![ORCID Icon]({"type":"image","src":"/templates/jsp/_style2/_tandf/images/orcid.svg"} "ORCID Icon"){kind=small}
https://orcid.org/0000-0001-5493-9636
ABSTRACT
ABSTRACT
This study aimed to determine the prevalence of and associations between anxiety, social isolation, and loneliness in a sample of older adults in Northern California during the stay-at-home order enacted during the COVID-19 pandemic. 514 older adults completed a 24-item survey. Perceived isolation and loneliness were reported in 56.4% and 36.0% of participants, respectively. Loneliness was found to be associated with both social isolation and COVID-19-related anxiety; however, social isolation and coronavirus-related anxiety were unrelated. Healthcare providers, social service providers, and families are encouraged to maintain or increase contact with older adults during the COVID-19 pandemic.
Introduction
Social isolation and loneliness are important issues and concerns for community-dwelling older adults because of their effects on overall health. Both loneliness and isolation have been shown to negatively affect one’s health (Coyle & Dugan, 2012), including increased risk of stroke and coronary heart disease (Valtorta et al., 2016), and increased likelihood of mortality (Holt-Lunstad et al., 2015). These documented negative sequelae are important for healthcare professionals, public health professionals, and community service providers to be aware of in order to identify older adults at risk of social isolation and loneliness and/or take action to reduce prevalence rates in this population.
Although commonly discussed alongside each other, the two constructs are distinct: Social isolation often refers to whether an individual is a member of a social network (e.g., through living with another person and/or participating in social activities). Conversely, loneliness refers to a perceived state of isolation, in that a person is not receiving the quality and/or quantity of social interactions they require (Cacioppo et al., 2014). Prevalence rates of social isolation and loneliness in older adults are commonly reported to be around 20–25% during daily life (e.g., Fokkema et al., 2012; Theeke, 2009; Victor & Yang, 2012), although a recent study found that social isolation and loneliness prevalence estimates were as low as 7.3% and 11.4%, respectively, in a national sample of Canadian adults aged 65 years and older (Menec et al., 2019), and 3% in English adults aged 65 years and older (Office for National Statistics, 2018).
In 2019, a novel coronavirus, COVID-19, emerged from Wuhan, China. The typical clinical presentation of COVID-19 was largely respiratory and varied in severity. Human-to-human transmission among those with COVID-19 and those without in close proximity was the primary form of spread. Because most with COVID-19 experience mild symptoms, transmission can affect more people (Vetter et al., 2020). In response to the growing number of cases of the virus, public health measures (similar to those taken during the 1918–1920 Spanish influenza; Bootsma & Ferguson, 2007) were implemented to slow the rate at which COVID-19 was spreading. Examples of these measures included washing hands frequently, avoiding touching the face and nose with one’s hands, and practicing social distancing (World Health Organization, 2020). Social distancing occurred in the community (e.g., cancellation and/or closure of schools, places of worship, non-essential businesses and events) and at the individual level (e.g., maintaining six feet of physical distance between people and limiting close contact with other people) as it is effective in mitigating pandemic influenza (Glass et al., 2006).
Prior to the COVID-19 outbreak, loneliness was a major public health concern among older Americans (Gerst-Emerson & Jaywardhana, 2015). The potential implications of the COVID-19-related stay-at-home order and measures of social distancing, particularly for the older adult population who were already vulnerable, must be given attention. Given the increased risk of adverse consequences associated with social isolation (Morrow-Howell et al., 2020), it is necessary to investigate the prevalence of and risk factors associated with feelings of social isolation and loneliness during the COVID-19 pandemic. However, given the unusual circumstances of the current study (namely, the stay-at-home order), no formal hypotheses were specified. Rather, the current study tested two research questions: First, to estimate the prevalence of feelings of loneliness and feelings of social isolation in a sample of older adults during the COVID-19 pandemic. Second, to investigate associations between objective measures of loneliness, isolation, and COVID-19-related anxiety during the stay-at-home order (i.e., a novel and unusual context). As past research has not occurred during mass quarantine/self-isolation, it is unknown whether these constructs are associated with each other.
Methods
The study was reviewed and approved by the Institutional Review Board at California State University, Sacramento (IRB-19-20-259).
Participants
Participants were 551 adults who were members of an organization for older adults interested in lifelong learning in Northern California. This organization, which is composed of more than 1900 members, provides over 200 seminars and forums on a range of topics on the California State University, Sacramento campus each year. Thirty-seven participants did not complete the survey and were excluded from all analyses. As such, the final sample size was 514.
Measures
Participants completed a total of 24 survey items regarding demographic information, loneliness, isolation, and COVID-19-related anxiety (see Supplemental Materials for complete list of items). To measure loneliness and COVID-19-related anxiety, the short-form UCLA Loneliness Scale (ULS-6; Neto, 2014) and the Coronavirus Anxiety Scale (CAS; Lee, 2020) were administered, respectively. These measures have been previously validated and achieved Cronbach’s α of 0.74 and 0.80 in the current study. Isolation was measured by asking participants about their marital status, household size, and whether they use social media. Additionally, participants were asked questions pertaining to feelings of isolation/loneliness, change in loneliness, frequency of in-person interactions, and change in that frequency were considered subjective measures. The survey was created in Qualtrics and took approximately five minutes to complete. The survey was sent to prospective participants via e-mail on May 28 2020, approximately two months after the March 19, 2020, stay-at-home order was enacted throughout California, and 74 days after recommendations for people aged 65 and older to stay at home. The organization of which the participants were members had also suspended its programs for 80 days.
Statistical analyses
All analyses were conducted using R 4.0.0 (R Core Team, 2020), and confirmatory factor analysis (CFA) was conducted and visualized using the lavaan 0.6.-6 (Rosseel, 2012) and semPlot 1.1.2 (Epskamp, 2015) packages. CFA was conducted to investigate associations between factors of loneliness (measured by the ULS-6), isolation (measured by marital status, household size, and social media use), and COVID-19-related anxiety (measured by the CAS). The ULS-6 and CAS were coded in accordance with the original papers describing the measures (Lee, 2020; Neto, 2014) such that a higher score indicated greater loneliness and greater COVID-19-related anxiety, respectively. For the CFA, marital status was recoded to give “married” a value of 0, and all other responses (“widowed”, “never married”, and “divorced or separated”) a value of 1. Social media use was coded so that a “yes” response, where participants indicate they do use social media, was assigned a value of 0, and a “no” response was assigned a value of 1. Household size was multiplied by −1. In short, a higher score for each item entered into the CFA was associated with worse symptoms.
The inter-factor correlations were of key interest. In order to evaluate the associations between the loneliness, isolation, and COVID-19-related anxiety factors for dimensional distinctness. The full three-factor model, where the loneliness, isolation, and COVID-19-related anxiety factors were free to correlate, was compared to an independent three-factor model (where the inter-factor correlations were constrained to 0), a one-factor model (where all survey items loaded onto a single factor), and a series of two-factor models where two factors were merged into one and then correlated with the third (e.g., a loneliness and isolation factor is correlated with the COVID-19-related anxiety factor). Additionally, the factors were considered to be distinct if the 95% upper-bound correlation confidence intervals (CIs) did not intersect with 1.0. Following the guidelines summarized by Scheizer (2010), the CFA model was considered to have acceptable fit based on the following criteria: Comparative Fit Index (CFI) ≥ 0.90; Tucker-Lewis Index (TLI) ≥ 0.90; standardized root mean residual (SRMR) ≤ 0.10; and the root mean square error of approximation (RMSEA) ≤ 0.08. The 90% CIs of the RMSEA were also reported. The Bayesian Information Criterion (BIC) was also used for model comparison. A smaller BIC value indicates a better fitting model. For thoroughness, the model chi-square statistics were also included, although this statistic is considered to be overly sensitive with large sample sizes (Scheizer, 2010). All models were tested using maximum likelihood estimation, although standard errors and confidence intervals were estimated via bias-corrected bootstrapping with 2000 replications to help ensure robustness to any deviations from normality.
Results
Prevalence of isolation and loneliness
Descriptive statistics of all items are presented in Table 1. Over half (56.4%) of participants reported feeling isolated as a result of the COVID-19 outbreak, and more than one third (36.0%) reported feeling lonely. Additionally, just over half of participants (50.1%) reported an increase in loneliness since the COVID-19 outbreak. Over half (54.5%) of participants reported having daily in-person interactions before the stay-at-home order, and 88.1% reported a decrease in their in-person interactions. Nearly half (40.9%) of participants reported living alone, and the majority reported using some form of social media.
Table 1. Descriptive statistics of survey items (n = 514)
Associations between loneliness, isolation, and COVID-19-related anxiety
Correlations between items entered into the CFA are presented in Table 2. The full three-factor model provided acceptable or close to acceptable model fit (see Table 3 for model fit statistics), and provided a better fit of the data than the next best fitting model (three independent factors), Δχ2 (3) = 62.388, p < .001, ΔBIC = −43.662. Figure 1 presents the full three-factor CFA model.
Table 2. Correlations between measures of loneliness, isolation, and COVID-19-related Anxiety (n = 514)
Table 3. Fit indices for the full confirmatory factor analysis model and reduced models of loneliness, isolation, and COVID-19-related anxiety (n = 514)
Published online:
04 October 2020Figure 1. The estimated three-factor model. Single-headed arrows have standardised factor loadings next to them. Double-headed arrows have inter-factor correlations next to them. All factor loadings are significant to p < .05 except for the social media use loading
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/wasp20/2021/wasp20.v033.i04-05/08959420.2020.1824541/20211004/images/medium/wasp_a_1824541_f0001_b.gif"})
Figure 1. The estimated three-factor model. Single-headed arrows have standardised factor loadings next to them. Double-headed arrows have inter-factor correlations next to them. All factor loadings are significant to p < .05 except for the social media use loading
Upon examination of the inter-factor correlations, it was noted that the correlation between the isolation and anxiety factors was very low (r = .020, 95% CI = −.081 – .143). Removal of this correlation did not significantly worsen model fit in comparison to the full three-factor model, Δχ2 (1) = 0.126, p = .723, and the BIC also decreased, ΔBIC = −6.116. Therefore, it was implied that the loneliness factor was associated with isolation (r = .319, 95% CI = .191-.453) and with COVID-19-related anxiety (r = .303, 95% CI = .173-.433), but that the isolation and COVID-19-related anxiety factors were independent. The final model fit statistics were close to acceptable or acceptable, CFI = 0.890, TLI = 0.867, RMSEA = 0.071 (90% CI = 0.062– 0.080), SRMR =0.058, χ2 (75) = 267.795, BIC = 12,796.354. Lastly, it should be noted that the loading of social media use on to the isolation factor was non-significant. However, as removing it made no difference to the outcome of any analysis, it was kept in all models.
Discussion
The current study aimed to estimate the prevalence of feelings of loneliness and feelings of social isolation and to investigate associations between objective measures of loneliness, isolation, as well as COVID-19-related anxiety during the stay-at-home order (i.e., a novel and unusual context). The findings of this survey reveal that over half of the older adults in the study sample felt isolated. Given that estimates of isolation and loneliness in older adults commonly report prevalence rates of 25% or lower during daily life (Fokkema et al., 2012; Menec et al., 2019; Theeke, 2009; Victor & Yang, 2012), the value observed in the current study is markedly higher. This difference may be attributed to the higher engagement that the participants likely had as a result of belonging to a lifelong learning organization. Perhaps surprisingly, this value was not as high for feelings of loneliness; that is, more participants feel isolated than lonely during the COVID-19 outbreak (although loneliness prevalence in the current study was still higher than 25%).
Additionally, CFA models found that loneliness was significantly associated with both isolation and COVID-19-related anxiety. However, isolation and COVID-19-related anxiety were not associated. Specifically, the CFA models found that loneliness and isolation are related, yet distinct, constructs. Isolation, in this case, refers more to the physical aspect. Aspects of loneliness may include not seeing family and friends often, wishing for more friends, feeling lonely the majority of the time, and having no friends nearby; whereas social isolation may include living alone, never visiting anyone, having no contact with neighbors, being alone for nine or more hours a day, and never leaving the house (Wenger & Burholt, 2004). These manifestations are of particular relevance to the current study and situation, in which participants are likely falling under the description of social isolation. The stay-at-home order has affected social interactions by reducing the number of people who can gather in a group, maintaining about six feet between people, and limiting trips and travel to only essential places, such as to medical appointments or grocery stores. Given that loneliness commonly refers to a perceived state of isolation (Cacioppo et al., 2014), it appears that those who perceive themselves to be isolated also worry the most about COVID-19. Prior to the COVID-19 outbreak, the majority of participants reported daily in-person interactions. With the stay-at-home order, however, opportunities to have in-person interactions and meetings have decreased, causing a large proportion of participants to notice a marked change. This may be resulting in increased anxiety with regards to COVID-19 (although it should be noted that this study is purely correlational and that the direction of associations between constructs cannot be determined).
With the aforementioned risks and effects of social isolation in mind, healthcare professionals and social service providers need to be aware of this finding. Older adults may already be at risk for loneliness, but the stay-at-home orders associated with COVID-19 have increased their vulnerability to isolation (see Office for National Statistics, 2018, however, for data showing that self-reported loneliness is more prevalent in young than older adults), and worry about the COVID-19 pandemic. To address this potential issue, it is important to maintain or increase contact (i.e., not in-person interactions) with older adults’ social networks through social media, phone calls, and video-based calls (Ouslander, 2020). Additionally, providers should encourage older adults to quickly reestablish in-person interactions and social activities once the COVID-19 pandemic has ended; doing so could potentially alleviate any lingering feelings of isolation and/or loneliness from the stay-at-home order. Interventions that are focused on social connectedness can be effective for alleviating feelings of loneliness (Masi et al., 2011). To address isolation, interventions that provide opportunities for social support and activity, particularly in group formats, may be beneficial (Dickens et al., 2011). However, these approaches are highly dependent on the characteristics and needs of individuals (Fakoya et al., 2020). Recommendations specific to feelings resulting from the COVID-19 pandemic would likely require additional research and consideration of related health risks for how to best tailor the interventions for older adults, keeping in mind that feelings of loneliness vary between individuals.
Limitations and future research
Although this study provides estimates of prevalence of feelings of loneliness and isolation in a sample of older adults, as well as investigating associations between loneliness, isolation, and COVID-19-related anxiety, there are some limitations that should be noted. First, this sample was not nationally representative, as the sample was based in Northern California. As well as potential variation in demographic factors, stay-at-home orders varied by county and state, so experiences of those in other regions or states are likely different. Given the racial/ethnic and socio-economic disparities in COVID-19-related outcomes (Hooper et al., 2020; Van Lancker & Parolin, 2020), it could be the case that COVID-19-related anxiety may also vary substantially between demographic groups. Second, the participants were members of a lifelong learning program, so they were likely social and active participants who may have been more affected than other community-dwelling older adults because the program’s classes were suspended during the pandemic. Older adults who are not as active, not community-dwelling, or who have smaller social networks may experience greater loneliness and/or isolation as a result of the stay-at-home order. On the other hand, those who are not as socially active as the participants in this study may feel no change in their perceived isolation. Third, the study design was cross-sectional, so it is possible that these feelings and the other variables of interest may change during the stay-at-home order. This survey captured the experiences of the participants about two months after the stay-at-home orders were enacted. In addition to differences in location, respondents may have different answers if this survey had been administered at other time points, such as after one month or three months. Future research could potentially follow up with older adults repeatedly during this period to see if feelings of loneliness and isolation change, and could also investigate the moderating effects of personality, as it could be a factor for determining one’s risk for isolation.
Conclusion
In conclusion, this study demonstrates that feelings of isolation and loneliness appear to be higher than usual as a result of the COVID-19 pandemic and the resulting stay-at-home orders, and that loneliness (but not isolation) is associated with anxiety about COVID-19. Although the research is relatively exploratory, it suggests that older adults who report feeling lonely appear to be most at risk for being anxious about the COVID-19 pandemic. Reducing isolation – both objective and perceived – is important within the context of this pandemic. Families and friends should place greater emphasis on connecting with older adults in their social networks, and providers should follow up with their patients to ensure that they are participating in activities that address their social needs.
| Survey Item | Responses |
|---|---|
| M = 73.25 years, SD = 6.56, range = 57-93 |
| 387 females; 127 males |
| 471 White; 14 Asian; 11 Mixed; 5 Black or African American; 2 Native Hawaiian or Pacific Islander; 1 American Indian or Alaska Native; 0 Hispanic; 10 Other |
| 275 married; 129 divorced or separated; 83 widowed; 27 never married |
| 290 yes; 224 no |
| 185 yes; 329 no |
| 108 never; 168 rarely; 181 sometimes; 57 often |
| 9 never; 42 rarely; 188 sometimes; 275 often |
| 80 never; 261 rarely; 148 sometimes; 25 often |
| 80 never; 156 rarely; 219 sometimes; 59 often |
| 147 never; 184 rarely; 153 sometimes; 30 often |
| 110 never; 191 rarely; 154 sometimes; 59 often |
| 75 a lot more; 183 slightly more; 210 unchanged; 34 slightly less; 12 a lot less |
| 210 one person; 262 two people; 31 three people; 4 four people; 7 five or more people |
| 399 yes; 115 no |
| 280 daily; 136 4-6 times a week; 79 2-3 times a week; 19 once a week or less |
| 280 daily; 126 4-6 times a week; 73 2-3 times a week; 19 once a week; 16 less than once a week |
| 291 much lower; 113 moderately lower; 49 slightly lower; 40 about the same; 6 slightly higher; 9 moderately higher; 6 much higher |
| 474 not at all; 27 rare, less than a day or two; 9 several days; 2 more than seven days; 2 nearly every day over the last two weeks |
| 345 not at all; 119 rare, less than a day or two; 35 several days; 9 more than seven days; 6 nearly every day over the last two weeks |
| 425 not at all; 67 rare, less than a day or two; 18 several days; 1 more than seven days; 3 nearly every day over the last two weeks |
| 471 not at all; 33 rare, less than a day or two; 9 several days; 0 more than seven days; 1 nearly every day over the last two weeks |
| 455 not at all; 42 rare, less than a day or two; 14 several days; 1 more than seven days; 2 nearly every day over the last two weeks |
| Open-ended responses entered in a text box. |
| Measure | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1. ULS 1 | - | ||||||||||||
| 2. ULS 2 | .25*** | - | |||||||||||
| 3. ULS 3 | .36*** | .24*** | - | ||||||||||
| 4. ULS 4 | .46*** | .25*** | .43*** | - | |||||||||
| 5. ULS 5 | .36*** | .21*** | .42*** | .52*** | - | ||||||||
| 6. ULS 6 | .31*** | .16*** | .32*** | .32*** | .27*** | - | |||||||
| 7. Marital status | .49*** | .03 | .07 | .09* | .10* | .07 | - | ||||||
| 8. Household size | .36*** | .00 | .10* | .10* | .10* | .08 | .58*** | - | |||||
| 9. Social media use | .08 | .16*** | .08 | .05 | .02 | -.06 | .08 | .05 | - | ||||
| 10. CAS 1 | .08 | .04 | .10* | .13** | .12** | .11* | -.02 | -.01 | .03 | - | |||
| 11. CAS 2 | .13** | .09* | .14** | .25*** | .25*** | .12** | .05 | .05 | .03 | .28*** | - | ||
| 12. CAS 3 | .10* | .03 | .10* | .18*** | .20*** | .10* | -.03 | -.03 | -.02 | .50*** | .35*** | - | |
| 13. CAS 4 | .13** | .07 | .05 | .15*** | .18*** | .08 | .04 | .04 | .07 | .52*** | .48*** | .44*** | - |
| 14. CAS 5 | .09* | .03 | .12** | .16*** | .16*** | .14** | -.01 | -.01 | -.02 | .53*** | .44*** | .48*** | .57*** |
* p <.05. ** p <.01. *** p<.001.
| Model | CFI | TLI | RMSEA (90% CI) | SRMR | χ2 | df | BIC |
|---|---|---|---|---|---|---|---|
| Full three-factor | 0.890 | 0.865 | 0.071 (0.062-0.081) | 0.058 | 267.669 | 74 | 12802.470 |
| Independent three-factor | 0.856 | 0.830 | 0.080 (0.071-0.089) | 0.101 | 330.057 | 77 | 12846.132 |
| One-factor | 0.483 | 0.389 | 0.151 (0.143-0.160) | 0.135 | 985.380 | 77 | 13501.455 |
| Two-factor (Loneliness = Isolation) | 0.788 | 0.743 | 0.098 (0.089-0.106) | 0.077 | 448.325 | 76 | 12970.642 |
| Two-factor (Loneliness = Anxiety) | 0.600 | 0.521 | 0.134 (0.126-0.143) | 0.124 | 779.345 | 76 | 13301.662 |
| Two-factor (Isolation = Anxiety) | 0.755 | 0.707 | 0.105 (0.096-0.114) | 0.093 | 506.456 | 76 | 13028.774 |
Disclosure statement
Laura Gaeta is a seminar co-leader in the Renaissance Society (volunteer position).
Key points:
Older adults are thought to be at increased risk of loneliness and social isolation.
A survey about loneliness during the coronavirus pandemic was sent to older adults.
The majority of respondents felt lonelier during the stay-at-home order.
Loneliness, but not isolation, is related to COVID-19-related anxiety.
Safely increasing social interactions may alleviate feelings of loneliness.