US trends in social isolation, social engagement, and companionship ⎯ nationally and by age, sex, race/ethnicity, family income, and work hours, 2003–2020

Abstract

Social connectedness is essential for health and longevity, while isolation exacts a heavy toll on individuals and society. We present U.S. social connectedness magnitudes and trends as target phenomena to inform calls for policy-based approaches to promote social health. Using the 2003–2020 American Time Use Survey, this study finds that, nationally, social isolation increased, social engagement with family, friends, and ‘others’ (roommates, neighbors, acquaintances, coworkers, clients, etc.) decreased, and companionship (shared leisure and recreation) decreased. Joinpoint analysis showed that the pandemic exacerbated upward trends in social isolation and downward trends in non-household family, friends, and ‘others’ social engagement. However, household family social engagement and companionship showed signs of progressive decline years prior to the pandemic, at a pace not eclipsed by the pandemic. Work hours emerged as a structural constraint to social engagement. Sub-groups allocated social engagement differently across different relationship roles. Social engagement with friends, others, and in companionship plummeted for young Americans. Black Americans experienced more social isolation and less social engagement, overall, relative to other races. Hispanics experienced much less social isolation than non-Hispanics. Older adults spent more time in social isolation, but also relatively more time in companionship. Women spent more time with family while men spent more time with friends and in companionship. And, men's social connectedness decline was steeper than for women. Finally, low-income Americans are more socially engaged with ‘others’ than those with higher income. We discuss potential avenues of future research and policy initiatives that emerge from our findings.

Full text

SSM - Population Health 21 (2023) 101331
Available online 25 December 2022
2352-8273/© 2022 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
US trends in social isolation, social engagement, and companionship ⎯
nationally and by age, sex, race/ethnicity, family income, and work
hours, 2003–2020
Viji Diane Kannan
a
,
*
, Peter J. Veazie
b
a
Department of Psychiatry, University of Rochester, 300 Crittenden Boulevard, Rochester, NY 14642, USA
b
Department of Public Health Sciences, University of Rochester, 265 Crittenden Blvd., Rochester, NY 14642, USA
ARTICLE INFO
Keywords:
Friends
Family
Health disparities
American Time Use Survey
ABSTRACT
Social connectedness is essential for health and longevity, while isolation exacts a heavy toll on individuals and
society. We present U.S. social connectedness magnitudes and trends as target phenomena to inform calls for
policy-based approaches to promote social health. Using the 2003–2020 American Time Use Survey, this study
nds that, nationally, social isolation increased, social engagement with family, friends, and ‘others’ (roommates,
neighbors, acquaintances, coworkers, clients, etc.) decreased, and companionship (shared leisure and recreation)
decreased. Joinpoint analysis showed that the pandemic exacerbated upward trends in social isolation and
downward trends in non-household family, friends, and ‘others’ social engagement. However, household family
social engagement and companionship showed signs of progressive decline years prior to the pandemic, at a pace
not eclipsed by the pandemic. Work hours emerged as a structural constraint to social engagement. Sub-groups
allocated social engagement differently across different relationship roles. Social engagement with friends,
others, and in companionship plummeted for young Americans. Black Americans experienced more social
isolation and less social engagement, overall, relative to other races. Hispanics experienced much less social
isolation than non-Hispanics. Older adults spent more time in social isolation, but also relatively more time in
companionship. Women spent more time with family while men spent more time with friends and in compan-
ionship. And, men’s social connectedness decline was steeper than for women. Finally, low-income Americans
are more socially engaged with ‘others’ than those with higher income. We discuss potential avenues of future
research and policy initiatives that emerge from our ndings.
1. Introduction
Humans are one of the most social of all animals (Tomasello, 2014)
and seek frequent, on-going social engagement (Baumeister & Leary,
1995). Social isolation (i.e., social decits indicated by infrequent or
insufcient engagement with others) is linked to decrements in health
and longevity (Holt-Lunstad, 2020b). Isolated individuals are at
elevated risk for cardiovascular disease (Hakulinen et al., 2018; Valtorta,
Kanaan, Gilbody, Ronzi, & Hanratty, 2016), dementia (Penninkilampi,
Casey, Singh, & Brodaty, 2018), infectious disease (Cohen, 2021), low
functional status (Fothergill et al., 2011; Shankar, McMunn, Demakakos,
Hamer, & Steptoe, 2017), anxious or depressed mood (Fothergill et al.,
2011), biological markers of poor health (e.g., C-reactive protein, brin-
ogen levels) (Heffner, Waring, Roberts, Eaton, & Gramling, 2011;
Shankar, McMunn, Banks, & Steptoe, 2011), and mortality (Holt-Lun-
stad, Smith, Baker, Harris, & Stephenson, 2015; Holt-Lunstad, Smith, &
Layton, 2010) including overdose (Schell et al., 2021) and suicide (Heuser
& Howe, 2019; Trout, 1980). Isolation is comparable to or rivals other
well-known mortality risk factors like air pollution, smoking, and
inactivity (Holt-Lunstad, 2020b). However, as social engagement in-
creases, health and longevity improve in a dose-response fashion (Yang
et al., 2016).
Given the toll on individuals and society, researchers and policy
makers have recommended cross-sectoral, policy-based approaches to
promote social connectedness (an umbrella term encompassing all
measures of social life) (Holt-Lunstad, 2020a; 2020b; United States
Congress Joint Economic Committee, 2017). Rather than only targeting
the most severely isolated in clinical settings, public policy has the
* Corresponding author.
E-mail addresses: viji_kannan@urmc.rochester.edu (V.D. Kannan), peter_veazie@urmc.rochester.edu (P.J. Veazie).
Contents lists available at ScienceDirect
SSM - Population Health
journal homepage: www.elsevier.com/locate/ssmph
https://doi.org/10.1016/j.ssmph.2022.101331
Received 26 August 2022; Received in revised form 29 November 2022; Accepted 23 December 2022

SSM - Population Health 21 (2023) 101331
2
potential to generate broad societal improvements in social connected-
ness across the risk trajectory. Although most interventions to reduce
social isolation report some success, currently, evidence for
individual-level interventions indicate weak efcacy (Gardiner, Gel-
denhuys, & Gott, 2018; Holt-Lunstad, 2020b; Marczak et al., 2019;
National Academies of Sciences Engineering And Medicine, 2020).
However, public policy has the capacity to have a population wide
impact and to target vulnerable sub-groups that may be less accessible
through individual-level interventions.
For example, preventing tobacco use through smoke-free-air spaces
and excise taxes is considered more effective in curbing related diseases
across the population than trying to get already addicted individuals to
quit smoking. Similarly, identifying public policies that can promote
social engagement and prevent isolation would be more effective across
the population than simply collecting that information at point of care
and addressing the needs of those found to already have high social
isolation (Holt-Lunstad, 2018). Furthermore, the inuence of these
policies varies by sub-group. For example, young, less educated, or
Medicaid recipient expectant mothers respond to excises taxes, whereas
more educated or high-income mothers respond to smoking bans in
restaurants (Markowitz, Adams, Dietz, Tong, & Kannan, 2013).
A public policy approach requires a priori establishment of patterns
and trends as target phenomena (Hodge, White, & Reeves, 2020;
Umberson & Karas Montez, 2010; United States Congress Joint Eco-
nomic Committee, 2017). Documenting patterns and trends related to
social connectedness, nationally and by sub-group, serves as a founda-
tion for theoretical explanations and strategies for effective structural
interventions. Trends reveal progress toward goals and how national
events like a pandemic affect social connectedness. Sub-group patterns
identify populations for targeted interventions and/or further study,
and, together with trends, are essential for designing effective structural,
policy-based solutions. Although the prevalence and hazard of social
isolation is similar to that of most mortality risk factors (Holt-Lunstad,
2020b), social isolation has not received comparable public health
attention nor are its magnitudes and trends at the national and
sub-group level sufciently documented. Thus, we report social
connectedness magnitudes and trends nationally and examine dispar-
ities across population sub-groups.
Studies of social connectedness trends have consisted of a variety of
measures that tap into emotional feelings such as loneliness and those
that point to frequency of social engagement or number of condants.
Loneliness trends among US adolescents increased in one study covering
the years 2000–2018 (Twenge et al., 2021) and decreased in another
study from 1991 to 2012 (Clark, Loxton, & Tobin, 2015). Over three
decades (1974–2008), Americans’ socializing more than once a month
increased slightly for friends (from 40% to 43%), remained stable for
relatives (around 58%), and decreased for neighbors (from 44% to 31%)
(Marsden & Srivastava, 2012). Having no condant with whom to
discuss important matters tripled between 1985 and 2004. However,
being able to conde in one’s spouse increased over those years
(McPherson, Smith-Lovin, & Brashears, 2006; McPherson, Smith-Lovin,
& Brashears, 2008). American adolescents experienced declining
in-person social interactions with peers between 1976 and 2017
(Twenge, Spitzberg, & Campbell, 2019) and declining leisure time,
non-digital social interactions between 2003 and 2017 (Twenge &
Spitzberg, 2020).
1.1. The present study
In this study of social connectedness trends, we use a self-reported,
continuous measure (number of minutes) that captures an individual’s
actual amount of social exposure [both isolation (where exposure is
zero) and engagement (where exposure is greater than zero)] over the
course of a dened time frame (one day). Activities performed over the
course of a day were recorded on that day and collected by the inter-
viewer the following day, minimizing recall bias and measurement
error. The primary inquiry asks about the duration of each activity, with
secondary questions regarding where, when, and with whom the ac-
tivity took place. Thus, unlike survey items that directly ask how much or
how frequently people are socially engaged, our data potentially mini-
mizes social desirability bias, since social exposure is not the main focus.
We examine trends in three aspects of social connectedness: (1) So-
cial Isolation; (2) Social Engagement (with household family, non-
household family, friends, and ‘others’ [neighbors, roommates, ac-
quaintances, clients, coworkers, and other unenumerated roles]); and
(3) Companionship, which refers to shared leisure for the sake of enjoy-
ment and provides an intrinsic satisfaction that need not serve any
extrinsic purpose such as social support (Rook, 1987; Rook & Ituarte,
1999).
For these three aspects of social connectedness, we limit our exam-
ination to in-person contact. While a few studies have found benets to
online interaction, there remain aspects of in-person interpersonal
interaction (e.g., touch, simultaneous expressions, mutually experienced
environment) that cannot be replicated online. Thus, documenting
trends specic to in-person social contact is important. Further, under-
standing changes to in-person social contact aids in determining the
extent to which online platforms serve as either a compliment or a
substitute to in-person contact.
Additionally, this study does not include the ambient presence of
others nor does it include interactions with strangers that the subject
might not report as ‘being with’. These types of social exposure do,
however, offer some benet. Social baseline theory suggests, at a min-
imum, being in relatively close proximity to others imparts physiological
benets (Beckes & Coan, 2011; Coan & Sbarra, 2015). And, studies show
that interaction with strangers such as chatting with the barista,
conversing with a fellow commuter on the bus, or greeting others in
public parks, provides hedonic and learning benets (Atir, Wald, &
Epley, 2022; Sandstrom & Dunn, 2014; Schroeder, Lyons, & Epley,
2022; Van Lange & Columbus, 2021). Future research investigating
these types of minimal social exposures could shed more light on the
dynamics of social connectedness trends.
We present temporal trends nationally and by sub-group in minutes
per day for each year from 2003 to 2020. For national trends, we use
joinpoint analysis to identify if and when signicant changes occurred.
Joinpoints serve as a useful tool for comparing trends with national
events (e.g., the Great Recession in 2008, the Covid pandemic in 2020).
For sub-group analyses, we examine trends by age, given that the tra-
jectory of relational networks and preferences differ by age (Antonucci,
Ajrouch, & Birditt, 2014; Carstensen, 2021). We also examine trends by
sex, race, ethnicity, and class which represent groups that are frequently
treated differently in society with resulting health consequences
(Homan, Brown, & King, 2021). And, we examine trends by number of
hours worked per week, which is cited in recent labor disputes as pre-
venting workers from developing meaningful relationships (Eidelson,
October 25, 2021) and is identied as one of seven structural sectors
inuencing social life in a recently developed systems-based framework
(Holt-Lunstad, 2022). To understand differences in magnitudes across
sub-groups, we report the average minutes per day of social connected-
ness, by sub-group, for each social connectedness measure, over the
2003–2019 period. And, to understand differences in trends across
sub-groups, we report the slope of the trendlines, by sub-group, for each
social connectedness measure, from 2003 to 2019.
2. Methods
2.1. Data
We use the 2003–2020 American Time Use Survey (ATUS), a na-
tionally representative sample of non-institutionalized Americans 15-
years and older (Bureau of Labor Statistics, 2021). ATUS collects data
on how Americans allocate their time over the course of a single,
randomly selected day. Respondents report the duration of each activity
V.D. Kannan and P.J. Veazie

SSM - Population Health 21 (2023) 101331
3
on that day in minutes and with whom the activity took place.
Rather than using scripted questions, interviewers engage in con-
versation as an interviewing technique to obtain precise, accurate
duration of activity measures. This exible interviewing style allows
interviewers to probe in a non-leading way, to guide respondents
through memory lapses, and allows respondents to describe their ac-
tivities with thoroughness. Whether or not the respondent was with
anyone when the activity took place is obtained by asking questions like
“Who was in the room with you” or “Who accompanied you?” for each
activity, excluding sleep, grooming (e.g., bathing), and work. Thus, the
ATUS measures indicate in-person social engagement.
2.2. Sampling and weights
The ATUS sample is distributed across US states in proportion to each
state’s population. Black and Hispanic households are oversampled to
improve the reliability of time-use data for these demographic groups.
The sampling process begins by stratifying households on race/
ethnicity, presence and age of children, and number of adults. Next,
households are randomly selected for each month. A person at least 15-
years of age from each household is then randomly selected. Each
month’s sample is divided into four randomly selected panels (one for
each week of the month). Respondents are then randomly assigned the
day of the week for which they will report their time-use.
The sample for each week is split evenly between weekdays and
weekends (i.e., 25% for each weekend day, Saturday and Sunday; and,
10% for each weekday, Monday through Friday). When sample weights
are applied, all seven days of the week are equally represented at
approximately 14.3% each. Holidays comprise 2% of the reported days.
Time-use diary reports are available for each day from January 1st, 2003
to December 31st, 2020, except for the day before a holiday. On average,
each day contained 35 time-use reports. On average, each year from
2003 to 2019 contained 355 days of time-use data.
The year 2020 contained 305 days of time-use data. ATUS data
collection was suspended for 52 days, from March 18, 2020 to May 9,
2020 — a period dened by sheltering in place. However, time-use data
is available for the other 10-months of 2020. Thus, the 2020 ATUS data
are not representative of a full year. However, ATUS provides a special
weight constructed to take into account sampling issues related to the
pandemic specically related to those days that were excluded from
2020 data collection. Nonetheless, we recommend viewing the 2020
results as somewhat underestimating social isolation and overestimating
social engagement, given those missing dates during the height of social
distancing and quarantine.
Sample weights account for the survey’s complex sampling design
and for non-response. Application of weights is required for computing
estimates with the ATUS data to avoid misleading results. Since some
demographic groups and certain days of the week are oversampled, the
sample weights ensure that each population subgroup and each day is
represented in summary calculations in proportion to the population,
the calendar week, and the calendar year. All our analyses use all ATUS
weights to ensure national and temporal representativeness. These
weights can also be used to estimate quarterly and annual averages
(Bureau of Labor Statistics, 2021). ATUS 2003–2020 contains 219,368
respondents.
2.3. Social connectedness measures
Each reported activity includes information on who the respondent
was with, if anyone, with the exceptions of: sleeping, grooming (e.g.,
bathing), and working. Thus, social connectedness variables in this
study reect non-sleep, non-grooming, and non-work time (in minutes)
during the course of a single 24-h day.
Social isolation is the total number of minutes spent with noone else.
While other people might be in the vicinity of the respondent (e.g., while
shopping alone at a grocery store), if the respondent was not “with” any
of those people, then the respondent was considered alone.
Social engagement is the total number of minutes the respondent spent
with household family members, non-household family members,
friends, and ‘others’ (i.e., roommates, neighbors, acquaintances, co-
workers, clients, and other unenumerated roles).
Companionship is the total number of minutes the respondent spent
with anyone while engaged in socializing, relaxing, leisure, sports, ex-
ercise, recreation, and eating or drinking at a restaurant or bar. Asso-
ciated travel time for these activities is included if spent with other
people. Analyzing companionship presents an opportunity to examine
social engagement with regard to leisure activities. These activities were
considered companionship only if performed with other people. Thus,
social engagement and companionship should not be considered
mutually exclusive. Rook describes companionship as shared leisure for
the sake of enjoyment and proposes that the activities of social
engagement that comprise companionship provide an intrinsic satis-
faction and a sense of belonging (Rook, 1987; Rook & Ituarte, 1999;
Sorkin, Rook, & Lu, 2002).
2.4. Sub-groups
In addition to calculating national social connectedness estimates,
we calculated estimates by sex (male, female), race (white, black, other),
ethnicity (Hispanic, non-Hispanic), age (15–24, 25–34, 35–44, 45–54,
55–64, ≥65 years), family income (≤$25,000; $25,000–$49,999;
$50,000–$99,999; ≥$100,000), and number of hours typically worked
per week (none, 1–25, 26–50, 51–100). ATUS collected information on
the combined income of all family members over the last year including
money from work; net income from business, farm, or rent; pensions;
dividends; interest; Social Security payments; and any other money in-
come received by family members. Hours typically worked per week
included all jobs. Respondents reporting greater than 100 work hours
per week (n =63) were excluded from this analysis.
2.5. Analysis
2.5.1. National analyses
For each year, from 2003 to 2020, we calculate the average number
of minutes per day of social connectedness. These are daily averages for
each year. Thus, a 1-min difference in the annual daily average across
years is equivalent to just over 6-h difference in the total yearly average.
We present these annual daily averages in minutes graphically to show
the temporal trend in social connectedness from 2003 to 2020.
National temporal trends were analyzed using the Joinpoint
Regression Program, (National Cancer Institute, 2022) which calculates
joinpoints — years at which statistically signicant changes to the slope
of the trendlines occur. We used the Weighted BIC Model selection
method. The joinpoints connect consecutive linear segments on a log
scale drawn through the actual trendlines. The program also calculates
the annual percent change (APC) for those linear segments. The program
ts the trend data into the simplest model that best summarizes the data.
For each APC, the program calculates 95% condence intervals and tests
whether the APC is signicantly different from zero at
α
=0.05, based on
a t-distribution. Although, for some linear segments, the program is
unable to calculate these statistics; we report these incidents as [test
statistics unavailable].
2.5.2. Sub-group analyses
Annual daily averages of social connectedness were calculated by
sex, race/ethnicity, age, family income, and hours worked per week and
presented as trendlines. For family income, the annual daily averages
are adjusted for age, sex, race, and ethnicity since income varies by these
demographic characteristics. For hours worked per week, the annual
daily averages are adjusted for age and for family income. Young people
and older adults tend to work fewer hours than adults in mid-life. And,
among Americans who work long hours, those with high income have
V.D. Kannan and P.J. Veazie

SSM - Population Health 21 (2023) 101331
4
potentially more time available for social engagement than those with
low income, since they are able to pay for personal services, such as
cleaning, yard work, shopping, and cooking. Analyses for family income
and hours worked per week were conducted from 2010 to 2020 due to
completeness limitations in the data.
From 2003 to 2019, for each social connectedness measure and for
each sub-group, we calculated: (1) the means and 95% condence in-
tervals to compare magnitudes across sub-groups, and (2) the slope of the
linear trend and 95% condence intervals to compare trends across sub-
groups. In calculating the slope, we normalized the time variable to a
range from 0 to 1 using the formula [(‘year’ – 2003)/16], (i.e., the year
2003 equals zero, the year 2019 equal one, and all years in between take
on values between zero and one in equal increments). Normalizing the
time variable this way allows us to interpret the slope coefcient as a
change across the entire 17-year period. Means and slopes for family
income are adjusted for age, sex, race, and ethnicity. Means and slopes
for hours worked per week are adjusted for age and family income. Since
the 2020 data do not represent the entire year and since social
connection during the pandemic may not be representative of over-
arching trends, that year was omitted from the calculations of sub-group
means and slopes. Whereas in the trendline gures the year 2020 stands
on its own, we did not want to insert 2020 into calculations that
included other years.
Tables in the online supplement contain the numbers used to
construct trendline gures.
3. Results
3.1. Summary statistics of weighted sample
The weighted ATUS sample consisted of 48% male and 52% female
participants. Race and ethnicity composition was: 81% white, 13%
black, 6% other-race, 15% Hispanic, and 85% non-Hispanic. Each age
category was 17% of the sample, except the 55-64-year age group which
was 15%. Nineteen percent of participants had an annual family income
less than $25,000; 24% made $25,000–$49,999; 32% made $50,000–
$99,999; and 25% made ≥$100,000. Thirty-seven percent of partici-
pants worked zero hours per week, 12% worked 1–25 h, 43% worked
26–50 h, and 7% worked 50–100 h.
3.2. National social connectedness trends
Fig. 1 presents national trendlines and joinpoint analyses for all six
measures of social connectedness. Nationally, the average time spent
alone increased from 285-min/day in 2003 to 309-min/day in 2019 and
continued to increase to 333-min/day in 2020. The 24-min per day
difference between 2003 and 2019 represents 146-h more social isola-
tion in 2019 than in 2003. This 146-h increase in social isolation over the
course of 17-years was repeated over the course of one-year between
2019 and 2020. At the national level, joinpoint analyses show that the
social isolation slope changed signicantly in 2018. In 2003–2018, APC
=0.32 [95%CI=(0.2,0.5); t-statistic =4.4; p =0.001] and in
2018–2020, APC =5.65 [95%CI=(1.2,10.3); t-statistic =2.7; p =
0.017].
Average time spent socially engaged with household family
decreased from 262-min/day in 2003 to 243-min/day in 2019, but
increased to 252-min/day in 2020; representing 122-h less in 2019 than
in 2003; and, 61-h more in 2020 than in 2019. Joinpoint analyses for
household family social engagement indicate a declining linear trend
over the entire observed period with no joinpoints, APC = − 0.31 [95%
CI=(–0.4,–0.2); t-statistic = − 5.2; p <0.001].
Average time spent socially engaged with non-household family
decreased overall from 35-min/day in 2003 to 28-min/day in 2019 and
22-min/day in 2020 representing 43-h less in 2019 than in 2003; and,
37-h less between 2020 and 2019. Joinpoint analyses for non-household
family social engagement indicate that the slope changed signicantly in
2010 and 2019. The 2003–2010 APC =0.78 [95%CI=(–1.0,2.5); t-sta-
tistic =1.0; p =0.348], the 2010–2019 APC = − 2.33 [95%CI=
(–4.0,–0.6); t-statistic = − 2.9; p =0.014], and the 2019–2020 APC =
−20.69 [test statistics unavailable].
Average time spent socially engaged with friends decreased overall
from 60-min/day in 2003 to 34-min/day in 2019 and continued to
decrease to 20-min/day in 2020 representing 158-h less in 2019 than in
2003; and, 85-h less in 2020 than in 2019. Joinpoint analyses for social
engagement with friends indicate that the slope changed signicantly in
2007, 2013, and 2019. The 2003–2007 APC = − 4.38 [95%CI=
(–7.3,–1.4); t-statistic = − 3.4; p =0.010], 2007–2013 APC =1.58 [95%
CI=(–1.0,4.2); t-statistic =1.4; p =0.190], 2013–2019 APC = − 6.89
[95%CI=(–9.6,–4.1); t-statistic = − 5.6; p =0.001] and 2019–2020 APC
= − 45.83 [test statistics unavailable].
Fig. 1. US Social Connectedness Trends, 2003–2020. Annual Daily Average in Minutes are in blue trendlines. Joinpoint lines are black with red-bordered square
points indicating years at which the slope the trendline changes signicantly.
V.D. Kannan and P.J. Veazie

SSM - Population Health 21 (2023) 101331
5
Average time spent socially engaged with ‘others’ decreased overall
from 54-min/day in 2003 to 43-min/day in 2019 and continued to
decrease to 34-min/day in 2020 representing 67-h less in 2019 than in
2003; and, 55-h less in 2020 than in 2019. Joinpoint analyses for social
engagement with ‘others’ indicate that the slope changed signicantly in
2005, 2007, and 2019. The 2003–2005 APC =2.63 [95%CI=
(–5.6,11.6); t-statistic =0.7; p =0.501], 2005–2007 APC = − 9.63 [test
statistics unavailable], 2007–2019 APC = − 0.45 [95%CI=(–1.2,0.3); t-
statistic = − 1.4; p =0.196] and 2019–2020 APC = − 23.23 [test sta-
tistics unavailable].
Average companionship time decreased overall from 202-min/day in
2003 to 182-min/day in 2019 and continued to decrease to 174-min/
day in 2020 representing 122-h less companionship in 2019 than in
2003; and, 49-h less companionship in 2020 than in 2019. Joinpoint
analyses for companionship indicate that the slope changed signicantly
in 2013. The 2003–2013 APC = − 0.16 [95%CI=(–0.5,0.2); t-statistic =
−1.1; p =0.294] and 2013–2020 APC = − 1.4 [95%CI=(–2.1,–0.7); t-
statistic = − 4.3; p =0.001].
3.3. Sub-group social connectedness trends
Trendlines by sex, race/ethnicity, age, family income, and hours
worked per week are depicted in Figs. 2–6, respectively. Table 1 presents
means and Table 2 presents slopes, for each social connectedness mea-
sure by sub-group, across the years 2003–2019. The sub-group results
described here draw from Tables 1 and 2 as well as from the corre-
sponding gures. We describe sub-group social connectedness statistics
in hours per year based on the average daily minutes in Tables 1 and 2
3.3.1. Sex (Fig. 2)
From 2003 to 2019, on average, women experienced 37-h/year more
social isolation than men. Women spent substantially more time with
family (365-h/year more) than men. Men spent slightly more time with
friends and ‘others’ than women. Men also experienced more time in
companionship (91-h/year more) than women. Social isolation
increased for both men and women. The increase in men’s social isola-
tion (176-h over the observed period) was steeper than for women (73-
h). All measures of social engagement decreased for both men and
women. The decline in social engagement with ‘others’ and in
companionship was steeper for men than for women. Importantly, if
current trends continue, men will surpass women in social isolation and
fall to or below women in social engagement with friends and ‘others’
and in companionship.
3.3.2. Race/ethnicity (Fig. 3)
Black Americans experienced more social isolation, on average, than
all other racial and ethnic categories: 359-h/year more than white
Americans, 444-h/year more than other-race Americans, and 663-h/
year more than Hispanic Americans. In total, black Americans also
experienced less social engagement across all roles (inuenced primarily
by household family social engagement), on average, than all other
racial and ethnic categories: 377-h/year less than other-race Americans,
395-h/year less than white Americans, and 505-h/year less than His-
panic Americans. Hispanics spent less time socially isolated and more
time engaged with household family than non-Hispanics. Time with
friends was similar across race, but higher among non-Hispanics than
Hispanics. Companionship was highest among white Americans
compared to non-white races and Hispanics. Trends over time show
larger increases in social isolation and larger declines in companionship
for non-white races and Hispanics compared to white Americans.
3.3.3. Age (Fig. 4)
Social isolation was highest for the oldest age category (≥65-years)
which experienced, on average, 554-h/year more social isolation than
those ages 55-64-years, 925-h/year more than those ages 45-54-years,
and 1405-h/year more than those ages 25-34-years. Of all age groups,
the youngest age category (15-24-years) spent the least amount of time
with household family and the most amount of time with friends and
others. On average, the youngest age group also spent the most amount
of time in companionship — followed by the oldest age group. However,
as shown in Fig. 4, from 2015 to 2020, the oldest age group eclipsed the
youngest age group in companionship. So, although adults 65-years and
older experienced the most social isolation, they also had relatively high
levels of companionship. From 2003 to 2019, social engagement
plummeted with friends (377-h), ‘others’ (195-h), and in companionship
(298-h) for the youngest age group.
3.3.4. Annual family income (Fig. 5)
Annual family income analyses are adjusted for age, sex, race, and
ethnicity and start in 2010 due to data completeness limitations. Social
isolation was inversely related to family income. From 2010 to 2019, on
average, the lowest income group (<$25K) experienced 310-h/year
Fig. 2. By sex: US social connectedness trends, annual daily average in Minutes, 2003–2020. Men (blue), Women (red).
V.D. Kannan and P.J. Veazie

SSM - Population Health 21 (2023) 101331
6
more social isolation than the $25K–$50K income group, 462-h/year
more social isolation than the $50K–$100K income group, and 596-h/
year more social isolation than the ≥$100K income group. Time spent
with household family was proportional to family income, whereas time
spent with non-household family was inversely related to family income.
The lowest income group also experienced the largest decline in social
engagement with household family and in companionship (420-h and
377-h decline over the observed period, respectively). And, the lowest
income group spent slightly more time with ‘others’ than the higher
income groups.
3.3.5. Hours worked per week (Fig. 6)
Analyses for hours worked per week are adjusted for age and family
income and start in 2010. Social isolation, social engagement overall,
and time in companionship are all inversely related to the number of
hours worked per week. Less hours spent at work potentially affords
Americans more time to spend alone as well as more time to spend with
other people. Those who work 25 h or less experienced greater declines
in friend social engagement than those who work more than 25 h;
perhaps, because this group had more “wiggle room” — that is, by
already spending more time with friends in 2003, they could potentially
lose more time in that social relationship. Otherwise, while the four
categories of work hours differed somewhat in their slopes, the overall
pattern was one of similarity in trends over time.
4. Discussion
This study was motivated by the need for a population wide account
of trends in various social connectedness measures and across various
sub-groups with the purpose of situating our current understanding
Fig. 3. By race & ethnicity: US social connectedness trends, annual daily average in Minutes, 2003–2020. White (blue), Black (red), Other (yellow), Hispanic (green).
Fig. 4. By age: US social connectedness trends, annual daily average in Minutes, 2003–2020. 15-24 years (blue), 25-34 years (red), 35-44 years (yellow), 45-54 years
(green), 55-64 years (orange), 65+years (purple).
V.D. Kannan and P.J. Veazie

SSM - Population Health 21 (2023) 101331
7
within this broad perspective and for stimulating structural, policy-
based proposals to improve social connectedness. We nd Americans’
social connectedness declined over almost two decades — social isola-
tion increased, social engagement decreased across all roles, and
companionship decreased.
The prevailing trend for most social connectedness measures was
exacerbated by the pandemic. However, household family social
engagement and companionship showed signs of progressive decline
years prior to the pandemic, at a pace not eclipsed by the pandemic.
Social connectedness may be affected by the pandemic for years to
come. However, since social connectedness trends were declining even
before the pandemic started, simply ‘getting back to normal’ is insuf-
cient. A limitation of note is that, for 52 days during the height of social
distancing (March 18, 2020–May 9, 2020), no data was collected. Thus,
the uptick in social isolation and social engagement with household
family and the decline in all other forms of social engagement are likely
underestimated for 2020. Our ability to accurately assess the impact of
the pandemic on social connectedness will require re-examining these
trends over the next several years.
The most dramatic trends in social connectedness were seen in the
plummeting social engagement with friends, ‘others’, and companion-
ship for the youngest group (15-24-years) relative to all other ages.
Previous studies suggest that adolescents and young adults may be
substituting online, digital social interaction for in-person, face-to-face
social engagement (Twenge et al., 2019; Twenge & Spitzberg, 2020).
Recent cohorts of adolescents and young adults will age with having
experienced less peer social engagement and companionship in their
youth than previous cohorts. The decline in social engagement with
Fig. 5. By Annual Family Income (adjusted for age, sex, race, and ethnicity): US Social Connectedness Trends, Annual Daily Average in Minutes, 2010–2020. $<25K
(blue), $25K-<50K (red), $50K-<100K (yellow), $≥100K (green).
Fig. 6. By Typical Work Hours per Week (adjusted for age and annual family income): US Social Connectedness Trends, Annual Daily Average in Minutes,
2010–2020. Zero hours (blue), 1–25 h (red), 26–50 h (yellow), 50–100 h. (green).
V.D. Kannan and P.J. Veazie

SSM - Population Health 21 (2023) 101331
8
friends and ‘others’ was not replaced by more social engagement with
family. Youth is when people tend to be more socially engaged with
friends, ‘others’, and in companionship than at any other time in life as
evident in our data. If, as research indicates, adolescence and young
adulthood are sensitive life-stages for socializing with non-family (Bla-
kemore & Mills, 2014), then the current youth cohort is experiencing
substantial loss in socialization experiences. Since social experiences in
older adulthood are a function of relational histories over the life-course
Table 1
Average number of daily minutes of social connectedness from 2003 to 2019
a
— means [95% condence intervals].
Group Characteristic Social Isolation Household Family Non–Household Family Friends All Others Companionship
Sex
Male 288 [286, 290] 232 [230, 235] 26 [25, 27] 54 [53, 55] 49 [49, 50] 204 [202, 206]
Female 294 [292, 296] 278 [276, 280] 40 [39, 41] 48 [46, 49] 46 [45, 47] 189 [187, 190]
Race
White 285 [282, 287] 265 [262, 268] 32 [31, 33] 50 [49, 52] 47 [46, 49] 201 [199, 203]
Black 344 [340, 347] 190 [187, 194] 42 [41, 44] 50 [48, 52] 47 [46, 49] 171 [168, 173]
Other 271 [266, 275] 264 [259, 269] 24 [22, 26] 54 [51, 56] 49 [47, 51] 177 [174, 181]
Ethnicity
Non-Hispanic 301 [298, 304] 250 [247, 254] 33 [32, 35] 52 [50, 53] 47 [46, 49] 196 [194, 199]
Hispanic 235 [232, 238] 288 [285, 292] 30 [28, 31] 45 [43, 46] 49 [47, 50] 195 [193, 197]
Age
15–24 years 234 [232, 237] 197 [194, 200] 31 [30, 32] 115 [114, 117] 83 [82, 85] 227 [225, 229]
25–34 years 211 [208, 213] 285 [282, 288] 27 [25, 28] 49 [48, 51] 50 [48, 51] 191 [189, 193]
35–44 years 229 [226, 232] 310 [307, 313] 22 [21, 23] 34 [33, 36] 40 [38, 41] 178 [176, 180]
45–54 years 290 [287, 293] 244 [241, 247] 34 [32, 35] 31 [29, 32] 38 [37, 40] 175 [173, 177]
55–64 years 351 [348, 354] 233 [230, 237] 42 [41, 43] 31 [29, 33] 37 [35, 38] 185 [183, 187]
≥65 years 442 [439, 444] 266 [263, 269] 43 [42, 44] 38 [37, 40] 35 [34, 37] 219 [217, 221]
Family Income
b
$ <25K 245 [242, 248] 130 [127, 134] 29 [28, 31] 100 [99, 102] 82 [81, 84] 183 [181, 186]
$ 25K - <50K 194 [191, 197] 166 [163, 170] 20 [19, 22] 96 [95, 98] 73 [72, 75] 188 [186, 191]
$ 50K - <100K 169 [167, 172] 179 [176, 182] 15 [14, 16] 94 [93, 96] 72 [71, 74] 187 [185, 189]
$ ≥100K 147 [145, 150] 198 [194, 201] 9 [8, 10] 97 [95, 99] 73 [71, 74] 185 [183, 188]
Hours of Work per Week
c
zero hours 287 [284, 290] 234 [231, 238] 44 [43, 46] 99 [98, 101] 80 [78, 81] 222 [220, 225]
1–25 h 234 [230, 238] 179 [174, 184] 40 [38, 42] 97 [94, 99] 84 [82, 86] 186 [182, 189]
26–50 h 180 [177, 183] 149 [145, 152] 34 [33, 36] 74 [73, 76] 69 [68, 71] 162 [159, 164]
>50 h 143 [137, 148] 105 [98, 111] 31 [28, 33] 69 [65, 72] 67 [64, 69] 131 [126, 135]
Note.
a
Except family income and hours worked per week which cover the years 2010–2019.
b
Analyses for family income are adjusted for age, sex, race, and ethnicity.
c
Analyses for hours worked per week are adjusted for age and family income.
Table 2
Trends in daily minutes of social connectedness from 2003 to 2019
a
— slopes [95% condence intervals].
Group Characteristic Social Isolation Household Family Non– Household Family Friends All Others Companionship
Sex
Male 29 [24, 34] −11 [–16, −5] −6 [–8, −4] −20 [–23, −17] −14 [–17, −12] −22 [–26, −18]
Female 12 [7, 16] −18 [–23, −13] −5 [–7, −2] −16 [–19, −14] −8 [–10, −6] −12 [–15, −8]
Race
White 17 [13, 20] −12 [–16, −7] −4 [–6, −2] −17 [–19, −15] −11 [–12, −9] −12 [–15, −9]
Black 33 [23, 43] −20 [–29, −10] −10 [–15, −6] −21 [–26, −16] −14 [–18, −10] −28 [–35, −20]
Other 36 [22, 51] −28 [–45, −12] −14 [–20, −8] −25 [–33, −16] −9 [–16, −1] −41 [–52, −30]
Ethnicity
Non-Hispanic 21 [17, 25] −12 [–16, −8] −6 [–8, −4] −20 [–22, −18] −9 [–11, −7] −14 [–17, −11]
Hispanic 35 [26, 43] −40 [–51, −30] −2 [–6, 2] −4 [–9, 1] −22 [–27, −18] −29 [–36, −22]
Age
15–24 years 31 [22, 40] −3 [–13, 8] −8 [–13, −4] −62 [–71, −54] −32 [–39, −25] −49 [–58, −41]
25–34 years 17 [10, 23] −45 [–55, −35] −5 [–8, −1] −5 [–10, 0] −5 [–9, −1] −18 [–25, −12]
35–44 years −18 [–25, −12] 9 [1, 18] −7 [–10, −4] −7 [–10, −4] −13 [–16, −10] −9 [–14, −3]
45–54 years −11 [–19, −3] −2 [–11, 6] −8 [–12, −5] −4 [–8, −1] −6 [–9, −3] −3 [–9, 3]
55–64 years 15 [5, 24] −39 [–49, −30] −8 [–13, −4] −9 [–12, −5] −3 [–7, 0] −22 [–29, −16]
≥65 years 3 [–7, 12] −5 [–14, 4] −5 [–9, −1] −11 [–14, −8] 2 [–1, 5] −7 [–13, 0]
Family Income
b
$ <25K 43 [24, 62] −69 [–88, −49] −21 [–30, −12] −32 [–41, −23] 12 [3, 20] −62 [–77, −48]
$ 25K - <50K 34 [17, 51] −22 [–41, −3] −13 [–20, −5] −32 [–41, −24] −6 [–13, 2] −30 [–44, −17]
$ 50K - <100K 40 [26, 54] −21 [–37, −5] −5 [–12, 1] −37 [–45, −29] −6 [–12, 1] −12 [–24, −1]
$ ≥100K 21 [6, 36] −13 [–32, 6] −6 [–12, 1] −45 [–55, −35] −9 [–16, −1] −22 [–36, −9]
Hours of Work per Week
c
zero hours 39 [24, 54] −29 [–45, −13] −15 [–22, −8] −39 [–47, −32] 3 [–3, 10] −36 [–48, −25]
1–25 h 34 [12, 56] −36 [–61, −10] 3 [–8, 13] −59 [–74, −45] 11 [–2, 24] −40 [–58, −21]
26–50 h 36 [26, 46] −40 [–53, −27] −9 [–14, −4] −28 [–35, −22] −13 [–18, −8] −31 [–40, −22]
>50 h 24 [2, 46] −9 [–38, 21] −17 [–28, −5] −27 [–41, −13] 4 [–7, 15] −5 [–26, 16]
Notes.
a
Except family income and hours worked per week which cover the years 2010–2019.
b
Analyses for family income are adjusted for age, sex, race, and ethnicity.
c
Analyses for hours worked per week are adjusted for age and family income.
V.D. Kannan and P.J. Veazie

SSM - Population Health 21 (2023) 101331
9
(Antonucci, Fiori, Birditt, & Jackey, 2010), reductions in friend and
‘other’ social engagement and in companionship for young people may
have health and longevity implications for this cohort in future years as
they age.
Having more leisure (non-work) time seems to allow for people to
allocate more time toward both social engagement and social isolation.
This is evident for people who work zero hours per week and for older
adults ≥65-years who are likely retired and, thus, working zero or
reduced hours. People who work zero hours per week had both the
highest magnitude of social isolation and the highest or second highest
magnitudes for all social engagement types and for companionship.
Similarly, those ≥65-years (and likely retired) had high levels of both
social isolation and companionship. Having larger amounts of leisure
time implies that the amount of social engagement and the amount of
social isolation can be generated more from personal decision making
rather than from externally imposed time constraints. Among those with
greater leisure time, this pattern of increased time allocation for both
engagement and isolation suggests that some amount of social isolation
is welcome and benecial to the individual. Time for oneself affords the
individual the opportunity to engage in self-care and personal interest
activities. Indeed, some amount of time spent with oneself, absent of
other people, is in alignment with self-care (Denyes, Orem, & Bekel,
2001; Levin & Idler, 1983). Thus, large amounts of social isolation
should not necessarily be viewed as detrimental in the absence of in-
formation about available leisure time and amount of social engage-
ment. However, it is important to note that older adults have additional
constraints beyond work imposed by declining health and disability.
Labor conditions should be studied as a structural constraint to social
connectedness. In fact, workers cite difculties with meaningful re-
lationships in recent labor strikes that have centered around long work
hours and mandatory overtime (Eidelson, October 25, 2021). And, the
surgeon general states that excessive work hours contributes to isola-
tion, but could be remedied by employers willing to protect workers’
time outside of work (McGregor, 2017). We recommend policy initia-
tives that disincentivize employers from extracting long work-hours or
paying wages low enough to require second and third jobs (thus,
increasing total work-hours).
Women spent more time with family but less time with friends and in
companionship than men. Different relationship types could potentially
impact health and longevity differently for different groups. For
example, among young adults, support from friends has the strongest
positive impact on mental health, strain from family has the strongest
negative impact, and friend support has a protective effect buffering
family strain (McLaughlin, Horwitz, & Raskin White, 2002; Obradovi´
c,
Tirado-Strayer, & Leu, 2013). For older adults, the quality of friend re-
lationships contributed more to life satisfaction than the quality of re-
lationships with their children (O’Connor, 1995). And, for older adults,
while family activities increase positive affect, it also increased negative
affect; whereas friend activities increased positive affect and decreased
negative affect and, further, also increased life satisfaction (Huxhold,
Miche, & Schüz, 2014). Additionally, the importance of friendships have
been increasing for recent generations of older adults (Fiori, Windsor, &
Huxhold, 2020). Role and normative expectations inuence individuals’
social relationships (Antonucci et al., 2010). Women’s social engage-
ment patterns may reect the social and biological expectations women
face regarding family life. Men’s social engagement patterns could
reect cultural norms around masculinity.
Nuances arise regarding differential allocation of social exposure and
consequent health outcomes. For example, Hispanics spend markedly
greater time with household family than non-Hispanics. This pattern
may be a consequence of Hispanic attitudinal and behavioral familism
(Cahill, Updegraff, Causadias, & Korous, 2021; Ruiz, 2005; Sabogal,
Marín, Otero-Sabogal, Marín, & Perez-Stable, 1987). Perhaps, family
relationships are more salubrious for women, whereas men may benet
more from time spent with friends. Or, possibly, women suffer from the
added stress of familial duties and obligations on top of less time spent in
voluntary associations, e.g., with friends and in companionship. Further
studies examining social connectedness tendencies and preferences
related to relationship types could illuminate the importance of rela-
tionship roles and cultural norms in explanations linking social exposure
to health and longevity.
Social connectedness research has focused heavily on older adults,
often characterizing late-life as socially isolated. Indeed, 20% of older
adults (approximately 6.4 million people) report being socially isolated,
while 1.3 million older adults are characterized as severely socially iso-
lated (Cudjoe et al., 2020). This study shows that both high social
isolation and high companionship levels characterize older adulthood.
This pattern potentially indicates an equilibrium between self-time (for
self-care or pursuit of one’s own preferred activities) and social-time
during retirement. And, at the same time that older adults experience
the loss of social network members, they also experience network
growth by cultivating new social ties, adding new condant relation-
ships, increased socializing with neighbors, and increased community
involvement (Cornwell, Goldman, & Laumann, 2021; Cornwell & Lau-
mann, 2015; Cornwell, Laumann, & Schumm, 2008). These changes in
the social networks of older adults may facilitate greater companion-
ship. Despite social network losses, when older adults cultivate new
condants, their mental and physical health improve (Cornwell &
Laumann, 2015). Socioemotional selectivity theory proposes that older
adults intentionally prune their social networks to create space for more
emotionally meaningful relationships (Carstensen, 2021) which could
increase their time spent in companionship.
Black Americans experienced both high social isolation and low so-
cial engagement. One structural explanation worth future investigation
is architectural exclusion. Black Americans sometimes live in ‘walled-
off’ neighborhoods; are often excluded from access to features of the
built environment that promote socialization such as parks, public pools,
and sidewalks; and, design elements such as bridges and one-way streets
are used to limit movement to and from black communities (Einhorn &
Lewis, July 19, 2021; Ka´
zmierczak, 2013; Leyden, 2003; Schindler,
2015; Travieso, 2020). Black Americans tend to experience greater
threat from the police (Alang, 2018; Alang, McAlpine, McCreedy, &
Hardeman, 2017). Simultaneously, black Americans also express greater
fear for their safety in their own neighborhoods — a fear that is, para-
doxically, deepened by greater neighborhood social capital (Roman &
Chaln, 2008). Thus, obstacles to social connection exist inside and
outside of black communities. Further, non-Hispanic blacks work
non-standard shifts (i.e., evenings, nights, and rotating or highly vari-
able work shifts) to a greater extent than their Hispanic or white
counterparts (Presser, 2003). As mentioned previously, labor conditions
might be an important constraint on social engagement. Thus, poten-
tially, a wide range of economic and social policies may be necessary to
improve social connectedness for this group.
The lowest income group had more social engagement with ‘others’
than higher income groups and was the only group to show a statistically
signicant, positive linear trend in social engagement with ‘others’. This
‘others’ category includes acquaintances, co-workers, neighbors, and
roommates, and could be an indicator of the degree to which individuals
are either pressed to tap into or have the leisure to engage socially with a
wide array of social connections. Most studies in social connectedness
investigate isolation or engagement with friends and family. Future
research on social engagement with ‘others’ could reveal information
about who taps into this social resource, why, and under which social
and economic conditions.
The steady decline of household family social engagement overtime
could be due to changes in marriage formation. On average, since the
baby boomers, Americans have married at increasingly later ages, if they
marry at all (Bloome & Ang, 2020). Declining marriage trends have been
especially steep for low-income individuals and for black Americans
across all economic backgrounds (Bloome & Ang, 2020). Our ndings
show that, on average, low-income Americans have lower and rapidly
declining household family social engagement than other income
V.D. Kannan and P.J. Veazie

SSM - Population Health 21 (2023) 101331
10
groups, and that black Americans have lower household family social
engagement than other race groups. Thus, the changing landscape of
marriage nationally and across sub-groups could contribute to differ-
ences in magnitudes and trends for household family social engagement.
5. Conclusion
Examining temporal trends in social connectedness nationally, we
see overall increases in time spent alone and overall decreases in time
spent with family, friends, others (roommates, neighbors, acquain-
tances, coworkers, clients, etc.), and in companionship. Thus, the
answer to the question, “is social connectedness improving over time?”
is a resounding, “no”. Overall, 2020 exacerbated these patterns. Sub-
group analysis showed that Black Americans experienced the greatest
overall disparity in social connectedness. Of importance to structural
solutions, less hours of work obligation allowed people to apportion
their time both for themselves and for social engagement. In fact, labor
conditions may be an important obstacle to both social engagement and
socially isolated time needed to care for oneself. Social isolation should
be studied with respect to total available leisure time, indicating the
total amount of personal time available for making time allocation de-
cisions. Social isolation should also be studied in relation to amount of
social engagement, alone time spent in self-care, and time spent in
personal interest activities, rather than as monolithically detrimental.
Future research could assess how relationship types contribute to health
and longevity and if sub-groups respond differently to different rela-
tionship types. Finally, digital media may be changing the socialization
dynamics of young people with the implications for social connection in
mid- and late-life to be observed in future decades.
Author statement
Viji Diane Kannan: Conceptualization, Methodology, Formal anal-
ysis, Writing - Original Draft, Writing - Review & Editing.
Peter Veazie: Methodology, Writing - Review & Editing, Supervision.
Funding
We have no nancial interests to disclose.
Ethics approval
There was no need for IRB approval as publicly available, secondary
data was used.
Declaration of competing interest
We have no conicts of interest.
Data availability
Data will be made available on request.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.ssmph.2022.101331.
References
Alang, S. (2018). The more things change, the more things stay the same: Race, ethnicity,
and police brutality. American Journal of Public Health, 108(9), 1127–1128.
Alang, S., McAlpine, D., McCreedy, E., & Hardeman, R. (2017). Police brutality and black
health: Setting the agenda for public health scholars. American Journal of Public
Health, 107(5), 662–665.
Antonucci, T. C., Ajrouch, K. J., & Birditt, K. S. (2014). The convoy model: Explaining
social relations from a multidisciplinary perspective. The Gerontologist, 54(1), 82–92.
Antonucci, T. C., Fiori, K. L., Birditt, K., & Jackey, L. M. (2010). Convoys of social
relations: Integrating life-span and life-course perspectives. In R. M. Lerner,
M. E. Lamb, & A. M. Freund (Eds.), The handbook of life-span development (Vol. 2, pp.
434–473). Hoboken, NJ: Wiley.
Atir, S., Wald, K. A., & Epley, N. (2022). Talking with strangers is surprisingly
informative. Proceedings of the National Academy of Sciences, 119(34), Article
e2206992119.
Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal
attachments as a fundamental human motivation. Psychological Bulletin, 117(3),
497–529.
Beckes, L., & Coan, J. A. (2011). Social baseline theory: The role of social proximity in
emotion and economy of action. Soc. Personality Psychol. Compass., 5(12), 976–988.
Blakemore, S. J., & Mills, K. L. (2014). Is adolescence a sensitive period for sociocultural
processing? Annual Review of Psychology, 65, 187–207.
Bloome, D., & Ang, S. (2020). Marriage and union formation in the United States: Recent
trends across racial groups and economic backgrounds. Demography, 57(5),
1753–1786.
Bureau of Labor Statistics. (2021). American time use survey user’s guide: Understanding
ATUS 2003 to 2020 (Retrieved from).
Cahill, K. M., Updegraff, K. A., Causadias, J. M., & Korous, K. M. (2021). Familism values
and adjustment among hispanic/latino individuals: A systematic review and meta-
analysis. Psychological Bulletin, 147(9), 947.
Carstensen, L. L. (2021). Socioemotional selectivity theory: The role of perceived endings
in human motivation. The Gerontologist, 61(8), 1188–1196.
Clark, D. M. T., Loxton, N. J., & Tobin, S. J. (2015). Declining loneliness over time:
Evidence from American colleges and high schools. Personality and Social Psychology
Bulletin, 41(1), 78–89.
Coan, J. A., & Sbarra, D. A. (2015). Social baseline theory: The social regulation of risk
and effort. Current opinion in psychology, 1, 87–91.
Cohen, S. (2021). Psychosocial vulnerabilities to upper respiratory infectious illness:
Implications for susceptibility to coronavirus disease 2019 (COVID-19). Perspectives
on Psychological Science, 16(1), 161.
Cornwell, B., Goldman, A., & Laumann, E. O. (2021). Homeostasis revisited: Patterns of
stability and rebalancing in older adults’ social lives. The Journals of Gerontology:
Series B, 76(4), 778–789.
Cornwell, B., & Laumann, E. O. (2015). The health benets of network growth: New
evidence from a national survey of older adults. Social Science & Medicine, 125,
94–106.
Cornwell, B., Laumann, E. O., & Schumm, L. P. (2008). The social connectedness of older
adults: A national prole. American Sociological Review, 73(2), 185–203.
Cudjoe, T. K., Roth, D. L., Szanton, S. L., Wolff, J. L., Boyd, C. M., & Thorpe, R. J., Jr.
(2020). The epidemiology of social isolation: National health and aging trends study.
The Journals of Gerontology: Serie Bibliographique, 75(1), 107–113.
Denyes, M. J., Orem, D. E., & Bekel, G. (2001). Self-care: A foundational science. Nursing
Science Quarterly, 14(1), 48–54.
Eidelson, J. (2021). ‘Suicide shifts,’ 7-day weeks fuel rare are-up. U.S. Strikes. Bloomberg.
Einhorn, E., Lewis, O., & July 19. (2021). Built to keep Black from white: Eighty years
after a segregation wall rose in Detroit, America remains divided. That’s not an
accident. by Erin Einhorn and Olivia Lewis, July 19, 2021. Retrieved from https
://www.nbcnews.com/specials/detroit-segregation-wall/.
Fiori, K. L., Windsor, T. D., & Huxhold, O. (2020). The increasing importance of
friendship in late life: Understanding the role of sociohistorical context in social
development. Gerontology, 66(3), 286–294.
Fothergill, K. E., Ensminger, M. E., Robertson, J., Green, K. M., Thorpe, R. J., & Juon, H.-
S. (2011). Effects of social integration on health: A prospective study of community
engagement among african American women. Social Science & Medicine, 72(2),
291–298.
Gardiner, C., Geldenhuys, G., & Gott, M. (2018). Interventions to reduce social isolation
and loneliness among older people: An integrative review. Health and Social Care in
the Community, 26(2), 147–157.
Hakulinen, C., Pulkki-Råback, L., Virtanen, M., Jokela, M., Kivim¨
aki, M., & Elovainio, M.
(2018). Social isolation and loneliness as risk factors for myocardial infarction,
stroke and mortality: UK biobank cohort study of 479 054 men and women. Heart,
104(18), 1536–1542.
Heffner, K. L., Waring, M. E., Roberts, M. B., Eaton, C. B., & Gramling, R. (2011). Social
isolation, C-reactive protein, and coronary heart disease mortality among
community-dwelling adults. Social Science & Medicine, 72(9), 1482–1488.
Heuser, C., & Howe, J. (2019). The relation between social isolation and increasing
suicide rates in the elderly. Quality in Ageing and Older. Adults, 20(1), 2–9.
Hodge, J. G., White, E. N., & Reeves, C. M. (2020). Legal and policy interventions to
address social isolation. Journal of Law Medicine & Ethics, 48(2), 360–364.
Holt-Lunstad, J. (2018). Why social relationships are important for physical health: A
systems approach to understanding and modifying risk and protection. Annual
Review of Psychology, 69, 437–458.
Holt-Lunstad, J. (2020a). The double pandemic of social isolation and COVID-19: Cross-
sector policy must address both. Health Affairs Blog, 22.
Holt-Lunstad, J. (2020b). Social isolation and health. Health affairs brief.
Holt-Lunstad, J., Smith, T. B., Baker, M., Harris, T., & Stephenson, D. (2015). Loneliness
and social isolation as risk factors for mortality: A meta-analytic review. Perspectives
on Psychological Science, 10(2), 227–237.
Holt-Lunstad, J., Smith, T. B., & Layton, J. B. (2010). Social relationships and mortality
risk: A meta-analytic review. PLoS Medicine, 7(7), Article e1000316.
Homan, P., Brown, T. H., & King, B. (2021). Structural intersectionality as a new
direction for health disparities research. Journal of Health and Social Behavior, 62(3),
350–370.
V.D. Kannan and P.J. Veazie

SSM - Population Health 21 (2023) 101331
11
Huxhold, O., Miche, M., & Schüz, B. (2014). Benets of having friends in older ages:
Differential effects of informal social activities on well-being in middle-aged and
older adults. Journals of Gerontology Series B: Psychological Sciences and Social
Sciences, 69(3), 366–375.
Ka´
zmierczak, A. (2013). The contribution of local parks to neighbourhood social ties.
Landscape and Urban Planning, 109(1), 31–44.
Levin, L. S., & Idler, E. L. (1983). Self-care in health. Annual Review of Public Health, 4(1),
181–201.
Leyden, K. M. (2003). Social capital and the built environment: The importance of
walkable neighborhoods. American Journal of Public Health, 93(9), 1546–1551.
Marczak, J., Wittenberg, R., Doetter, L. F., Casanova, G., Golinowska, S., Guillen, M.,
et al. (2019). Preventing social isolation and loneliness among older people.
Eurohealth, 25(4), 3–5.
Markowitz, S., Adams, E. K., Dietz, P. M., Tong, V. T., & Kannan, V. (2013). Tobacco
control policies, birth outcomes, and maternal human capital. Journal of Human
Capital, 7(2), 130–160.
Marsden, P. V., & Srivastava, S. B. (2012). Trends in informal social participation,
1974–2008. In Social trends in American life: Findings from the general social survey
since 1972 (pp. 240–263).
McGregor, J. (2017). This former surgeon general says there’s a ‘loneliness epidemic’and
work is partly to blame. Washington Post, 10(4).
McLaughlin, J., Horwitz, A. V., & Raskin White, H. (2002). The differential importance of
friend, relative and partner relationships for the mental health of young adults. In
J. A. Levy, & B. A. Pescosolido (Eds.), Social networks and health (advances in medical
sociology (Vol. 8). Bingley: Emerald Group Publishing Limited.
McPherson, M., Smith-Lovin, L., & Brashears, M. E. (2006). Social isolation in America:
Changes in core discussion networks over two decades. American Sociological Review,
71(3), 353–375.
McPherson, M., Smith-Lovin, L., & Brashears, M. (2008). The ties that bind are fraying.
Contexts, 7(3), 32.
National Academies of Sciences Engineering And Medicine. (2020). Social isolation and
loneliness in older adults: Opportunities for the health care system. National Academies
Press.
February National Cancer Institute. (2022). Joinpoint trend analysis software version
4.9.0.1. Statistical Methodology and applications branch. Surveillance Research Program.
Retrieved from https://surveillance.cancer.gov/joinpoint/.
O’Connor, B. P. (1995). Family and friend relationships among older and younger adults:
Interaction motivation, mood, and quality. The International Journal of Aging and
Human Development, 40(1), 9–29.
Obradovi´
c, J., Tirado-Strayer, N., & Leu, J. (2013). The importance of family and friend
relationships for the mental health of Asian immigrant young adults and their
nonimmigrant peers. Research in Human Development, 10(2), 163–183.
Penninkilampi, R., Casey, A.-N., Singh, M. F., & Brodaty, H. (2018). The association
between social engagement, loneliness, and risk of dementia: A systematic review
and meta-analysis. Journal of Alzheimer’s Disease, 66(4), 1619–1633.
Presser, H. B. (2003). Race-ethnic and gender differences in nonstandard work shifts.
Work and Occupations, 30(4), 412–439.
Roman, C. G., & Chaln, A. (2008). Fear of walking outdoors: A multilevel ecologic
analysis of crime and disorder. American Journal of Preventive Medicine, 34(4),
306–312.
Rook, K. S. (1987). Social support versus companionship: Effects on life stress, loneliness,
and evaluations by others. Journal of Personality and Social Psychology, 52(6), 1132.
Rook, K. S., & Ituarte, P. H. (1999). Social control, social support, and companionship in
older adults’ family relationships and friendships. Personal Relationships, 6(2),
199–211.
Ruiz, E. (2005). Hispanic culture and relational cultural theory. Journal of Creativity in
Mental Health, 1(1), 33–55.
Sabogal, F., Marín, G., Otero-Sabogal, R., Marín, B. V., & Perez-Stable, E. J. (1987).
Hispanic familism and acculturation: What changes and what doesn’t? Hispanic
Journal of Behavioral Sciences, 9(4), 397–412.
Sandstrom, G. M., & Dunn, E. W. (2014). Is efciency overrated? Minimal social
interactions lead to belonging and positive affect. Social Psychological and Personality
Science, 5(4), 437–442.
Schell, R. C., Allen, B., Goedel, W. C., Hallowell, B. D., Scagos, R., Li, Y., … Cerda, M.
(2021). Identifying predictors of opioid overdose death at a neighborhood level with
machine learning. American Journal of Epidemiology, 191(3), 526–533.
Schindler, S. (2015). Architectural exclusion: Discrimination and segregation through
physical design of the built environment. The Yale Law Journal, 1934–2024.
Schroeder, J., Lyons, D., & Epley, N. (2022). Hello, stranger? Pleasant conversations are
preceded by concerns about starting one. Journal of Experimental Psychology: General,
151(5), 1141.
Shankar, A., McMunn, A., Banks, J., & Steptoe, A. (2011). Loneliness, social isolation,
and behavioral and biological health indicators in older adults. Health Psychology, 30
(4), 377.
Shankar, A., McMunn, A., Demakakos, P., Hamer, M., & Steptoe, A. (2017). Social
isolation and loneliness: Prospective associations with functional status in older
adults. Health Psychology, 36(2), 179.
Sorkin, D., Rook, K. S., & Lu, J. L. (2002). Loneliness, lack of emotional support, lack of
companionship, and the likelihood of having a heart condition in an elderly sample.
Annals of Behavioral Medicine, 24(4), 290–298.
Tomasello, M. (2014). The ultra-social animal. European Journal of Social Psychology, 44
(3), 187–194.
Travieso, C. (2020). A nation of walls. Places Journal. https://placesjournal.org/article
/a-nation-of-walls/. Accessed 2022.
Trout, D. L. (1980). The role of social isolation in suicide. Suicide and Life-Threatening
Behavior, 10(1), 10–23.
Twenge, J. M., Haidt, J., Blake, A. B., McAllister, C., Lemon, H., & Le Roy, A. (2021).
Worldwide increases in adolescent loneliness. Journal of Adolescence, 93, 257–269.
Twenge, J. M., & Spitzberg, B. H. (2020). Declines in non-digital social interaction among
Americans, 2003–2017. Journal of Applied Social Psychology, 50(6), 363–367.
Twenge, J. M., Spitzberg, B. H., & Campbell, W. K. (2019). Less in-person social
interaction with peers among US adolescents in the 21st century and links to
loneliness. Journal of Social and Personal Relationships, 36(6), 1892–1913.
Umberson, D., & Karas Montez, J. (2010). Social relationships and health: A ashpoint
for health policy. Journal of Health and Social Behavior, 51(1_suppl), S54–S66.
United States Congress Joint Economic Committee. (2017). Social capital project.
Retrieved from https://www.jec.senate.gov/public/index.cfm/republicans/s
ocialcapitalproject.
Valtorta, N. K., Kanaan, M., Gilbody, S., Ronzi, S., & 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.
Van Lange, P. A., & Columbus, S. (2021). Vitamin S: Why is social contact, even with
strangers, so important to well-being? Current Directions in Psychological Science, 30
(3), 267–273.
Yang, Y. C., Boen, C., Gerken, K., Li, T., Schorpp, K., & Harris, K. M. (2016). Social
relationships and physiological determinants of longevity across the human life
span. Proceedings of the National Academy of Sciences, 113(3), 578–583.
V.D. Kannan and P.J. Veazie