Depression is a common and debilitating mental health issue that can significantly affect individual well-being, such as through changes in mood, physical symptoms (e.g., fatigue, headaches, digestive problems, and chronic pain), suicidal behavior, and cognitive function [1]. According to the World Health Organization [2], approximately 4.4% of individuals experience depression worldwide, with the prevalence of depression being notably higher among women, as 5.1% of women experience the disorder compared to 3.6% of men [2]. Depression is also more prevalent among older adults, affecting approximately 7.5% of women aged 55-74 and 5.5% of men in the same age group [2]. In South Korea, which is one of the fastest aging countries in the world, a significant proportion of adults aged 60 and older have reported depressive symptoms (11%-13%) [3], experiencing loneliness (19.6%-23.4%) [4], and a lack of social support (28.4%-31.1%) [4]. The increased societal consequences resulting from depression, such as suicidal behavior, has already come to the forefront of public health concerns, thus underscoring the need to understand factors that may protect against the development or continuation of depressive symptoms.

Social isolation has been identified as a crucial risk factor for depressive symptoms [5-9]. Research has consistently documented the effect of social isolation on depressive symptoms, and studies have proposed behavioral (e.g., reduced physical activity), psychosocial (e.g., lacking a feeling of belonginess), and physiological (e.g., inflammation) mechanisms to explain these associations [7,10]. Prior research has also distinguished between two components of social isolation, namely, social disconnectedness and perceived isolation (i.e., objective isolation and subjective isolation), which have separate and distinct associations with mental health [5,11]. Whereas social disconnectedness refers to the degree of physical separation from other people and society as measured by social network size, frequency of interactions, and social participation, perceived social isolation refers to an individual’s subjective experience with and feelings in their relationships with others, such as loneliness or a lack of perceived social support, regardless of physical separation [5,11-13]. These two components of social isolation may operate together in their association with mental health outcome, or they may stand alone and have fully separate patterns of association. Individuals that are physically connected through socializing with others and participating in social organizations may still feel lonely and a lack of social support, which can then have a deleterious effect on their mental health. By contrast, other individuals who are socially disconnected but not lonely may still maintain good mental health. Indeed, some previous studies have found that socially disconnected older adults have worse mental health only to the extent that they perceive themselves to be isolated [5,11]. However, less is known about the extent to which social disconnectedness and loneliness change differently and consequently have separate patterns of change in their association with depressive symptoms.

Although previous studies distinguished between the two components of social isolation and also shown that they have distinct associations with mental health, the majority of research in this area focused predominantly on older adults [5,11,14], thus leaving a significant gap in the empirical literature regarding the varying associations across different age groups. This gap is particularly concerning considering the differential impact of the coronavirus disease-2019 (COVID-19) pandemic on social connectedness and loneliness across age groups [4,15]. Examining the pre-pandemic levels of social connectedness and loneliness is expected to provide valuable insights into the varying degrees of social isolation experienced across different age groups during the pandemic, and the subsequent impact of this isolation on mental health outcomes. Given the mandatory stay-at-home measures and restrictions on face-to-face interactions, understanding these relative differences is crucial for addressing social isolation and its associated mental health consequences.

Social connectedness among Korean middle to older adults exhibits distinct features that are seldom observed in other countries which should be noted. For instance, while older adults aged 75 and above in the United States mainly socialize with relatives, neighbors, old friends, and members from social organizations [16,17], their Korean counterparts predominantly socialize with members from senior community centers or social welfare centers for seniors in addition to their relatives, neighbors, and friends. 18 Senior community centers represent a unique aspect of old age in Korea, as approximately 30 to 40 percent of older adults utilize these facilities. These centers serve as multifunctional spaces that provide various amenities for accommodation, meals, and leisure activities [19]. To capture the unique context of Korea, specific response categories such as “no-in-jung” or “kyung-ro-dang” in Korean for senior centers and “no-in-bokji-kwan” for social welfare centers are incorporated into representative panel surveys involving older adults, such as the Korean Longitudinal Study on Ageing (KLoSA) [20]. Notably, these senior centers and social service centers were freqeuntly selected by older adults responding to the KLoSA survey [18]. In the context of the COVID-19 pandemic, where minimal activities and interactions were allowed in places like workplaces, it is expected that older adults whose primary interactions revolved around locations that were forced to close (e.g., senior centers, daycare centers, leisure clubs, and religious organizations) experienced an inevitable increase in social disconnectedness. Therefore, the impact of social distancing measures after the outbreak of the COVID-19 pandemic may be more severe among older adults compared to middle-aged adults.

The purpose of this study is to examine the impact of COVID-19 on the associations between social isolation and depressive symptoms of adults aged 45 or older using four waves of data from the KLoSA (2014-2020). The primary focus of our study is to investigate potential variations in the relationships between the two dimensions of social isolation and depressive symptoms across three distinct age groups(specially, 45-64, 65-74, and 75 or over) both before and during the COVID-19 pandemic. We will use longitudinal panel data to apply fixed-effect (FE) method to control for time-invariant unobserved characteristics that could bias the estimation in ordinary least square regression. To our knowledge, this study is the first to estimate the longitudinal effects of social connectedness and perceived isolation on depressive symptoms among middle-aged to older adults in Korea, using data collected both before and during the COVID-19 period.

Table 1 lists the summary statistics for the sample of individuals aged 45 to 64, 65 to 74, and 75 or above. The data shows that, compared to younger individuals in the sample, older adults are more likely to be female and less likely to be married due to divorce and bereavement. Individuals aged 75 or above are also more likely to have lower educational attainment and less likely to be actively working, thus resulting in a lower average income compared to younger individuals. Older adults aged 75 or above with lower socioeconomic status also exhibit lower rates of smoking and alcohol consumption. However, due to their higher age, they are still more vulnerable to health shocks. Therefore, they face a higher prevalence of chronic diseases, and their self-rated health is poorer compared to that of any of the other older age groups. In terms of the outcome variables, the level of loneliness is found to increase with age and is highest among individuals aged 75 or above. However, older adults aged 75 or above, who benefit from the unique community center (No-in-jung) system for older adults in Korea, have a higher frequency of interactions. Despite having higher levels of objective connectedness, the proportion of individuals experiencing loneliness and the degree of associated depression symptoms are both significantly higher in the 75 and above age group compared to any other of the older age groups.

Difference p-value in Table 1 displays the F-statistics and related p-values used to test the equality of each variable between age groups. Different age groups exhibit distinct characteristics across all variables, including control and outcome variables. The results of a test of joint significance across all variables are shown to reject the null hypothesis of equality between age groups. Considering that all age groups have different characteristics across all observed variables, they might also possess distinct unobserved characteristics that can influence outcome variables. Therefore, we use the panel fixed model to control for unobserved time-invariant characteristics.

Table 2 presents the results of estimating equation (1), which focuses on the interaction term between aged groups and years. We demonstrate how each age group responds differently to the COVID-19 pandemic. Table 2 presents RE and FE estimates of the effect of sample period and age on depressive symptom (CES-D). The results of the FE model (RE model) are as follows: first, throughout the sample period, individuals aged 75 or above show higher depressive symptoms scores compared to those in the 45 to 64 age group by 0.36 (0.28). Second, when examining the impact of time, it becomes evident that the year 2020 showed a significant 0.48 (0.49) increase in depression symptoms (CES-D) compared to the data from 2014, thus indicating a substantial impact of the pandemic across all age group. While some fluctuations were observed between 2014 and 2020, it is noteworthy that the substantial surge in scores observed specifically in 2020 surpasses the increases noted in other years. However, older adults aged 75 or above are less likely to be negatively influenced by the COVID-19 pandemic in terms of depressive symptoms. Individuals aged 75 or above had lower depressive symptom scores compared to other age groups. Specifically, their CES-D score is lower than the corresponding score for those aged 45 to 64 in 2014 by 0.4440 (0.4141) points.

Figure 1 depicts the trends over time and by age groups for depressive symptoms (CES-D), which aligns with the FE results presented in Table 2. Individuals aged 45 to 64 and those aged 65 to 74 experienced a worsening of depressive symptoms in 2020. However, the group aged 75 and older did not experience higher depressive symptoms during the COVID-19 period compared to other periods.

To understand what dimension of social isolation affects depressive symptoms more before and during the COVID-19 pandemic, Table 3 presents the effects of social connectedness and perceived isolation on depressive symptoms. The frequencies of socializing and loneliness are the main explanatory variables in model (1) and model (2), respectively. Perceived isolation (loneliness) is associated with higher depressive symptom (0.6613, p<0.001), while social connectedness (frequency of social interactions) does not have a statistically significant relationship with depressive symptoms due to its small magnitude. Model (3) examines the effects of both frequency of social interaction and loneliness on CES-D. The relationship between frequency of social interaction and depressive symptom remains weak (0.0051, p<0.05), but the relationship between perceived isolation (loneliness) and depressive symptoms is strongly negative (0.6630, p<0.001).

Each age group has data from two sample periods: before and during the COVID-19 pandemic. Table 4 presents separate results for each age group and sample period regarding the effect of social connectedness and loneliness on depressive symptom. To understand how different age groups differentially respond to the COVID-19 pandemic, the results indicate the effect of frequency of socializing and loneliness on depressive symptoms for each age group. There are two details related to the results that are worth mentioning. First, perceived isolation shows a stronger link with depressive symptoms than it does with social connectedness for all age groups. Second, the strength of that link has changed before and during the COVID-19 pandemic for each age group. While the effect of loneliness became larger during the pandemic for individuals aged 45 to 64, perceived isolation (loneliness) in older age groups had a similar or even reduced impact on depressive symptoms. The coefficients of Loneliness decreased from 0.8651 to 0.7774 for older adults aged 75 or above. Coefficients for older adults aged 65 to 74 have slightly decreased from 0.6332 to 0.6074 between the periods before and during COVID-19.

We expected that the COVID-19 pandemic would have the most significant impact on the social disconnectedness of individuals aged 75 or above, thus leading to an increase in loneliness. We therefore expected that this heightened loneliness would intensify the association between loneliness and depressive symptoms. To explore these unexpected findings and how COVID-19 affects social disconnectedness and loneliness across age groups, we also conducted a supplementary analysis of the effects of COVID-19 on social connectedness and loneliness across the time. Supplementary Table 1 (in the online-only Data Supplement) illustrates how the frequency of social interaction and loneliness changed in response to the COVID-19 pandemic among different age groups. Note that individuals aged 75 or above exhibit a higher frequency of social interactions than individuals aged 45 to 64 by 1.32. However, individuals aged 75 or above experienced the most significant reduction in the number of social interactions in the year 2020 (-1.70, p<0.001). Surprisingly, despite this decrease, they also experienced the largest reduction in the level of loneliness (-0.1381, p<0.001). Although the COVID-19 pandemic affected individuals aged 75 or above by decreasing their frequency of social interaction, this did not result in an increase in their experience of loneliness.

Consistent with Supplementary Table 1 and Supplementary Figure 1 (in the online-only Data Supplement) presents the predicted value of frequency of social interaction by age group. The trend from 2016 to 2020 is characterized by a sharp decrease in the frequency of social interaction for those aged 65 to 74 as well as those aged 75 or above during COVID-19. On the other hand, individuals aged 45 to 65 maintained their level of social interaction compared to 2018. Therefore, the group that was most affected by social distancing measures was older adults aged 75 or over. On the other hand, Supplementary Figure 2 (in the online-only Data Supplement) shows the level of loneliness over time and by age group. Notably, although individuals aged 45 to 64 maintained their social interactions during 2020, the extent to which they felt lonely increased in this age group. However, individuals aged 65 and older (aged 65 to 74 and aged 75 or older) experienced a decrease in the level of loneliness during the COVID-19 period.

Social isolation and depressive symptom are both growing concerns in many countries. While previous studies have found that the two distinct forms of social isolation—i.e., social disconnectedness and loneliness—have profound effects on depressive symptoms [11,14], less is known the extent to which these associations change differently across age groups. Moreover, in a context where individuals are compelled to adhere to shelter-in-place orders and minimize in-person interactions during the pandemic, namely, when everyone was forced to undergo enforced physical separation from others, the preexisting variations in social connectedness and loneliness among different age groups can significantly contribute to the emergence of social isolation during the pandemic and therefore impact mental health outcomes. To investigate this issue, we employed the FE models using four waves (2014-2020) of the KLoSA, a nationally representative sample of individuals age 45 and over collected before and during the pandemic. The FE estimates reveals that loneliness, and not social disconnectedness, is associated with depressive symptoms. This finding is generally consistent with recent research using the longitudinal representative data of older adults in the United States that disentangle the impact of perceived isolation and social connectedness on mental health [11,14]. However, our research findings uncover distinct age-specific patterns regarding the impact of COVID-19 on depressive symptoms, and they also help elucidate the relationship between social isolation and depressive symptoms. These findings sharply contrast with recent findings from other countries using similarly designed longitudinal data (e.g., HRS; Canadian Longitudinal Study; and the SHARE) and therefore represent a significant contribution to the existing body of knowledge and offer novel insights into the subject matter.

First, our findings provide strong evidence that depressive symptoms increased during the pandemic, but only for middle-aged (age 45-64) and young older adults (age 65-74). In fact, the level of depressive symptoms of the oldest group (age 75 or over) remained at a relatively stable level before and during the pandemic. These findings are consistent with the results of a number of studies that have reported slightly lower levels of depressive symptoms among older adults after the COVID-19 outbreak. However, it should be noted that these studies often had limitations such as small sample sizes [24], cross-sectional designs [25], or a lack of differentiation between distinct age groups [26,27]. Consequently, these studies were unable to thoroughly investigate age-specific changes in psychological responses—such as loneliness—to the pandemic and their subsequent impact on mental health.

Our findings also indicate that the impact of loneliness on depressive symptoms outweighs that of social disconnectedness, with particular emphasis on the age-specific variations in the changes observed in the association between loneliness and depressive symptoms both before and during the pandemic. Specifically, we observed an intensified impact of loneliness on depressive symptom among individuals aged 45 to 64 during the pandemic. By contrast, the influence of loneliness on depressive symptoms among older age groups remained either consistent (age 65-74) or even reduced (age 75 or above). Two other studies using longitudinal representative data collected before and during the pandemic reveal inconclusive findings: whereas Canadian older adults aged 65 or older exhibit an attenuated association between loneliness and depressive symptoms [27], Irish older adults aged 54 or older demonstrate amplified associations [28]. This inconclusive finding could be partially attributed to unspecified age groups. In other words, contrary to the concern raised by public health experts and other relevant researchers regarding the potential increase in loneliness and subsequent impact on mental health, particularly among older adults as a result of mandated physical separation, our findings reveal that it is in fact middle-aged adults who are most likely to feel lonely and depressed as well as experience the greatest impact of loneliness on depressive symptoms.

These findings may be attributable to age-specific challenges within the context of aging in Korea and hold significant policy implications. Notably, a substantial portion of social connections among older adults is dependent on interactions within senior centers [19]. The significant decline in socializing among older adults can be attributed to the mandatory closure of these centers along with the suspension of related activities in response to social distancing policies. Further, the observed decrease in loneliness and its impact on depressive symptoms among individuals aged 75 or above, despite their reduced socializing following the onset of COVID-19, raises concerns about the adequacy of social relationships within these settings in terms of the social needs of bonding and attachment. It is plausible that older adults aged 75 or above may visit senior centers primarily for purposes of casual socialization, engaging in informal conversations, dining, and passing time. However, the quality of these relationships warrants further examination. Recent findings from studies involving Dutch older adults suggest that interventions that effectively combine physical distancing and well-being are closely tied to the quality of social contacts, including interactions with societal institutions [26]. Therefore, there is a need for additional research to thoroughly explore the quality of social relationships among Korean older adults whose primary social interactions revolve around senior centers.

The finding that young older adults (age 65-74) exhibited higher levels of depressive symptoms during the pandemic, despite a decline in loneliness, along with the sustained strength of the association between loneliness and depressive symptoms, suggests that senior centers or senior welfare centers may hold significance beyond mere socialization for this particular group. This observation may be linked to their relatively higher labor force participation, as young older adults often engage in employment to supplement their post-retirement living expenses [3]. Notably, the closure of social welfare centers, which are frequented by young older adults and offer elderly employment programs, has been found to heighten concerns regarding income and employment security, thus reflecting the need for economic support in this age group [29].

On the other hand, in the case of middle-aged adults (age 45-64), our findings indicate an increase in both depressive symptoms and the intensity of the association between loneliness and depressive symptoms, despite the observed consistency in their level of social interactions prior to and during the pandemic. For individuals in this particular group, they may maintain minimum level of social interactions based on their worksite (as an example), but the abrupt restriction in personal social interactions may amplify feelings of loneliness and its correlation with depressive symptoms. Further investigations are warranted to determine whether the heightened levels of loneliness and depressive symptoms that are currently observed are of a transient, acute nature or if they have evolved into chronic conditions.

This study has some limitations. First, the measure of loneliness was based on a single item. Although a single-item measure of loneliness is widely used in the literature, the multiple items can capture different dimensions of loneliness (e.g., intimate, relational, and collective) 6 that match different dimensions of social connectedness and thus have different association with health outcomes. Second, the data is based on community-dwelling adults. Residents in nursing homes were completely cut off from visitors due to pandemic-related restrictions, and they may have witnessed fellow residents dying more frequently [30]. Therefore, feelings of loneliness and depression can be more severe among that population compared to the population in this study.

Despite these limitations, our study is one of the first studies to examine changes in the relationship between social isolation and depressive symptoms before and during the COVID-19 pandemic, by using a nationally representative sample of older adults in Korea (KLoSA). This study also offers valuable insights into whether different age groups, including middle-aged individuals (45-64), young older adults (65-74), and the oldest adults (75 or more), exhibited distinct responses to the COVID-19 pandemic, thus resulting in varying associations with mental health outcomes. Moreover, the data utilized in this study are structured in a manner similar to those used in well-established studies such as HRS, ELSA, and SHARE. This design choice allows for meaningful comparisons of COVID-19 responses among older adults across various countries. Consequently, this study not only offers valuable policy implications during the global pandemic, but it also contributes to a comprehensive understanding of the unprecedented aging challenges faced by Korea within a global context.