Introduction
Social connection is vital to health and well-being. Isolation and loneliness are related, yet distinct measures of social disconnection. Social isolation is defined “as the objective lack or paucity of social contacts and interactions with family members, friends or the wider community” (Valtorta and Hanratty, Reference Valtorta and Hanratty2012). Loneliness is the subjective negative perception of being alone or disconnected from others (Hawkley and Cacioppo, Reference Hawkley and Cacioppo2010). Social disconnection is associated with increased mortality and poor physical and mental health outcomes (Cacioppo et al., Reference Cacioppo, Hawkley and Thisted2010; Courtin and Knapp, Reference Courtin and Knapp2017; Hawkley and Cacioppo, Reference Hawkley and Cacioppo2010; Holt-Lunstad et al., Reference Holt-Lunstad, Smith and Layton2010; Leigh-Hunt et al., Reference Leigh-Hunt2017; Luanaigh and Lawlor, Reference Luanaigh and Lawlor2008). Estimates suggest that both isolation and loneliness are increasing in prevalence among adults (Varrella, Reference Varrella2021) and are serious threats to public health on par with or even exceeding obesity, physical inactivity, smoking, and air pollution (Holt-Lunstad, et al., Reference Holt-Lunstad, Smith and Layton2010).
Disruption of social connection due to the physical distancing recommendations to mitigate the spread of the novel coronavirus disease 2019 (COVID-19) brought increased attention to social isolation and loneliness among all age groups. The “stay-at-home” orders and social distancing guidelines during the initial phase of the pandemic resulted in the abrupt cancellation, postponement, or modification of opportunities for social interaction that provide companionship, support, and resources (Smith et al., Reference Smith, Steinman and Casey2020). Thus, a paradox emerged wherein interventions to lower risk of exposure to the coronavirus at the same time increased risk for isolation and loneliness (Smith et al., Reference Smith, Steinman and Casey2020). Due to greater risk of complications and mortality if infected with the novel coronavirus (Centers for Disease Control and Prevention, 2021), along with other characteristics (e.g. more likely to live alone or to be retired or physical impaired; Holt-Lunstad, Reference Holt-Lunstad2017), older adults were more likely to be isolated and/or choose to give up social activities to minimize risk of exposure.
This report focuses on older adults assessed during the first 120 days of pandemic-related restrictions in 2020, which included the most restrictive periods of the pandemic to-date beginning with the state of Pennsylvania’s Governor’s “stay-at-home” order on March 23, 2020, and including the “red” and “yellow” phases of restrictions through June 5, 2020, in Allegheny County, PA where the study participants reside (https://www.governor.pa.gov/process-to-reopen-pennsylvania/). We sought to examine the psychosocial health effects of pandemic-related restrictions on older adults during this period by specifically describing the characteristics of those who felt lonely and the cross-sectional associations of feelings of loneliness with activity disruption (potential predictors/antecedents of loneliness) and depression and anxiety (potential outcomes of loneliness).
Methods
Study design and participants
The Monongahela-Youghiogheny Healthy Aging Team (MYHAT), based in Allegheny County, Pennsylvania, follows an age-stratified randomly selected population-based cohort annually for the development of mild cognitive impairment and dementia. This report includes MYHAT participants from the original cohort aged 65+ years enrolled between 2006 and 2008 (Ganguli et al., Reference Ganguli, Snitz, Vander Bilt and Chang2009) as well as a new, younger subcohort aged 65–74 years enrolled between 2016 and 2019 meeting the same eligibility criteria. At study entry and annually thereafter, participants complete a detailed assessment including basic demographic information, health history and lifestyle behaviors, health service utilization, self-report of memory function, a battery of neuropsychological tests, mental health and psychosocial well-being, a neurological exam, medication review, and blood pressure, as previously described (Ganguli et al., Reference Ganguli, Hughes, Jia, Lingler, Jacobsen and Chang2020).
In March 2020, after “stay-at-home” orders were imposed by the Governor of the state of Pennsylvania, a telephone survey was added to the MYHAT protocol to assess the impact of the COVID-19 pandemic restrictions on emotional health, activities, services, and general well-being. The survey was administered by trained interviewers, along with other select segments of the study assessment that were modified from in-person to telephone administration. The COVID-19 survey and all study procedures were approved by the University of Pittsburgh Institutional Review Board; all participants had previously provided written informed consent.
This study focuses on data from the COVID-19-related telephone survey and depressive and anxiety symptoms collected during the first 120 days of the pandemic-related restriction period between March 23 and July 20, 2020.
Measures
Feelings of loneliness
Participants were asked “In general, to what extent are the restrictions making you feel lonely?” Response choices included “not at all,” “somewhat,” or “to a great extent.” We combined “somewhat “and “to a great extent” into one category and compared it to “not at all” based on the distribution of responses.
Disruption of social activities
Participants reported which specific activities from a list that were disrupted due to the restrictions, including having to give up or change (i.e. do in a different way, such as virtually). These activities included working, volunteering, in-person visits, or group socializing with family or friends living outside the household, attending church or religious services, going out for entertainment, exercising in a gym or other facility, walking outdoors, going to restaurants and bars, nonessential shopping, and planned travel. Participants who did not engage in a given activity before the pandemic indicated that the question did not apply and were excluded from the analyses for that activity.
Feelings of depression and anxiety
Participants reported the extent (not at all, somewhat, to a great extent) that the pandemic-related restrictions made them feel depressed or nervous/anxious. The “somewhat” and “to a great extent” responses were combined and compared to “not at all.”
The study annual assessment also includes a modified version of the Center for Epidemiological Studies Depression scale (mCES-D; Radloff, Reference Radloff1977) to measure number of depressive symptoms in the past week (1 = present, 0 = absent; range 0–20), and the generalized anxiety disorder assessment (GAD-7; Spitzer et al., Reference Spitzer, Kroenke, Williams and Löwe2006) to measure frequency of anxiety symptoms in the past two weeks (0 = not at all, 1 = several days, 2 = more than half the days, 3 = nearly every day; range 0–21). We categorized depressive symptoms as <4 vs. ≥4 and anxiety symptoms as <7 vs. ≥7, which is the 90th percentile of this sample for each measure.
Statistical analyses
We first describe the proportion of participants who felt lonely, who experienced disruption (either gave up or changed) in each social activity after excluding those who did not participate in that activity before the pandemic, and who felt depressed or anxious due the pandemic restrictions, as well as the proportion who had ≥4 depressive and ≥7 anxiety symptoms. We used χ2 analyses or Fisher’s exact test to characterize differences by sociodemographic measures (age, sex, education, race, marital status, and living arrangement) for these measures. Logistic regression models, adjusted for age, sex, race, and education, were used to examine the association between disruption in each social activity and feeling lonely among those who engaged in the activity prior to the pandemic. The odds of feeling lonely (somewhat/to a great extent vs. not at all) in relation to having to give up or change engagement in each social activity was compared to no change (reference group). Since the size of the effect may differ depending on if the activity was changed or given up, we examined each separately relative to no change in engagement. Similarly, logistic regression analyses, adjusted for age, sex, race, and education, were used to examine if the participant’s perception of feeling lonely due to the restrictions was associated with depression or anxiety, in four separate models. Two models had depression as the outcome (i) based on the extent that the pandemic restrictions made participants feel depressed (not at all vs. somewhat or great extent) and (ii) the mCES-D binary score (<4 vs. ≥4 symptoms). Two models had anxiety as the outcome (iii) based on the extent that the pandemic restrictions made participants feel anxious (not at all vs. somewhat or great extent) and (iv) the GAD-7 binary score (<7 vs. ≥7 symptoms). Analyses were completed using R statistical package, version 3.6.3 (R Core Team, 2020).
Results
A total of 360 participants’ annual assessments occurred in the first 120 days of pandemic restrictions (March 23–July 20, 2021) in Southwestern, Pennsylvania. Each participant completed the modified assessment over the telephone, including the newly added COVID-19 survey, once during this period. Among these participants, 198 (55%) were 65–74, 102 (28%) were 75–84, and 60 (17%) were 85 + years old; 221 (61%) were female; 227 (63%) had completed more than high school education (vs. ≤ high school); and 334 (93%) identified as White (vs. non-White). Partial data were available for marital status (n = 281) and living arrangement (n = 280) where 44.8% were married (vs. unmarried) and 39.3% lived alone (vs. not alone).
Sociodemographic characteristics of participants feeling lonely due to the pandemic-related restrictions
Feeling at least somewhat lonely due to the restrictions was reported by 131 (36.4%) of participants with 13 (3.6%) reporting that they felt lonely to a great extent. Females, those not currently married, and those living alone were significantly more likely to feel lonely compared to males, those who were currently married, and living with others, respectively (Table 1).
Sociodemographic characteristics of participants who reported feeling lonely in the first 120 of the COVID-19 pandemic-related restrictions
p-value estimated using χ2 analyses or Fisher’s exact test. p-value <0.05 considered statistically significant.
Sociodemographic characteristics of participants who reported disruption of social activities due to pandemic-related restrictions and the association with feeling lonely
The majority [334 (92.8%)] of participants reported that the recommendation for social distancing changed the way that they spent their time. Among those who engaged in each activity prior to the pandemic, we found that changing or giving up going to restaurants and bars and going out for entertainment was reported by nearly all participants, with attending church, visiting friends, exercising, shopping for nonessential items, volunteering, and visiting family also being given up or changed by the majority of participants. Fewer participants gave up or changed work, travel plans, or walking outdoors. We found no differences in disruption of any social activities by sociodemographic characteristics (Supplemental Table 1).
Table 2 shows the proportion of those who gave up or changed their engagement in each activity that also felt lonely due to the restrictions. Results from the adjusted logistic regression models show that giving up in-person visits with family was significantly associated with an approximately twofold higher odds of feeling lonely due to the restrictions, compared to no change in this activity. Disruption in other activities had similar or larger effect sizes in relation to feeling lonely, but were not statistically significant. These included having to give up or change visiting with friends in-person and giving up walking outdoors (Table 2).
Association between disruption of social activities (vs. no change) and feeling lonely in the first 120 days of COVID-19 pandemic-related restrictions
^Logistic regression adjusted for age, gender, education, and race.
Sociodemographic characteristics of participants feeling depressed or anxious and the association with feeling lonely
Table 3 shows that among all 360 participants, 71 (19.7%) felt depressed and 112 (31.1%) felt anxious/nervous due to the restrictions. Regarding depressive and anxiety symptoms in the preceding weeks, 50 (13.9%) participants had 4 or more depressive symptoms in the past week (mean 1.42 (SD 2.41)), and 37 (10.3%) participants had scores of 7 or higher for anxiety (mean 2.39 (SD 3.06)). No significant differences were found by sociodemographic characteristics in feeling depressed or anxious either due to restrictions or when measured with the mCES-D or GAD-7 (χ2 or Fisher’s Exact test; data not shown). However, both depressive feelings and anxiety were associated with significantly higher odds, with about equal strength, of feeling lonely in models adjusted for age, sex, education, and race (Table 3).
Associations between feeling lonely in the first 120 days of the COVID-19 pandemic-related restrictions and depression and anxiety
^Logistic regression adjusted for age, sex, education, and race.
mCES-D = modified Center for Epidemiologic Studies; GAD-7 = Generalized Anxiety Disorder Assessment-7.
Discussion
In this population-based study of older adults from a small town region in Southwestern, Pennsylvania, USA, we found over a third felt lonely due to restrictions implemented to reduce the spread of the novel SARS-CoV-2 virus during the first 120 days of the pandemic beginning with the Governor’s stay-at-home order. Females, those not currently married, and those living alone were more likely to report feelings of loneliness due to the pandemic restrictions compared to their comparison groups. Disruption of in-person visits with family was associated with feelings of loneliness, and feeling lonely was concurrently associated with feelings of depression and anxiety. While our cross-sectional study design does not allow inferences about the direction of the associations between these measures, these results do provide evidence of the extent that feelings of loneliness, depression, and anxiety as well as and disruption in social activities were related to the pandemic restrictions.
Prior to the pandemic, there were sociodemographic differences in loneliness among older adults (Dahlberg et al., Reference Dahlberg, Agahi and Lennartsson2018). We set out to examine if these differences persisted or changed in the early phase of the pandemic restrictions. In our sample, loneliness was not equally distributed across sociodemographic groups. This is similar to pre-pandemic differences (Dahlberg et al., Reference Dahlberg, Agahi and Lennartsson2018) and aligns with the work of others that the COVID-19 pandemic affected some older adults more than others, including females and those living alone (AARP Foundation & United Health Foundation, 2020; Seifert and Hassler, Reference Seifert and Hassler2020). However, one study reported that men had greater odds of loneliness compared to women during the pandemic (Choi et al., Reference Choi, Farina, Wu and Ailshire2021). We found no age differences in loneliness, comparing the 65–74, 75–84, and 85+ years age groups within this sample of adults. Prepandemic studies of loneliness in later life suggest that loneliness is more likely to affect the oldest-old (Beam and Kim, Reference Beam and Kim2020) and that this may persist into the pandemic (Frenkel-Yosef et al., Reference Frenkel-Yosef, Maytles and Shrira2020). However, it has also been suggested that the oldest-old were less likely to feel lonely due to the restrictions (or to experience an increase in feeling lonely) since their engagement in social activities was already low before the pandemic (Luchetti et al., Reference Luchetti2020). Although our sample did not include adults younger than 65 years, it is noteworthy that reports suggest that pandemic-related loneliness was higher among younger adults compared to older adults (Losada-Baltar et al., Reference Losada-Baltar, Jiménez-Gonzalo, Gallego-Alberto, del Pedroso-Chaparro, Fernandes-Pires and Márquez-González2021), which may due to greater resilience among older adults (Fuller and Huseth-Zosel, Reference Fuller and Huseth-Zosel2021) as compared to younger individuals. The overall proportion feeling lonely is within the range of other studies examining loneliness among older adults (Choi et al., Reference Choi, Farina, Wu and Ailshire2021; Fierloos et al., Reference Fierloos2021; Polenick et al., Reference Polenick, Perbix, Sawli, Maust, Birditt and Brooks2021) and is lower than reports of the proportion of younger adults feeling lonely during the early pandemic (Varma et al., Reference Varma, Junge, Meaklim and Jackson2021).
We also examined the extent that social activities were disrupted by the restrictions and whether disruption in any particular activity was associated with feeling lonely during this most restrictive period of the pandemic. In our sample, we found that those who gave up visiting with family in-person were more likely to feel lonely compared to those who did not change this activity. Although the effects for disruption in visiting friends in person and walking outdoors did not reach statistical significance in relation to feeling lonely, the size of the effects are noteworthy and should be followed up with in a larger cohort. The nationally representative Understanding America Study (UAS) found that canceling or postponing social activities was associated with increased feelings of loneliness among adults aged 50 years and older, but not all types of social restrictions were related to loneliness (Choi et al., Reference Choi, Farina, Wu and Ailshire2021). Similarly, we found that disruption in some, but not all activities, is associated with loneliness. Having to give up in-person visits with family was especially important during this restrictive period (Frenkel-Yosef et al., Reference Frenkel-Yosef, Maytles and Shrira2020). Social interactions with close ties may be particularly meaningful for older adults to feel connected (Carstensen, Reference Carstensen1995; Taylor, Reference Taylor2020). Having to give up walking outdoors may also provide a sense of connection with others through casual encounters with “weak” or “peripheral” ties (Huxhold et al., Reference Huxhold, Fiori, Webster and Antonucci2020). In exploratory analyses, we also found that giving up or changing shopping for nonessential items was significantly associated with feeling socially isolated due to the pandemic restrictions in a logistic regression model adjusted for age, sex, education, and race (OR 2.58, p = 0.002). Similar to walking outdoors, exchanging social pleasantries through casual encounters while shopping may lower risk of feeling disconnected or lonely. Moreover, these encounters provide exchanges with a variety of social ties that is linked to better well-being in late life (Fingerman et al., Reference Fingerman, Huo, Charles and Umberson2020; Holt-Lunstad et al., Reference Holt-Lunstad, Smith and Layton2010; Thomas, Reference Thomas2012). Future work should examine if disruption in different types of activities is differentially associated with emotional loneliness related to the loss of intimate relationships versus social loneliness related to the loss of social network (Lampraki et al., Reference Lampraki, Hoffman, Roquet and Jopp2022; Weiss, Reference Weiss1973).
Evidence suggests that loneliness is one of the strongest predictors of depression, anxiety, and post-traumatic stress disorder during the pandemic (González-Sanguino et al., Reference González-Sanguino2020; Kovacs et al., Reference Kovacs, Caplan, Grob and King2021; Palgi et al., Reference Palgi2020), but this evidence comes primarily from studies conducted in younger samples (Sampogna et al., Reference Sampogna2021; Weissbourd et al., Reference Weissbourd, Batanova, Lovison and Torres2021) and may be different for older adults (Vahia et al., Reference Vahia, Jeste and Reynolds2020; Xiong et al., Reference Xiong2020). Therefore, we examined the associations of feeling lonely due to the pandemic restrictions with depression and anxiety among adults aged 65 years and older residing in a lower socioeconomic, small-town region in Southwestern, Pennsylvania. Our results support an association between feeling lonely and feeling depressed or anxious due to restrictions during this most restrictive period of the pandemic. These results, along with others (Kotwal et al., Reference Kotwal2021; Polenick et al., Reference Polenick, Perbix, Sawli, Maust, Birditt and Brooks2021), highlight the potentially negative short-term impact of the COVID-19 social activity restrictions on symptoms of depression and anxiety for some older adults. The proportion feeling depressed or anxious due to the pandemic restrictions in this sample also aligns with national data where one in five older adults (aged 50–80 years) reported worse depression or sadness and 28% worse anxiety or worry (Gerlach et al., Reference Gerlach, Solway, Singer, Kullgren, Kirch and Malani2021). Further, since loneliness could affect longer term cognitive and physical health among older adults through pathways related to depression or anxiety (Holt-Lunstad, Reference Holt-Lunstad2017), future studies should examine if older adults who felt lonely in the early phases of the pandemic are at greater risk for poor physical and cognitive health outcomes over time (Vahia et al., Reference Vahia, Jeste and Reynolds2020).
We focused our analyses on participants whose annual assessments fell during the first 120 days of the Pennsylvania stay-at-home order since this included, but was not limited to, the most restrictive period to-date of the COVID-19 pandemic. Even within this brief period, we may have missed variations in the study measures related to changing restrictions both within and among participants since each participant was only assessed once. It is possible that the effect of the restrictions on the study measures could have worsened over the 120 days or, alternatively, could have weakened due to adaption over this time period. Due to the telephone assessment format during the period described here, we were unable to administer our standard in-person cognitive assessments, including the full health assessment, and some previously assessed variables (e.g. living arrangement) were not included from the beginning of modifying the assessment. Nonetheless, focusing on this time period provides a useful snapshot that sheds light on the emotional well-being of socially restricted older adults in this socioeconomically distressed small-town area. This snapshot will also serve as a backdrop for studying longer term outcomes of the pandemic restrictions in this population.
A strength of this work is that we assessed the specific impact of the pandemic-related restrictions on feeling lonely, engagement in social activities, and feeling depressed and anxious, using focused questions rather than instruments that more broadly assess these constructs and could be partially unrelated to the pandemic. It is possible that this method underestimated the impact of restrictions if participants did not recognize or want to admit the extent that the pandemic was affecting them. However, the opposite may also have been true since the pandemic forced isolation universally and may have reduced the stigma of loneliness and participants may have expected to be lonely (Van der Velpen et al., Reference Van der Velpen2022). The social activities included in these analyses varied in the degree to which they could involve social interaction with others, where some (e.g. exercise) could be done alone while others (e.g. walking outdoors, and shopping) could be more representative of being connected with the community and society. These findings should be interpreted in the context of sample characteristics as our results likely do not generalize to clinical samples, older adults in higher-income communities with different resources and environments, those in more ethnically diverse communities, or to those in congregate living situations. For example, the relatively lower level of depression and anxiety as measured by the mCES-D and GAD-7 scales, respectively, is typical for older adults living in the community, but cannot be generalized to older adults in other settings. Finally, these being cross-sectional data, we cannot determine the directions of the observed associations, e.g. whether loneliness led to feelings of depression or anxiety or feelings of anxiety or depression led to feeling lonely. However, these and other similar findings raise concern about the potential negative effects of social disconnection due to the pandemic restrictions among older adults (Killgore et al., Reference Killgore, Cloonan, Taylor and Dailey2020). Ongoing follow-up of the entire cohort using the full in-person assessment will allow us to address these larger questions.
In conclusion, the restrictions during the early days of the COVID-19 pandemic in Pennsylvania restricted the social activities of over 90% of our community-dwelling older adult sample, but only about a third reported feeling lonely. Loss of in-person contact with family and friends as well as peripheral ties with the community was likely linked to feelings of loneliness. Those who felt lonely were also more likely to also feel depressed or anxious due to the pandemic restrictions and to have more depressive or anxiety symptoms in the preceding weeks. Those in whom pandemic mitigation strategies increased feelings of loneliness may be at risk for both short- and long-term deleterious effects on health. These individuals may be a target for interventions, such as providing alternative means of socialization, e.g. telephone check-ins and virtual events for isolated older adults. Planning for future pandemics and other natural disasters should consider the emotional toll of social restrictions, especially for select subgroups of the population, and weigh the potential harm versus benefits of these restrictions when designing and implementing health and social policies.
Acknowledgements
Thank you to all MYHAT staff for recruitment, data collection, and data management and to the older adults who participated in the study and made this work possible.
Conflict of interest
None.
Source of funding
The Monongahela-Youghiogheny Healthy Aging Team study is supported in part by the National Institute on Aging under research grant R37#AG023651 (PI: Mary Ganguli).
IRB
The study was approved by the University of Pittsburgh Institutional Review Board protocol number 19040058.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1041610222000977
