United we thrive: friendship and subsequent physical, behavioural and psychosocial health in older adults (an outcome-wide longitudinal approach)

Abstract

Aims
Three factors converge to underscore the heightened importance of evaluating the potential health/well-being effects of friendships in older adulthood. First, policymakers, scientists, and the public alike are recognizing the importance of social relationships for health/well-being and creating national policies to promote social connection. Second, many populations are rapidly aging throughout the world. Third, we currently face what some call a ‘friendship recession’. Although, growing research documents associations between friendship with better health and well-being, friendship can also have a ‘dark side’ and can potentially promote negative outcomes. To better capture friendship’s potential heterogeneous effects, we took an outcome-wide analytic approach.

Methods
We analysed data from 12,998 participants in the Health and Retirement Study (HRS) – a prospective and nationally representative cohort of U.S. adults aged >50, and, evaluated if increases in friendship strength (between t 0 ; 2006/2008 and t 1 ; 2010/2012) were associated with better health/well-being across 35 outcomes (in t 2 ; 2014/2016). To assess friendship strength, we leveraged all available friendship items in HRS and created a composite ‘friendship score’ that assessed the following three domains: (1) friendship network size, (2) friendship network contact frequency and (3) friendship network quality.

Results
Stronger friendships were associated with better outcomes on some indicators of physical health (e.g. reduced risk of mortality), health behaviours (e.g. increased physical activity) and nearly all psychosocial indicators (e.g. higher positive affect and mastery, as well as lower negative affect and risk of depression). Friendship was also associated with increased likelihood of smoking and heavy drinking (although the latter association with heavy drinking did not reach conventional levels of statistical significance).

Conclusions
Our findings indicate that stronger friendships can have a dual impact on health and well-being. While stronger friendships appear to mainly promote a range of health and well-being outcomes, stronger friendships might also promote negative outcomes. Additional research is needed, and any future friendship interventions and policies that aim to enhance outcomes should focus on how to amplify positive outcomes while mitigating harmful ones.

Full text

Epidemiology and Psychiatric
Sciences
cambridge.org/eps
Original Article
Cite this article: Kim ES, Chopik WJ, Chen Y,
Wilkinson R, VanderWeele TJ (2023). United
we thrive: friendship and subsequent
physical, behavioural and psychosocial health
in older adults (an outcome-wide longitudinal
approach). Epidemiology and Psychiatric
Sciences 32, e65, 1–11. https://doi.org/
10.1017/S204579602300077X
Received: 10 July 2023
Revised: 08 September 2023
Accepted: 21 October 2023
Keywords:
friendship; health and retirement study;
outcome-wide epidemiology; physical health;
psychological well-being; public health
Corresponding author: Eric S. Kim;
Email: eric.kim@ubc.ca
© The Author(s), 2023. Published by
Cambridge University Press. This is an Open
Access article, distributed under the terms of
the Creative Commons Attribution licence
(http://creativecommons.org/licenses/by/4.0),
which permits unrestricted re-use,
distribution and reproduction, provided the
original article is properly cited.
United we thrive: friendship and
subsequent physical, behavioural and
psychosocial health in older adults
(an outcome-wide longitudinal approach)
E. S. Kim1,2,3, W. J. Chopik4, Y. Chen2,5, R. Wilkinson2and
T. J. VanderWeele2,5,6
1Department of Psychology, University of British Columbia, Vancouver, BC, Canada; 2Human Flourishing
Program, Institute for Quantitative Social Science, Harvard University, Cambridge, MA, USA; 3Lee Kum Sheung
Center for Health and Happiness, Harvard T.H. Chan School of Public Health, Boston, MA, USA; 4Department of
Psychology, Michigan State University, East Lansing, MI, USA; 5Department of Epidemiology, Harvard T.H. Chan
School of Public Health, Boston, MA, USA and 6Department of Biostatistics, Harvard T.H. Chan School of Public
Health, Boston, MA, USA
Abstract
Aims. ree factors converge to underscore the heightened importance of evaluating the
potential health/well-being eects of friendships in older adulthood. First, policymakers,
scientists, and the public alike are recognizing the importance of social relationships for
health/well-being and creating national policies to promote social connection. Second, many
populations are rapidly aging throughout the world. ird, we currently face what some call a
‘friendship recession’. Although, growing research documents associations between friendship
with better health and well-being, friendship can also have a ‘dark side’ and can potentially
promote negative outcomes. To better capture friendship’s potential heterogeneous eects, we
took an outcome-wide analytic approach.
Methods. We analysed data from 12,998 participants in the Health and Retirement Study
(HRS) – a prospective and nationally representative cohort of U.S. adults aged >50, and,
evaluated if increases in friendship strength (between t0; 2006/2008 and t1; 2010/2012) were
associated with better health/well-being across 35 outcomes (in t2; 2014/2016). To assess
friendship strength, we leveraged all available friendship items in HRS and created a com-
posite ‘friendship score’ that assessed the following three domains: (1) friendship network size,
(2) friendship network contact frequency and (3) friendship network quality.
Results. Stronger friendships were associated with better outcomes on some indicators of
physical health (e.g. reduced risk of mortality), health behaviours (e.g. increased physical activ-
ity) and nearly all psychosocial indicators (e.g. higher positive aect and mastery, as well as
lower negative aect and risk of depression). Friendship was also associated with increased
likelihood of smoking and heavy drinking (although the latter association with heavy drinking
did not reach conventional levels of statistical signicance).
Conclusions. Our ndings indicate that stronger friendships can have a dual impact on health
and well-being. While stronger friendships appear to mainly promote a range of health and
well-being outcomes, stronger friendships might also promote negative outcomes. Additional
research is needed, and any future friendship interventions and policies that aim to enhance
outcomes should focus on how to amplify positive outcomes while mitigating harmful ones.
ree factors converge to underscore the heightened importance of evaluating the potential
health and well-being eects of friendships in older adulthood. First, policymakers, scien-
tists and the public alike are recognizing the importance of social relationships for health and
well-being outcomes and creating national policies to promote social connection. For example,
the United Kingdom and Japan recently appointed their rst ‘ministers of loneliness’ to com-
bat loneliness at the national level (Fried et al.,2020), and the U.S. Surgeon General recently
published ‘Our Epidemic of Loneliness and Isolation: e U.S. Surgeon General’s Advisory on
the Healing Eects of Social Connection and Community’. Second, populations are rapidly
aging in many countries throughout the world (United Nations, Department of Economic and
Social Aairs, Population Division, 2020). For example, the number of people aged ≥65 years
in the United States is projected to increase by nearly 50% in the next 15 years (Colby and
Ortman, 2014). Coupled with this rapid pace of population aging, older adults are also more
likely to experience risk factors for loneliness and social isolation, such as: living alone, loss
of family and contemporaries, and illness (National Academies of Sciences, Engineering, and
https://doi.org/10.1017/S204579602300077X Published online by Cambridge University Press

2 Kim et al.
Medicine, 2020). ird, we are currently facing what some call
a ‘friendship recession’. Even before the COVID-19 pandemic,
research has shown that friendship levels have been declining over
time (McPherson et al.,2006; Ward, 2022). A recent report high-
lights our changing social landscape; in 2021, a nationwide survey
found that 12% of respondents reported having 0 close friends,
and 49% reported having three or fewer close friends. is con-
trasts markedly with 1990 when the rates were only 3% and 27%,
respectively (Cox, 2021).
As populations age, identifying factors that bolster health and
well-being is critical for stemming the growing wave of chronic
conditions and mounting healthcare costs (Kubzansky et al.,2018).
Although traditional biomedical eorts have focused on identi-
fying risk factors of disease (e.g. loneliness and social isolation)
(Holt-Lunstad, 2022; Hong et al.,2023; Valtorta et al.,2016; Wilson
et al.,2007), researchers and policymakers increasingly seek poten-
tially modiable health assets that uniquely enhance a person’s
ability to foster health and well-being (e.g. friendship) (Kubzansky
et al.,2018; VanderWeele, 2017). A large and growing body of
research has evaluated the adverse eects of loneliness and social
isolation on health and well-being outcomes. As illustrative exam-
ples, loneliness has been linked with a 50%, 29% and 26% elevated
risk of developing Alzheimer’s disease, coronary heart disease and
premature mortality, respectively (Holt-Lunstad, 2022; Valtorta
et al.,2016; Wilson et al.,2007). However, much less research has
focused on the potential inuence that friendships exert on health
and well-being outcomes, especially in older adults (Blieszner et al.,
2019).
Friendships are shaped by genetics, social structural factors
and changing life circumstances (Blieszner et al.,2019; Festinger
et al.,1950; Iervolino et al.,2002). However, various dimensions of
friendship (i.e. quantity, contact frequency, quality of interactions)
can be intervened upon through several approaches (Blieszner
et al.,2019; Hong et al.,2022) (e.g. a series of questions that foster
closeness through mutual vulnerability (i.e. ‘Fast Friends’) (Aron
et al.,1997); an online 12-week ‘Friendship Enrichment Program’
aimed at promoting relational competence, social skills and friend-
ship formation skills (Bouwman et al.,2017); and components of
other research and programs designed to reduce loneliness and
social isolation) (Hoang et al.,2022;Hong et al., In Press). Growing
research has documented friendship’s association with a range of
health and well-being outcomes. Various dimensions of friend-
ship have been associated with better: psychological health (e.g.
higher life satisfaction and lower depression) (Choi et al.,2020;
Fowler and Christakis, 2008; Santini et al.,2015), health behaviours
(e.g. increased: preventive healthcare service use, physical activity,
smoking cessation) (Christakis and Fowler, 2008; Han et al.,2019;
Watt et al.,2014) and physical health (e.g. reduced risk of: cognitive
decline and mortality) (Chopik, 2017; Sharian et al.,2020; Shor
and Roelfs, 2015). Interestingly, like other social factors, friend-
ship might act as an ‘amplication system’ for both benecial and
harmful health and well-being outcomes depending on the social
context. Said another way, networks can magnify whatever they are
seeded with (e.g. attitudes, norms, behaviours). us, friendships
can also have a ‘dark side’ and promote negative health and well-
being outcomes. However, research evaluating negative outcomes
is sparse.
Existing friendship and health/well-being studies have been
seminal and contributed substantially to the literature, but they
remain limited from a causal inference perspective. First, many
studies are cross-sectional, making it challenging to assess the
causal and temporal sequence of variables. Second, only a small
portion of studies in the eld of social relationships and health have
broken down results by relationship type, preventing us from iso-
lating the eects of friendships. ird, among longitudinal studies
evaluating friendship, many health behaviour and physical health
outcomes have not been evaluated in adult populations. Fourth,
some existing studies do not account for relevant potential con-
founders (e.g. depressive symptoms or baseline health). Fih, exist-
ing research on friendships and health and well-being is highly
fragmented. Studies oen focus on a single dimension of friend-
ship (e.g. network size, contact frequency or interaction quality)
and usually limit their scope to one specic health behaviour
or physical health outcome. is piecemeal approach leaves sub-
stantial gaps in our understanding of how various dimensions of
friendship might collectively or dierentially inuence a broader
range of health and well-being outcomes. Sixth, most longitudi-
nal studies have not controlled for friendship characteristics in the
pre-baseline wave. Doing so allows researchers to ask a dierent
question – one particularly important in this era of translational
research: What health and well-being outcomes might we observe
within a relatively short time horizon (over 4 years) if friendships
change?
To begin addressing this question we used an outcome-wide
analytic approach (VanderWeele et al.,2020). is hypothesis-
generating, data-driven approach aims to discover estimates of the
outcomes we might expect to observe if friendship was intervened
upon. Promising ndings can then undergo further investigation
in future studies. We leveraged a large, prospective and nation-
ally representative sample of U.S. older adults and tested if changes
in friendship were associated with better subsequent health and
well-being across 35 separate outcomes. ese outcomes were
chosen because they are frequently included in seminal geron-
tological models that characterize the antecedents, processes and
outcomes that foster people’s ability to age well (Depp and Jeste,
2006; Rowe and Kahn, 1987; Ry and Singer, 2009). To the best
of our knowledge, this is among the rst studies to evaluate how
changes in friendship are associated with changes in health and
well-being.
Methods
Study population
We used data from the Health and Retirement Study (HRS),
a nationally representative panel study of adults aged >50.
Beginning in 2006, HRS sta provided mail-in questionnaires to
study participants and began assessing psychosocial factors in a
randomly selected 50% of HRS participants. e remaining 50% of
HRS participants completed the questionnaire in the subsequent
wave (2008). Each sub-cohort alternates reporting so that each
participant reports psychosocial data every 4 years. In 2006, the
psychosocial questionnaire response rate was 88% and in 2008, it
was 84% (Smith et al.,2013). To increase sample size and power,
we combined data from both sub-cohorts.
Our study used data from three timepoints (t0,t1,t2). All
covariates (including prior levels of friendship) were assessed in the
pre-baseline wave (t0; 2006/2008). e exposure – friendship – was
assessed 4 years later in the baseline wave (t1; 2010/2012). All out-
comes were assessed another 4 years later in the outcome wave (t2;
2014/2016). We restricted the sample to those who completed the
psychosocial questionnaire at baseline because nearly half of our
study outcomes were assessed in the psychosocial questionnaire,
resulting in a nal analytic sample size of 12,998 people.
https://doi.org/10.1017/S204579602300077X Published online by Cambridge University Press

Epidemiology and Psychiatric Sciences 3
e University of Michigan’s Institute of Social Research coor-
dinates the study and provides extensive documentation about
the protocol, instrumentation, sampling strategy and statistical
weighting procedures (Smith et al.,2013; Sonnega et al.,2014).
e HRS has been approved by several ethics committees, includ-
ing the University of Michigan IRB. Further, informed consentwas
obtained from all HRS respondents.
Measures
Friendship
We leveraged all available friendship items in HRS and created a
composite ‘friendship score’ that assessed the following domains.
(1) Friendship Network Size: was assessed by asking respondents,
‘How many of your friends would you say you have a close rela-
tionship with?’ Respondents were asked to provide a numerical
answer in response. (2) Friendship Network Contact Frequency: was
assessed by asking respondents, ‘On average, how oen do you do
each of the following?’ (a) ‘Meet up (include both arranged and
chance meetings)’, (b) ‘Speak on the phone’, (c) ‘Write or email’.
For each category, we reverse coded responses so that higher val-
ues represented more frequent contact (0 = ≤ Every Few Months,
1=1x-2x/Month, 2 =1x-2x/Week, 3 = ≥ 3x/Week). (3) Friendship
Network Quality: was assessed by asking respondents to rate per-
ceived support and strain from their friends. On a 4-point Likert
scale, respondents rated the degree to which they endorsed three
support items (e.g. ‘How much can you rely on them if you have
a serious problem?’) and four strain items (e.g. ‘How oen do
they make too many demands on you?’). Responses to all items
were averaged within each dimension to create a perceived sup-
port score, and a separate strain score. To standardize the measure,
we rst rescored contra-indicative items and created z-scores for
each of the items. en, we created a composite friendship score
by averaging the z-scores for each measure (higher scores indi-
cated stronger friendships). To evaluate potential threshold eects,
we categorized scores into quartiles based on the distribution of
friendship scores in the sample. In secondary analyses, we sep-
arately evaluated each facet of the friendship composite score in
relation to our health and well-being outcomes. e HRS guides
(and Supplementary Text 1) provide further details about each
assessment.
Covariates
All covariates were assessed in the pre-baseline wave (t0;
2006/2008) by self-report. Covariates included (1) demographics
(age, sex, race/ethnicity [White, Black, Hispanic, Other]), mar-
ital status (married/not married), (2) annual household income
(<$50,000, $50,000–$74,999, $75,000–$99,999, ≥$100,000), (3)
total wealth (quintiles of the score distribution in this sample),
(4) educational attainment (no degree, GED/high school diploma,
≥college degree), (5) employment (yes/no), (6) health insurance
(yes/no), (7) geographic region (Northeast, Midwest, South, West),
(8) religious service attendance (none, <1x/week, ≥1x/week), (9)
personality (openness, conscientiousness, extraversion, agreeable-
ness, neuroticism) and (10) childhood abuse (yes/no). We also
adjusted for prior levels of friendship and all outcomes (listed
directly below) in the pre-baseline wave.
Outcomes
We considered 35 outcomes in the outcome wave (2014/2016
(t2)), including dimensions of physical health factors (all-cause
mortality, number of chronic conditions, diabetes, hypertension,
stroke, cancer, lung disease, arthritis, overweight/obesity, physi-
cal functioning limitations, cognitive impairment, chronic pain
and self-rated health), health behaviours (binge drinking, smok-
ing, physical activity and sleep problems), psychological well-being
(positive aect, life satisfaction, optimism, purpose in life, mas-
tery, health mastery and nancial mastery), psychological distress
(depression, depressive symptoms, hopelessness, negative aect
and perceived constraints) and social factors (loneliness, living
with spouse/partner, frequency of contact with: children, other
family or friends – each assessed separately). e HRS guides (and
Supplementary Text 2) provide further details about each assess-
ment (Fisher et al.,2005; Jenkins et al.,2008; Smith et al.,2013;
Sonnega et al.,2014).
Statistical analysis
We used an outcome-wide analytic approach (VanderWeele et al.,
2020), which has several characteristics not widely used outside
of biostatistics and causal inference. us, we summarize those
characteristics here.
First, we control for covariates in the wave prior to the expo-
sure (t0), because, if we assess potential confounders in the same
timepoint as the exposure (t1), it remains unclear if they are con-
founders or mediators. If we accidentally control for mediators
in the same timepoint, we may spuriously attenuate a true eect.
A pragmatic approach to avoiding this problem is to adjust for
potential confounders in the pre-baseline wave (t0).
Second, to enhance our ability to strive towards “no unmea-
sured confounding,” and “exchangeability” (as well as other key
criteria described in “disjunctive cause criterion” for selection of
covariates which includes potential causes of either the exposure
or the outcomes or both), which in turn enhances our ability
to make causal inference, we adjust for a rich set of potential
confounder variables to make these assumptions as plausible as
possible (Greenland and Robins, 1986; VanderWeele et al.,2020).
ird, to reduce potential reverse causality, we also adjust for all
outcome variables in the pre-baseline wave (t0).
Fourth, to evaluate potential “change” in friendship, we adjust
for friendship in the pre-baseline wave (t0). is helps “hold con-
stant” pre-baseline levels of friendship (see Supplementary Text S3
for proof and further explanation). Adjusting for pre-baseline lev-
els of friendship (t0) also has several other advantages, including
reducing the risk of reverse causality by “removing” the accumu-
lating eects that friendship already had on outcomes in the past
(“prevalent exposure”) and allowing readers to instead focus on the
eects of change in friendship (“incident exposure”), over 4 years,
on outcomes.
We ran separate models for each outcome. We ran (1) logis-
tic regression models for binary outcomes with <10% prevalence;
(2) generalized linear models (with a log link and Poisson dis-
tribution) for binary outcomes with ≥10% prevalence or (3) lin-
ear regression models for continuous outcomes. We standardized
all continuous outcomes (mean =0, standard deviation =1)
so their eect size can be interpreted as a standard deviation
change in the outcome. In our tables, and for ease of reviewing
results, we present multiple p-value cutos (both with and with-
out Bonferroni correction for multiple testing) because dierent
investigators use dierent threshold standards for interpreting evi-
dence based on current norms in their specic discipline. In our
Results section, we comment on traditional 0.05 p-value threshold
https://doi.org/10.1017/S204579602300077X Published online by Cambridge University Press

4 Kim et al.
(without Bonferroni correction). However, in all cases, we also
provide 95% condence intervals which are considered prefer-
able assessments of uncertainty since all thresholds are ultimately
arbitrary.
Additional analyses
We ran several additional analyses. First, to evaluate the robustness
of our results to potential unmeasured confounding, we conducted
E-value analyses to assess the minimum strength of association (on
the risk ratio scale) that an unmeasured confounder must have with
both the exposure and the outcome to explain away the observed
association (VanderWeele and Ding, 2017). Second, to evaluate
how our ndings might compare to past research, we re-analysed
all models using a more conventional set of covariates (e.g. sociode-
mographics, but no control for previous friendship scores or out-
comes, in the pre-baseline wave). is analytic approach asks a
dierent question: what are the potential long-term cumulative
eects that the whole history of friendship (approximated by its
current measure but not controlling for the past) has on outcomes?
ird, we re-analysed the main models but removed people who
had any history of a given physical condition at baseline. Fourth, we
separately evaluated each facet of the friendship composite score in
relation to our health and well-being outcomes.
Multiple imputation
We imputed all missing exposures, covariates and outcomes using
an imputation by chained equations approach and generated ve
datasets because it provides a potentially more accurate approach
than other methods of handling missing data and helps address
problems that arise due to attrition (Harel et al.,2018). All analyses
were conducted in Stata 18.0.
Results
In the covariate wave (t0; 2006/2008), the average age of partici-
pants was 65 years old (SD =10) and they were predominantly
women (59%), married (67%) and high school educated (55%).
Table 1 provides the distribution of covariates by categories of
friendship. As the composite friendship score increased from the
lowest quartile (Quartile 1) to the highest quartile (Quartile 4),
so did the scores of its subordinate elements in a dose–response
manner. For example, people in the highest versus lowest quartile
reported having more close friends (7.8 vs. 1.6), meeting friends
>1x/week more frequently (86.3% vs. 9.1%) and receiving more
social support from friends (3.6 vs. 2.3). Supplementary Table 1
describes change in friendship from the pre-baseline wave (t0) to
the baseline wave (t1).
Table 2 shows the associations between friendship and subse-
quent health and well-being outcomes over the 4-year follow-up
period. When considering physical health outcomes, friendship
was associated with 3 (out of 14) outcomes, including 24% reduced
risk of all-cause mortality (95% condence interval [CI]: 0.62,
0.95), 19% reduced risk of stroke (95% CI: 0.67, 0.97), and higher
self-rated health (𝛽 = 0.09, 95% CI: 0.04, 0.15). When considering
health behaviours, friendship was associated with 2 (out of 4) out-
comes including a 43% (95% CI: 1.03, 1.99) increased likelihood of
smoking and 9% increased likelihood of frequency physical activity
(95% CI: 1.00, 1.18).
Friendship was associated with all 7 (out of 7) psychological
well-being outcomes. e strongest associations were with higher:
positive aect (𝛽 = −0.22, 95% CI: 0.15, 0.28) and sense of mastery
(𝛽 = 0.20, 95% CI: 0.14, 0.27). Friendship was also associated with
all 5 (out of 5) psychological distress outcomes. e strongest asso-
ciations were with negative aect (𝛽 = −0.20, 95% CI: −0.26, −0.13)
and a 17% reduced risk of depression (95% CI: 0.68, 1.00). Finally,
friendship was associated with all 5 (out of 5) social factors. e
strongest associations were with 96% increased likelihood of con-
tacting friends ≥1x/week (95% CI: 1.77, 2.17) and 20% increased
likelihood of not living with a spouse/partner (95% CI: 1.09, 1.32).
Additional analyses
We conducted several additional analyses. First, E-value analy-
ses suggested that a few of the associations we observed were at
least moderately robust to unmeasured confounding (Table 3). For
example, an unmeasured confounder of both friendship and all-
cause mortality by risk ratios of 1.94, each, above and beyond the
large number of potential confounders already adjusted for, could
explain away the association. However, weaker joint confounder
associations could not. To shi the condence interval to include
the null, an unmeasured confounder associated with both friend-
ship and all-cause mortality by risk ratios of 1.30 each could suce,
but weaker joint confounder associations could not. Several other
associations were not especially robust to potential unmeasured
confounding. Second, conventionally adjusted covariate models
generally showed larger coecients than fully-adjusted models
(Supplementary Table 2). ese analytic dierences might empha-
size the eect of short-term change in friendship vs. accumulating
eects over time or might reect residual confounding in con-
ventional analyses. ird, when re-evaluating the fully-adjusted
models aer removing anyone with a history of a given physical
condition at baseline (t1), the coecients were generally larger (top
panel of Supplementary Table 2). When evaluating each dimen-
sion of the friendship composite score, frequency of meeting with
friends and negative social strain from friends appeared to have the
largest coecients (Supplementary Tables 3–8).
Discussion
In a large, diverse, prospective and nationally representative sam-
ple of people aged >50, we observed that stronger friendships
were associated with some indicators of better: physical health
outcomes (i.e. reduced risk of: stroke and mortality) and health
behaviours (i.e. increased physical activity), as well as better out-
comes on psychological well-being (i.e. increased: positive aect,
life satisfaction, optimism, purpose in life, mastery, health mastery,
nancial mastery), psychological distress (i.e. decreased: depres-
sion, depressive symptoms, hopelessness, negative aect, perceived
constraints) and social factors (e.g. better scores on: loneliness, liv-
ing with spouse, contact with other family and friends). Friendship
was not associated with other physical health outcomes and health
behaviours. It was also associated with increased likelihood of
smoking and heavy drinking (although the latter association did
not reach conventional levels of statistical signicance).
Our results both align with and deviate from results from
past work that evaluated associations between the “prevalence”
of friendship and outcomes. For example, consistent with past
research we observed that “incident” friendship was associ-
ated with some better outcomes: psychological well-being (e.g.
increased: life satisfaction and lower depression) (Choi et al.,2020;
Fowler and Christakis, 2008; Santini et al.,2015), health behaviours
(e.g. increased: physical activity and heavy drinking) (Rosenquist
et al.,2010; Watt et al.,2014) and physical health (e.g. reduced
risk of mortality) (Shor and Roelfs, 2015). However, our results
https://doi.org/10.1017/S204579602300077X Published online by Cambridge University Press

Epidemiology and Psychiatric Sciences 5
Table 1. Characteristics of participants at baseline by quartiles of friendship (N=10,087)a,b,c
Friendship
Quartile 1 (N=2,522) Quartile 2 (N=2,525) Quartile 3 (N=2,521) Quartile 4 (N=2,519)
Lowest quartile Highest quartile
Participant characteristics n%M SD n %M SD n %M SD n %M SD
Sociodemographic factors
Age (yr; range: 47−97) 66.4 9.3 67.1 9.2 67.3 9.2 67.7 9.1
Female (%) 1,239 49.1 1,401 55.5 1,573 62.4 1,810 71.9
Race/ethnicity (%)
White 1,856 73.6 2,004 79.4 2,061 81.8 2,087 82.9
Black 319 12.6 273 10.8 279 11.1 261 10.4
Hispanic 273 10.8 191 7.6 128 5.1 135 5.4
Other 74 2.9 56 2.2 53 2.1 36 1.4
Married (%) 1,775 70.4 1,780 70.5 1,654 65.6 1,546 61.4
Income (%)
<$50,000 1,515 60.1 1,357 53.7 1,366 54.2 1,395 55.4
$50,000–$74,999 425 16.9 434 17.2 418 16.6 397 15.8
$75,000–$99,999 230 9.1 270 10.7 274 10.9 252 10.0
≥$100,000 352 14.0 464 18.4 463 18.4 475 18.9
Total wealth (%)
1st Quintile 681 27.0 461 18.3 408 16.2 395 15.7
2nd Quintile 555 22.0 506 20.0 488 19.4 468 18.6
3rd Quintile 469 18.6 524 20.8 497 19.7 524 20.8
4th Quintile 418 16.6 517 20.5 572 22.7 553 22.0
5th Quintile 399 15.8 517 20.5 556 22.1 579 23.0
Education (%)
<High school 522 20.7 379 15.1 365 14.5 340 13.5
High school 1,441 57.3 1,447 57.5 1,393 55.4 1,376 54.8
≥College 553 22.0 691 27.5 756 30.1 796 31.7
Employed (%) 1,037 41.2 1,086 43.0 1,011 40.1 979 38.9
Health insurance (%) 2,358 93.6 2,415 95.6 2,421 96.0 2,425 96.3
Geographic region (%)
Northeast 359 14.2 374 14.8 366 14.5 380 15.1
Midwest 686 27.2 758 30.0 726 28.8 627 24.9
South 993 39.4 927 36.7 963 38.2 1,001 39.8
West 483 19.2 464 18.4 466 18.5 509 20.2
Childhood abuse (%) 203 8.1 154 6.2 162 6.5 165 6.6
Physical health
Diabetes (%) 523 20.7 453 17.9 413 16.4 385 15.3
Hypertension (%) 1,398 55.4 1,349 53.4 1,351 53.6 1,307 51.9
Stroke (%) 155 6.1 152 6.0 147 5.8 120 4.8
Cancer (%) 318 12.6 336 13.3 349 13.8 357 14.2
Heart disease (%) 554 22.0 517 20.5 533 21.1 447 17.7
Lung disease (%) 244 9.7 187 7.4 183 7.3 162 6.4
(Continued)
https://doi.org/10.1017/S204579602300077X Published online by Cambridge University Press

6 Kim et al.
Table 1. (Continued.)
Friendship
Quartile 1 (N=2,522) Quartile 2 (N=2,525) Quartile 3 (N=2,521) Quartile 4 (N=2,519)
Lowest quartile Highest quartile
Participant characteristics n%M SD n %M SD n %M SD n %M SD
Arthritis (%) 1,429 56.7 1,471 58.3 1,472 58.4 1,447 57.4
Overweight/obesity (%) 1,865 75.1 1,828 73.1 1,784 71.5 1,727 69.4
Physical limitations (%) 584 23.2 479 19.0 487 19.3 383 15.2
Cognitive impairment (%) 452 18.2 308 12.4 319 12.7 298 11.9
Chronic pain (%) 956 37.9 834 33.0 819 32.5 757 30.1
Self-rated health (range: 1−5) 3.1 1.1 3.3 1.0 3.4 1.0 3.5 1.0
Health behaviours
Binge drinking (%) 275 13.4 289 14.3 276 13.5 231 11.3
Smoking (%) 390 15.6 289 11.5 294 11.7 241 9.7
Frequent physical activity (%) 1,794 71.2 1,922 76.2 2,016 80.0 1,997 79.3
Sleep problems (%) 598 43.0 557 41.1 569 41.8 500 37.6
Religious service attendance (%)
Never 811 32.2 591 23.4 532 21.1 463 18.4
<1x/week 831 33.0 841 33.3 782 31.0 765 30.4
≥1x/week 879 34.9 1,092 43.3 1,205 47.8 1,289 51.2
Psychological well-being
Positive aect (range: 1−5) 3.4 0.8 3.6 0.7 3.7 0.7 3.9 0.7
Life satisfaction (range: 1−7) 4.7 1.5 5.1 1.4 5.2 1.4 5.5 1.4
Optimism (range: 1−6) 4.2 1.0 4.5 0.9 4.6 0.9 4.8 0.9
Purpose in life (range: 1−6) 4.4 0.9 4.6 0.9 4.8 0.9 5.0 0.8
Personal mastery (range: 1−6) 4.6 1.1 4.8 1.1 4.9 1.0 5.0 1.0
Health mastery (range: 1−10) 7.0 2.5 7.3 2.2 7.5 2.1 7.8 2.1
Financial mastery (range: 1−10) 6.9 2.8 7.3 2.5 7.6 2.4 7.8 2.4
Psychological distress
Depression (%) 471 18.7 309 12.2 269 10.7 233 9.2
Depressive symptoms (range: 0−8) 1.6 2.1 1.2 1.8 1.1 1.7 1.0 1.6
Hopelessness (range: 1−6) 2.7 1.4 2.3 1.2 2.1 1.2 1.9 1.1
Negative aect (range: 1−5) 1.8 0.7 1.7 0.6 1.6 0.6 1.5 0.5
Perceived constraints (range: 1−6) 2.5 1.2 2.1 1.1 2.0 1.1 1.8 1.0
Social factors
Loneliness (range: 1−3) 1.7 0.6 1.5 0.5 1.4 0.5 1.3 0.4
Not living with spouse/partner (%) 632 25.7 643 26.1 757 30.6 880 35.9
Contact children ≥1x/week (%) 1,707 69.3 1,857 75.2 1,903 76.9 1,950 79.6
Contact other family ≥1x/week (%) 1,055 42.5 1,277 51.1 1,336 53.5 1,519 61.3
Personality
Openness (range: 1−4) 2.8 0.5 2.9 0.5 3.0 0.5 3.1 0.5
Conscientiousness (range: 1−4) 3.3 0.5 3.4 0.4 3.4 0.4 3.5 0.4
Extraversion (range: 1−4) 3.0 0.6 3.1 0.5 3.3 0.5 3.4 0.5
Agreeableness (range: 1−4) 3.4 0.5 3.5 0.5 3.6 0.4 3.7 0.4
Neuroticism (range: 1−4) 2.2 0.6 2.1 0.6 2.0 0.6 1.9 0.6
(Continued)
https://doi.org/10.1017/S204579602300077X Published online by Cambridge University Press

Epidemiology and Psychiatric Sciences 7
Table 1. (Continued.)
Friendship
Quartile 1 (N=2,522) Quartile 2 (N=2,525) Quartile 3 (N=2,521) Quartile 4 (N=2,519)
Lowest quartile Highest quartile
Participant characteristics n%M SD n %M SD n %M SD n %M SD
Friendship factors
# of close friends (range: 0−124) 1.6 2.0 3.2 2.6 4.3 3.6 7.8 9.4
Meetup with friends ≥1x/week (%) 176 9.1 782 31.1 1447 57.9 2047 83.6
Call friends ≥1x/week (%) 228 11.7 1012 40.3 1851 74.1 2310 94.3
Writing friends ≥1x/week (%) 33 1.8 191 8.0 471 19.8 1134 53.0
Friend social support (range: 1−4) 2.3 0.6 2.9 0.6 3.3 0.6 3.6 0.5
Friend social strain (range: 1−4) 1.6 0.6 1.5 0.5 1.4 0.4 1.3 0.4
aThis table was created based on non-imputed data.
bAll of these variables were used as covariates, and assessed in the wave prior (2006/2008) to the exposure wave (2010/2012).
cThe percentages in some sections may not add up to 100% due to rounding.
also diverge with results from past research. For example, we did
not observe associations with some physical health outcomes (e.g.
reduced risk of cognitive impairment, lower number of physi-
cal health conditions) (Chopik, 2017; Sharian et al.,2020) that
past studies observed. However, when considering our results that
adjusted for conventional covariates, we observed associations that
align with this prior research.
Methodologically, the underlying reasons for diverging results
may stem from a range of sources including dierences in: (1)
which covariates were controlled for, (2) control for prior friend-
ship scores, (3) study population (e.g. nationally representative vs.
non-generalizable samples, younger vs. older samples), (4) study
design (e.g. cross-sectional vs. longitudinal), (5) measurement of
the exposure (friendship has been measured in dierent ways), (6)
measurement of the outcome (e.g. specic outcome vs. compos-
ite measures) and (7) type of analyses employed (e.g. sociometric
analyses vs. covariate-controlled regression approach). When eval-
uating results from conventionally adjusted models, many associ-
ations observed in the friendship literature were also observed in
our results. is suggests that our analyses emphasize short-term
change in friendship vs. accumulating eects over longer durations
of time.
We also observed that friendship was associated with a few
adverse outcomes. For example, it was associated with a 43%
increased likelihood of smoking and 48% increased likelihood of
heavy drinking (this latter association did not reach conventional
levels of statistical signicance but does align with results from
some prior research) (Rosenquist et al.,2010). Like other social
factors, friendship might act as an “amplication system” for both
benecial and harmful health and well-being outcomes depending
on the social context by magnifying whatever they are seeded with
(e.g. attitudes, norms, behaviours). us, friendships can also have
a “dark side” and promote negative health and well-being outcomes
by (Christakis and Fowler, 2007; Villalonga-Olives and Kawachi,
2017): (1) excessively straining group members by “requiring”
them to provide support to others, (2) restricting freedom because
of excessive informal control, (3) excluding out-group members,
(4) “down-levelling” norms so that individuals trying to break free
from negative group norms are penalized and (5) facilitating the
“contagion” of unhealthy behaviours from negative role models. As
an illustration of friendship’s dualistic nature on health outcomes,
friends can be a source of emotional and social support for one
another, yet if the exchange of support occurs in social contexts
where there is smoking and/or excessive consumption of alcohol,
the impact of friendship on health and well-being outcomes may
be both positive and negative.
However, the majority of our results suggest that friendship
has a salubrious association with health and well-being outcomes.
Several hypothesized mechanisms illustrate how friendships might
promote health and well-being (Blieszner et al.,2019; Holt-
Lunstad, 2022; Hong et al.,2022; Kawachi et al.,2008; Villalonga-
Olives and Kawachi, 2017). Friendships might promote health by:
(1) increased diusion of information about health (e.g. referrals
to high-quality healthcare practitioners); (2) social and psycho-
logical support (e.g. instrumental and emotional support in times
of distress); (3) maintenance of healthy norms through informal
social control (e.g. reinforcing norms that certain behaviours [e.g.
exercising] are desirable and the norm) and (4) providing compan-
ionship, fun and satisfaction through mutual interests and shared
activities.
Our study has several limitations, including potential self-
report and common method bias, as both friendship and nearly
all outcomes were self-reported. However, control for pre-baseline
outcomes and a wide range of potential confounders helps to mit-
igate these concerns. Confounding by unmeasured variables and
reverse causality are common concerns in observational research.
However, controlling for a large array of variables, including the
exposure in the pre-baseline wave, the prospective nature of our
data, and results from E-value analyses helps mitigate these con-
cerns. Our results could have been inuenced (moderated) by
numerous other factors such as: marital status, age, sex, race/eth-
nicity, socioeconomic status, personality, baseline health and oth-
ers. Future work should formally evaluate these potential mod-
erators of the friendship and health/well-being association. Our
study also featured several strengths, including the use of a large,
diverse, prospective and nationally representative sample of older
adults. Further, we attained stronger evidence of causality for our
question of interest because we adjusted for pre-baseline values of
the exposure and outcomes, as well as a robust range of covariates
(VanderWeele et al.,2020).
Policymakers, scientists and the public alike are recognizing
the importance of social relationships for health and well-being
https://doi.org/10.1017/S204579602300077X Published online by Cambridge University Press

8 Kim et al.
Table 2. Friendship and subsequent health and well-being (health and retirement study [HRS]: N=12,998)a,b,c,d
Friendship
Quartile 1 (n=3,249) Quartile 2 (n=3,246) Quartile 3 (n=3,246) Quartile 4 (n=3,247)
Outcomes (Reference) RR/OR/𝛽(95% CI) RR/OR/𝛽(95% CI) RR/OR/𝛽(95% CI)
Physical Health
All-cause mortality 1.00 0.96 (0.81, 1.14) 0.85 (0.71, 1.01) 0.76 (0.62, 0.95)*
Number of chronic conditions 0.00 −0.01 (−0.04, 0.02) −0.02 (−0.05, 0.02) −0.01 (−0.05, 0.02)
Diabetes 1.00 0.98 (0.89, 1.09) 1.01 (0.92, 1.12) 0.96 (0.86, 1.09)
Hypertension 1.00 1.01 (0.95, 1.07) 1.01 (0.94, 1.08) 1.01 (0.94, 1.09)
Stroke 1.00 0.94 (0.81, 1.10) 0.86 (0.73, 1.01) 0.81 (0.67, 0.97)*
Cancer 1.00 1.01 (0.90, 1.13) 1.04 (0.92, 1.18) 1.08 (0.94, 1.23)
Heart disease 1.00 0.99 (0.90, 1.08) 1.01 (0.91, 1.11) 0.99 (0.89, 1.10)
Lung disease 1.00 0.98 (0.85, 1.12) 0.98 (0.84, 1.15) 0.95 (0.79, 1.14)
Arthritis 1.00 1.00 (0.94, 1.06) 1.00 (0.94, 1.07) 1.00 (0.93, 1.08)
Overweight/obesity 1.00 1.00 (0.94, 1.06) 0.98 (0.92, 1.04) 0.99 (0.92, 1.07)
Physical functioning limitations 1.00 0.95 (0.86, 1.05) 0.95 (0.86, 1.05) 0.90 (0.80, 1.00)
Cognitive impairment 1.00 0.93 (0.84, 1.03) 0.90 (0.80, 1.01) 0.91 (0.79, 1.04)
Chronic pain 1.00 0.99 (0.90, 1.08) 1.00 (0.91, 1.10) 1.01 (0.90, 1.13)
Self-rated health 0.00 0.04 (−0.01, 0.09) 0.05 (0.00, 0.10)*0.09 (0.04, 0.15)**
Health Behaviours
Heavy drinking 1.00 1.22 (0.98, 1.52) 1.31 (0.91, 1.90) 1.48 (0.96, 2.30)
Smoking 1.00 1.11 (0.80, 1.54) 1.32 (0.97, 1.80) 1.43 (1.03, 1.99)*
Frequent physical activity 1.00 1.03 (0.96, 1.10) 1.06 (0.99, 1.13) 1.09 (1.00, 1.18)*
Sleep problems 1.00 0.98 (0.90, 1.08) 1.02 (0.92, 1.12) 1.05 (0.94, 1.17)
Psychological Well-Being
Positive aect 0.00 0.08 (0.04, 0.12)** 0.13 (0.08, 0.19)** 0.22 (0.15, 0.28)**
Life satisfaction 0.00 0.05 (−0.00, 0.10) 0.09 (0.04, 0.14)** 0.17 (0.10, 0.23)**
Optimism 0.00 0.08 (0.01, 0.14)*0.11 (0.03, 0.19)*0.18 (0.09, 0.27)**
Purpose in life 0.00 0.08 (0.02, 0.13)** 0.12 (0.07, 0.18)** 0.17 (0.12, 0.23)**
Mastery 0.00 0.11 (0.04, 0.19)** 0.14 (0.08, 0.20)** 0.20 (0.14, 0.27)**
Health mastery 0.00 0.05 (0.01, 0.10)*0.08 (0.03, 0.13)** 0.11 (0.03, 0.19)*
Financial mastery 0.00 0.09 (0.02, 0.17)*0.13 (0.06, 0.21)** 0.15 (0.06, 0.23)**
Psychological Distress
Depression 1.00 0.90 (0.76, 1.07) 0.92 (0.78, 1.09) 0.83 (0.68, 1.00)*
Depressive symptoms 0.00 −0.07 (−0.12, − 0.02)** −0.05 (−0.10, − 0.00)*−0.09 (−0.15, − 0.04)**
Hopelessness 0.00 −0.09 (−0.16, − 0.03)** −0.10 (−0.17, − 0.03)** −0.14 (−0.21, − 0.07)**
Negative aect 0.00 −0.11 (−0.16, − 0.07)** −0.15 (−0.20, − 0.10)** −0.20 (−0.26, − 0.13)**
Perceived constraints 0.00 −0.09 (−0.14, − 0.04)** −0.09 (−0.15, − 0.04)** −0.15 (−0.21, − 0.09)**
Social Factors
Loneliness 0.00 −0.09 (−0.15, − 0.02)*−0.13 (−0.19, − 0.08)** −0.21 (−0.29, − 0.13)**
Not living with a spouse/partner 1.00 1.08 (0.99, 1.17) 1.11 (1.02, 1.21)*1.20 (1.09, 1.32)**
Contact children ≥1x/week 1.00 1.03 (0.96, 1.09) 1.02 (0.95, 1.09) 1.02 (0.95, 1.10)
Contact other family ≥1x/week 1.00 1.11 (1.01, 1.22)*1.15 (1.05, 1.25)** 1.18 (1.07, 1.31)**
(Continued)
https://doi.org/10.1017/S204579602300077X Published online by Cambridge University Press

Epidemiology and Psychiatric Sciences 9
Table 2. (Continued.)
Friendship
Quartile 1 (n=3,249) Quartile 2 (n=3,246) Quartile 3 (n=3,246) Quartile 4 (n=3,247)
Outcomes (Reference) RR/OR/𝛽(95% CI) RR/OR/𝛽(95% CI) RR/OR/𝛽(95% CI)
Contact friends ≥1x/week 1.00 1.47 (1.35, 1.61)** 1.78 (1.63, 1.94)** 1.96 (1.77, 2.17)**
Abbreviations: CI, confidence interval; OR, odds ratio; RR, risk ratio.
aIf the reference value is “1,” the eect estimate is OR or RR; if the reference value is “0,” the eect estimate is 𝛽.
bThe analytic sample was restricted to those who had participated in the baseline wave (t1; 2010 or 2012). Multiple imputation was performed to impute missing data on the exposure,
covariates and outcomes. All models controlled for sociodemographic characteristics (age, sex, race/ethnicity, marital status, annual household income, total wealth, level of education,
employment status, health insurance, geographic region), pre-baseline childhood abuse, pre-baseline religious service attendance, pre-baseline values of the outcome variables (diabetes,
hypertension, stroke, cancer, heart disease, lung disease, arthritis, overweight/obesity, physical functioning limitations, cognitive impairment, chronic pain, self-rated health, binge drinking,
current smoking status, physical activity, sleep problems, positive aect, life satisfaction, optimism, purpose in life, mastery, health mastery, financial mastery, depressive symptoms,
hopelessness, negative aect, perceived constraints, loneliness, living with spouse/partner, contact children <1x/week, contact other family <1x/week, contact friends <1x/week),
personality factors (openness, conscientiousness, extraversion, agreeableness, neuroticism) and the pre-baseline value of the exposure. These variables were controlled for in the pre-
baseline was (in t0; 2006 or 2008).
cWe used an outcome-wide analytic approach and ran a separate model for each outcome. We ran a dierent type of model depending on the nature of the outcome: (1) for each binary
outcome with a prevalence of ≥10%, we used a generalized linear model (with a log link and Poisson distribution) to estimate a RR; (2) for each binary outcome with a prevalence of
<10%, we used a logistic regression model to estimate an OR; and (3) for each continuous outcome, we used a linear regression model to estimate a 𝛽.
dAll continuous outcomes were standardized (mean =0; standard deviation =1), and 𝛽was the standardized eect size.
*p <0.05 before Bonferroni correction; **p <0.01 before Bonferroni correction; ***p <0.05 aer Bonferroni correction (the p-value cuto for Bonferroni correction is p=0.05/35
outcomes =p<0.001).
Table 3. Robustness to unmeasured confounding (E-values) for the associa-
tions between friendship (4th quartile vs. 1st quartile) and subsequent health
and well-being (N=12,998)a
Eect estimateb
Confidence
interval limitc
Physical Health
All-cause mortality 1.94 1.30
Number of chronic conditions 1.13 1.00
Diabetes 1.23 1.00
Hypertension 1.10 1.00
Stroke 1.77 1.20
Cancer 1.36 1.00
Heart disease 1.11 1.00
Lung disease 1.29 1.00
Arthritis 1.07 1.00
Overweight/obesity 1.09 1.00
Physical functioning limitations 1.47 1.00
Cognitive impairment 1.44 1.00
Chronic pain 1.11 1.00
Self-rated health 1.39 1.22
Health Behaviours
Binge drinking 2.33 1.00
Smoking 2.22 1.22
Frequent physical activity 1.40 1.07
Sleep problems 1.28 1.00
Psychological Well-being
Positive aect 1.74 1.58
Life satisfaction 1.60 1.44
Optimism 1.63 1.42
Purpose in life 1.62 1.47
(Continued)
Table 3. (Continued.)
Eect estimateb
Confidence
interval limitc
Mastery 1.69 1.52
Health mastery 1.44 1.22
Financial mastery 1.55 1.33
Psychological Distress
Depression 1.72 1.06
Depressive symptoms 1.39 1.22
Hopelessness 1.52 1.34
Negative aect 1.68 1.52
Perceived constraints 1.55 1.40
Social Factors
Loneliness 1.72 1.53
Living with spouse/partner 1.69 1.40
Contact children ≥1x/week 1.17 1.00
Contact other family ≥1x/week 1.65 1.35
Contact friends ≥1x/week 3.32 2.93
aSee VanderWeele and Ding (2017) for the formula for calculating E-values.
bThe E-values for eect estimates are the minimum strength of association on the risk
ratio scale that an unmeasured confounder would need to have with both the exposure
and the outcome to fully explain away the observed association between the exposure and
outcome, conditional on the measured covariates.
cThe E-values for the limit of the 95% confidence interval (CI) closest to the null denote the
minimum strength of association on the risk ratio scale that an unmeasured confounder
would need to have with both the exposure and the outcome to shi the confidenceinterval
to include the null value, conditional on the measured covariates.
outcomes, and many countries have even begun creating national
policies to strengthen social bonds. Findings from our study sug-
gest that friendships are an important element to consider in these
eorts. Further, continuously iterating existing interventions that
target friendship (Aron et al.,1997; Blieszneret al.,2019; B ouwman
et al.,2017; Hoang et al.,2022; Hong et al.,2022), and creating new
interventions might be a promising method of enhancing several
aspects of health and well-being in our rapidly aging population.
https://doi.org/10.1017/S204579602300077X Published online by Cambridge University Press

10 Kim et al.
Supplementary material. e supplementary material for this article can
be found at https://doi.org/10.1017/S204579602300077X.
Availability of data and materials. Data are available for download at
https://hrsdata.isr.umich.edu/data-products.
Acknowledgements. We would like to acknowledge and thank the Health
and Retirement Study conducted by the Institute for Social Research at the
University of Michigan, with grants from the National Institute on Aging
(U01AG09740) and the Social Security Administration.
Author contributions. E.S.K. had full access to all the data in the study, takes
responsibility for the integrity of the data and the accuracy of the data analysis
and contributed to draing the original manuscript; all authors contributed to
the study concept and design; acquisition, analysis, or interpretation of data;
critical revision of the manuscript for important intellectual content. E.S.K. and
W.J.C. are co-1st authors.
Financial support. is work was supported by Michael Smith Health
Research BC and the John Templeton Foundation. e funding source had no
impact on the study design; on the collection, analysis and interpretation of
data; on the writing of the report; or on the decision to submit the article for
publication.
Competing interests. E.S.K. has consulted with AARP and UnitedHealth
Group. T.J.VW. reports receiving licensing fees from Flerish Inc. and
Flourishing Metrics.
Ethical standards. e University of Michigan’s Institute of Social Research
coordinates the study and provides extensive documentation about the proto-
col, instrumentation, sampling strategy and statistical weighting procedures.
e HRS has been approved by several ethics committees, including the
University of Michigan IRB. Further, informed consent was obtained from all
HRS respondents.
References
Aron A, MelinatE, Aron EN, Vallone RD and BatorRJ (1997) e exp erimen-
tal generation of interpersonal closeness: A procedureand some preliminary
ndings. Personality & Social Psychology Bulletin 23(4), 363–377.
Blieszner R, Ogletree AM and Adams RG (2019) Friendship in later life: A
research agenda. Innovation in Aging 3(1), igz005.
Bouwman TE, Aartsen MJ, van Tilburg TGand Stevens NL (2017) Do es stim-
ulating various coping strategies alleviate loneliness? Results from an online
friendship enrichment program. Journal of Social and Personal Relationships
34(6), 793–811.
Choi KW, Stein MB, Nishimi KM, Ge T, Coleman JR, Chen CY,
Ratanatharathorn A, Zheutlin AB and Dunn EC (2020) An exposure-wide
and Mendelian randomization approach to identifying modiable factors
for the prevention of depression. American Journal of Psychiatry 177(10),
944–954.
Chopik WJ (2017) Associations among relational values, support, health, and
well-being across the adult lifespan. Personal Relationships 24(2), 408–422.
Christakis NA and Fowler JH (2007) e spread of obesity in a large social
network over 32 years. e New England Journal of Medicine 357, 370–379.
Christakis NA and Fowler JH (2008) e collective dynamics of smoking in a
large social network. e New England Journal of Medicine 358, 2249–2258.
Colby SL and Ortman JM (2014) Projections of the size and composition of
the U.S. population: 2014 to 2060. US Census Bureau, P25–1143.
Cox DA (2021) e state of American friendship: Change, challenges, and loss.
Survey Center on American Life.
Depp CA and Jeste DV (2006) Denitions and predictors of successful aging:
A comprehensive review of larger quantitative studies. e American Journal
of Geriatric Psychiatry 14(1), 6–20.
Festinger L, Schachter S and Back K (1950) Social pressures in informal
groups; a study of human factors in housing.
Fisher GG, Faul JD, Weir DR and Wallace RB (2005) Documentation
of chronic disease measures in the Health and Retirement Study
(HRS/AHEAD). Ann Arbor, MI: Survey Research Center, University
of Michigan.
Fowler JH and Christakis NA (2008) Dynamic spread of happiness in a large
social network: Longitudinal analysis over 20 yearsin the Framingham Heart
Study. BMJ 337, a2338.
Fried L, Prohaska T, Burholt V, Burns A, Golden J, Hawkley L, Lawlor B,
Leavey G, Lubben J and O’Sullivan R (2020) A unied approach to loneli-
ness. e Lancet 395(10218), 114.
Greenland S and Robins JM (1986) Identiability, exchangeability, and epi-
demiological confounding. International Journal of Epidemiology 15(3),
413–419.
Han SH, Kim K and Burr JA (2019) Social support and preventive healthcare
behaviors among couples in later life. e Gerontologist 59(6), 1162–1170.
Harel O, Mitchell EM, Perkins NJ, Cole SR, Tchetgen Tchetgen EJ,
Sun B and Schisterman EF (2018) Multiple imputation for incomplete
data in epidemiologic studies. American Journal of Epidemiology 187(3),
576–584.
Hoang P, King JA, Moore S, Moore K, Reich K, Sidhu H, Tan CV, Whaley C
and McMillan J (2022) Interventions associated with reduced loneliness and
social isolation in older adults: A systematic review and meta-analysis.JAMA
Network Open 5(10), e2236676.
Holt-Lunstad J (2022) Social connection as a public health issue: e evidence
and a systemic framework for prioritizing the “social” in social determinants
of health. Annual Review of Public Health 43, 193–213.
Hong JH, Nakamura JS, Berkman LF, Chen FS, Shiba K, Chen Y, Kim ES
and VanderWeele TJ (2023) Are loneliness and social isolation equal threats
to health and well-being? An outcome-wide longitudinal approach. SSM-
Population Health 23, 101459.
Hong JH, Nakamura JS, Sahakari SS, Chopik WJ, Shiba K, VanderWeele TJ
and Kim ES (In Press) e silent epidemic of loneliness: Identifying
the antecedents of loneliness using a lagged exposure-wide approach.
Psychological Medicine.
Hong JH, Yeh CS, Sandy LG, Fellows A, Martin DC, Shaeer JA, Tkatch R,
Parker K and Kim ES (2022) Friendship and loneliness: A prototype
roadmap for health system action. American Journal of Preventive Medicine
63(1), 141–145.
Iervolino AC, Pike A, Manke B, Reiss D, Hetherington EM and Plomin R
(2002) Genetic and environmental inuences in adolescent peer socializa-
tion: Evidence from two genetically sensitive designs. Child Development
7391, 162–174.
Jenkins KR, Ofstedal MB and Weir D (2008). Documentation of health
behaviors and risk factors measured in the Health and Retirement Study
(HRS/AHEAD). Ann Arbor, MI: Survey Research Center, University of
Michigan.
Kawachi I, Subramanian SV and Kim D (2008) Social Capital and Health, 1st
edn. New York, NY: Springer.
Kubzansky LD, Human JC, Boehm JK, Hernandez R, Kim ES, Koga HK,
Feig E, Lloyd-Jones D, Seligman MEP and Labarthe D (2018) Positive psy-
chological well-being and cardiovascular health promotion. Journal of the
American College of Cardiology 72(12), 1382–1396.
McPherson M, Smith-Lovin L and Brashears ME (2006) Social isolation in
America: Changes in core discussion networks over two decades. American
Sociological Review 71(3), 353–375.
National Academies of Sciences, Engineering, and Medicine (2020) Social
Isolation and Loneliness in Older Adults: Opportunities for the Health Care
System. Washington, DC: National Academies Press.
Rosenquist JN, Murabito J, Fowler JH and Christakis NA (2010) e spread
of alcohol consumptionb ehavior in a large social network. Annalsof Internal
Medicine 152(7), 426–433.
Rowe JW and Kahn RL (1987) Human aging: Usual and successful. Science
237(4811), 143–149.
Ry CD and Singer B (2009) Understanding healthy aging: Key components
and their integration. Handbook of eories of Aging 2, 117–144.
Santini ZI, Koyanagi A, Tyrovolas S, Mason C and Haro JM (2015) e asso-
ciation between social relationships and depression: A systematic review.
Journal of Aective Disorders 175, 53–65.
Sharian N, Kraal AZ, Zaheed AB, Sol K and Zahodne LB (2020)
Longitudinal associations between contact frequency with friends and
https://doi.org/10.1017/S204579602300077X Published online by Cambridge University Press

Epidemiology and Psychiatric Sciences 11
with family, activity engagement, and cognitive functioning. Journal of the
International Neuropsychological Society 26(8), 815–824.
Shor E and Roelfs DJ (2015) Social contact frequency and all-cause mortality:
A meta-analysis and meta-regression. Social Science & Medicine 128, 76–86.
Smith J, Fisher G, Ryan L, Clarke P, House J and Weir D (2013). Psychosocial
and lifestyle questionnaire.Ann Arbor, MI: Survey Research Center, Institute
for Social Research. Survey Research Center, University of Michigan.
Sonnega A, Faul JD, Ofstedal MB, Langa KM, Phillips JWR and
Weir DR (2014) Cohort prole: e Health and Retirement Study
(HRS). International Journal of Epidemiology 43(2), 576–585.
United Nations, Department of Economic and Social Aairs,
Population Division (2020) World population ageing 2019. https://
www.un.org/en/development/desa/population/publications/pdf/ageing/
WorldPopulationAgeing2019-Highlights.pdf (accessed 14 July 2023).
Valtorta NK, Kanaan M, Gilbody S, Ronzi S and Hanratty B (2016)
Loneliness and social isolation as risk factors for coronary heart disease and
stroke: Systematic review and meta-analysis of longitudinal observational
studies. Heart 102(13), 1009–1016.
VanderWeele TJ (2017) On the promotion of human ourishing. Proceedings
of the National Academy of Sciences 114(31), 8148–8156.
VanderWeele TJ and Ding P (2017) Sensitivity analysis in observational
research: Introducing the E-Value. Annals of Internal Medicine 167(4),
268–274.
VanderWeele TJ, Mathur MB and Chen Y (2020) Outcome-wide longitudinal
designs for causal inference: A new template for empirical studies. Statistical
Science 35(3), 437–466.
Villalonga-Olives E and Kawachi I (2017) e dark side of social capital: A
systematic review of the negative health eects ofs ocial capital.Social Science
& Medicine 194, 105–127.
Ward B (2022) Opinion: Americans are choosing to be alone. Here’s why
we should reverse that. e Washington Post. 23 November. https://www.
washingtonpost.com/opinions/2022/11/23/americans-alone-thanksgiving-
friends/ (accessed 4 September 2023).
Watt RG, Heilmann A, Sabbah W, Newton T, Chandola T, Aida J, Sheiham A,
Marmot M, Kawachi I and Tsakos G (2014) Social relationships and
health-related behaviors among older US adults. BMC Public Health
14, 533.
Wilson RS, Krueger KR, Arnold SE, Schneider JA, Kelly JF, Barnes LL,
Tang Y and Bennett DA (2007) Loneliness and risk of Alzheimer disease.
Archives of General Psychiatry 64(2), 234–240.
https://doi.org/10.1017/S204579602300077X Published online by Cambridge University Press