kawachi); Enquiry Department of Epidemiology and Public Health, Faculty of Population Health Sciences, University College London, London, United kingdom (George David Batty, Mika Kivimäki); Institute of Behavioral Sciences, University of Helsinki, Helsinki, Finland (Marko Elovainio); National Institute for Health and Welfare, Helsinki, Republic of finland (Marko Elovainio); Clinicum, Faculty of Medicine, University of Helsinki, Helsinki, Finland (Jaana Pentti, Markku Koskenvuo, Mika Kivimäki); Public Health Unit, Department of Clinical Medicine, Kinesthesia of Medicine, Academy of Turku, Turku, Finland (Jussi Vahtera); and Turku Academy Infirmary, Turku, Republic of finland (Jussi Vahtera).
J.V. and M.Chiliad. are joint senior authors.
This research was supported by the University of Finland (grants 258598, 265977, 267727, 292824, and 311492), NordForsk, the Nordic Research Program on Wellness and Welfare, the Finnish Piece of work Surround Foundation, and the Medical Research Council (Great britain) (grant K013351). Funding bodies for the elective cohort studies (FPS and HeSSup) are listed on the studies' websites.
Each constituent study with individual participant data was canonical past the relevant local or national ideals committee, and all participants gave informed consent to participate. Statistical syntax and exposure data from the accomplice studies are available. Sharing of record-linkage data is not permitted.
M.K. affirms that this manuscript is an honest, accurate, and transparent business relationship of the study existence reported; that no of import aspects of the report have been omitted; and that any discrepancies from the study equally planned (and, if relevant, registered) take been explained. The sponsors played no role in the design and conduct of the written report; in the drove, management, analysis, and interpretation of the information; or in the preparation, review, and approval of the manuscript.
Conflict of interest: none declared.
Abbreviations
-
CI
-
FPS
Finnish Public Sector Written report
-
HeSSup
Health and Social Support Written report
-
Hr
-
MET
REFERENCES
i
Berkman
LF
,
Syme
SL
.Social networks, host resistance, and mortality: a nine-twelvemonth follow-up study of Alameda County residents
.
Am J Epidemiol
.
1979
;
109
(
two
):
186
–
204
.
2
Holt-Lunstad
J
,
Smith
TB
,Layton
JB
.Social relationships and mortality risk: a meta-analytic review
.
PLoS Med
.
2010
;
7
(
7
):
e1000316
.
3
Smith
KP
,
Christakis
NA
.Social networks and health
.
Annu Rev Sociol
.
2008
;
34
:
405
–
429
.
4
Elovainio
Thousand
,
Hakulinen
C
,Pulkki-Råback
L
, et al. .Contribution of risk factors to excess bloodshed in isolated and lonely individuals: an assay of data from the UK Biobank cohort study
.
Lancet Public Wellness
.
2017
;
two
(
6
):
e260
–
e266
.
5
Umberson
D
,
Montez
JK
.Social relationships and wellness: a flashpoint for health policy
.
J Health Soc Behav
.
2010
;
51
(
suppl
):
S54
–
S66
.
six
Umberson
D
,
Crosnoe
R
,Reczek
C
.Social relationships and health behavior across life course
.
Annu Rev Sociol
.
2010
;
36
:
139
–
157
.
seven
Yang
YC
,
Boen
C
,Gerken
Thou
, et al. .Social relationships and physiological determinants of longevity beyond the human life bridge
.
Proc Natl Acad Sci United states
.
2016
;
113
(
3
):
578
–
583
.
8
Antonucci
TC
,
Akiyama
H
,Takahashi
1000
.Attachment and close relationships beyond the life span
.
Attach Hum Dev
.
2004
;
6
(
four
):
353
–
370
.
nine
Manzoli
L
,
Villari
P
,Pirone
GM
, et al. .Marital condition and mortality in the elderly: a systematic review and meta-analysis
.
Soc Sci Med
.
2007
;
64
(
1
):
77
–
94
.
10
Rasulo
D
,
Christensen
K
,Tomassini
C
.The influence of social relations on mortality in later life: a study on elderly Danish twins
.
Gerontologist
.
2005
;
45
(
5
):
601
–
608
.
11
Rutledge
T
,
Matthews
K
,Lui
LY
, et al. .Social networks and marital status predict mortality in older women: prospective prove from the Study of Osteoporotic Fractures (SOF)
.
Psychosom Med
.
2003
;
65
(
iv
):
688
–
694
.
12
Granovetter
MS
.
The forcefulness of weak ties
.
Am J Sociol
.
1973
;
78
(
6
):
1360
–
1380
.
13
Hakulinen
C
,
Pulkki-Råback
L
,Jokela
K
, et al. .Structural and functional aspects of social support equally predictors of mental and physical health trajectories: Whitehall II cohort study
.
J Epidemiol Community Wellness
.
2016
;
70
(
seven
):
710
–
715
.
14
Sandstrom
GM
,
Dunn
EW
.Social interactions and well-being: the surprising power of weak ties
.
Per Soc Psychol Bull
.
2014
;
40
(
7
):
910
–
922
.
15
Christakis
NA
,
Fowler
JH
.Social contagion theory: examining dynamic social networks and human beliefs
.
Stat Med
.
2013
;
32
(
4
):
556
–
577
.
16
Antonucci
TC
.
Hierarchical mapping technique
.
Generations
.
1986
;
x
:
x
–
12
.
17
Kivimäki
M
,
Lawlor
DA
,Davey Smith
Yard
, et al. .Socioeconomic position, co-occurrence of behavior-related adventure factors, and coronary heart illness: the Finnish Public Sector Study
.
Am J Public Health
.
2007
;
97
(
5
):
874
–
879
.
18
Suominen
S
,
Koskenvuo
K
,Sillanmäki
Fifty
, et al. .Non-response in a nationwide follow-up postal survey in Finland: a register-based mortality assay of respondents and non-respondents of the Health and Social Support (HeSSup) Report
.
BMJ Open up
.
2012
;
two
(
2
):
e000657
.
nineteen
Vahtera
J
,
Kivimäki
M
,Uutela
A
, et al. .Hostility and sick health: role of psychosocial resource in two contexts of working life
.
J Psychosom Res
.
2000
;
48
(
one
):
89
–
98
.
twenty
Halonen
JI
,
Kivimäki
M
,Virtanen
M
, et al. .Living in proximity of a bar and risky booze behaviours: a longitudinal study
.
Addiction
.
2013
;
108
(
ii
):
320
–
328
.
21
Ainsworth
Exist
,
Haskell
WL
,Herrmann
SD
, et al. .2011 Compendium of Physical Activities: a second update of codes and MET values
.
Med Sci Sports Exerc
.
2011
;
43
(
eight
):
1575
–
1581
.
22
Kouvonen
A
,
Vahtera
J
,Oksanen
T
, et al. .Chronic workplace stress and insufficient concrete activity: a cohort report
.
Occup Environ Med
.
2013
;
70
(
i
):
iii
–
viii
.
24
Harrell
F
,
Lee
K
. Verifying assumptions of the Cox proportional hazards model. In:Proceedings of the Eleventh Almanac SAS User's Group International Briefing
.
Cary, NC
:
SAS Found, Inc.
;
1986
:
823
–
828
.
25
Hedges
LV
,
Vevea
JL
.Fixed- and random-furnishings models in meta-analysis
.
Psychol Methods
.
1998
;
three
(
4
):
486
–
504
.
26
Bravata
DM
,
Olkin
I
.Simple pooling versus combining in meta-assay
.
Eval Health Prof
.
2001
;
24
(
2
):
218
–
230
.
27
Kawachi
I
,
Colditz
GA
,Ascherio
A
, et al. .A prospective study of social networks in relation to total mortality and cardiovascular disease in men in the United states
.
J Epidemiol Customs Health
.
1996
;
50
(
3
):
245
–
251
.
28
Orth-Gomér
Grand
,
Johnson
JV
.Social network interaction and mortality. A six yr follow-up written report of a random sample of the Swedish population
.
J Chronic Dis
.
1987
;
40
(
x
):
949
–
957
.
29
Litwin
H
,
Shiovitz-Ezra
S
.Network type and bloodshed risk in later life
.
Gerontologist
.
2006
;
46
(
6
):
735
–
743
.
30
Giles
LC
,
Glonek
GF
,Luszcz
MA
, et al. .Effect of social networks on 10 twelvemonth survival in very old Australians: the Australian Longitudinal Written report of Crumbling
.
J Epidemiol Community Health
.
2005
;
59
(
7
):
574
–
579
.
31
Ellwardt
50
,
van Tilburg
T
,Aartsen
M
, et al. .Personal networks and bloodshed chance in older adults: a twenty-year longitudinal written report
.
PLoS One
.
2015
;
x
(
3
):
e0116731
.
32
Barefoot
JC
,
Grønbaek
Yard
,Jensen
G
, et al. .Social network diversity and risks of ischemic centre illness and total mortality: findings from the Copenhagen City Heart Study
.
Am J Epidemiol
.
2005
;
161
(
10
):
960
–
967
.
33
Walen
Hr
,
Lachman
ME
.Social back up and strain from partner, family unit, and friends: costs and benefits for men and women in adulthood
.
J Soc Pers Relat
.
2000
;
17
(
1
):
5
–
30
.
34
Wellman
B
,
Wong
RY
,Tindall
D
, et al. .A decade of network modify: turnover, persistence and stability in personal communities
.
Soc Netw
.
1997
;
19
(
1
):
27
–
50
.
35
Cornwell
B
,
Laumann
EO
.The wellness benefits of network growth: new evidence from a national survey of older adults
.
Soc Sci Med
.
2015
;
125
:
94
–
106
.
36
Eng
PM
,
Rimm
EB
,Fitzmaurice
M
, et al. .Social ties and change in social ties in relation to subsequent total and crusade-specific bloodshed and coronary heart disease incidence in men
.
Am J Epidemiol
.
2002
;
155
(
8
):
700
–
709
.
© The Author(south) 2018. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg Schoolhouse of Public Wellness. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
Source: https://academic.oup.com/aje/article/187/4/746/4093016
0 Response to "Social Relationships and Mortality Risk a Meta-analytic Review"
Postar um comentário