Social isolation is the objective separation from others, and loneliness is the subjective feeling of being alone or separated. Associations between social isolation, loneliness, and various adverse health outcomes have been reported (1–5). Moreover, scholars have argued that the health risks associated with isolation and loneliness are equivalent to the well-established detrimental effects of smoking and obesity (6).

The recent outbreak of the coronavirus disease 2019 (COVID-19) forced people to socially isolate themselves (7) and to feel lonely (8, 9) as a result of strategies enacted in efforts to reduce the risk of infection. This was because people were unable to engage in casual interactions under social distancing and stay-at-home orders (10, 11). One pandemic-related review confirmed that visiting restrictions during the pandemic led to social isolation, loneliness, and negative consequences such as reduced nutritional intake and physical inactivity (12). Several studies have also reported both social isolation (13) and loneliness (14–16) during the pandemic as potential risk factors for poor mental health. Additionally, Murayama et al. reported that those who became socially isolated during the pandemic were likely to experience greater loneliness (7).

Because both social isolation and loneliness had deleterious impacts on people's health conditions not only before but also during the pandemic, the trends in social isolation and loneliness due to the unique context of the COVID-19 pandemic should be carefully monitored. Pedersen et al. investigated time trends in social isolation and loneliness over the first 16 months of the COVID-19 pandemic (between March 2020 and July 2021) in Denmark using nationally representative, repeated cross-sectional data (17). Overall, poorer levels of social isolation and loneliness were observed during the strict lockdown periods, whereas better levels were observed during the reopening phases. Sugaya et al. reported no improvement in social isolation and loneliness between May 2020 and February 2021 among Japanese people, whereas psychological distress and depression significantly decreased (18).

However, this area of research can be improved in at least two ways. First, trends of social isolation and loneliness during the pandemic have not been well-documented because much of the research related to these conditions and experiences was derived from a cross-sectional design. Second, detailed information, such as the changes in people's characteristics and residential environments, is not available, warranting further analysis. This information could contribute to developing effective strategies not only for the current COVID-19 pandemic but also for future pandemics.

Did the COVID-19 pandemic have a negative impact on people's social connections with others and society? To address the gaps mentioned above, this study aimed to describe changes in the prevalence of social isolation and loneliness between the first year (2020) and the second year (2021) of the COVID-19 pandemic among Japanese people using data from large-scale, web-based nationwide surveys. In particular, we compared the prevalence of social isolation and loneliness by demographic characteristics, socioeconomic status, health conditions, and outbreak situations in residential prefectures of Japan.

Table 1 presents the weighted prevalence of social isolation and loneliness among the participants. The proportion (95% CI) of individuals with social isolation in the total sample was 27.4% (25.9, 28.9) in 2020 and 22.7% (21.9, 23.5) in 2021. The percentage difference from 2020 to 2021 was −4.7 percentage points (−6.3, −3.1), indicating that social isolation decreased during the interval. Concerning loneliness, the weighted mean score of the UCLA Loneliness Scale was 5.03 (4.86, 5.20) in 2020 and 5.86 (5.81, 5.91) in 2021, and the score difference was 0.83 points (0.66, 1.00). This suggests that, contrary to social isolation, loneliness increased from 2020 to 2021.

The weighted prevalence of social isolation and loneliness, according to sex and age, is shown in Table 1. Social isolation was significantly greater in men than in women both in 2020 and 2021, but the decrease was similar for men and women [men: −4.4 percentage points (−7.0, −1.8); women: −4.9 percentage points (−7.0, −2.8)]. The level of loneliness was not different by sex in 2020, but in 2021, women had greater loneliness than men: 5.01 (4.71, 5.30) in men and 5.05 (4.90, 5.21) in women in 2020, and 5.62 (5.55, 5.68) in men and 6.09 (6.02, 6.16) in women in 2021. The level of increase in loneliness tended to be greater in women than men [men: 0.61 points (0.31, 0.91); women: 1.04 points (0.88, 1.20)].

The proportion of social isolation in younger ages, particularly 20–29 years, was lower than that in other age subgroups in both 2020 and 2021. The decrease was greatest in the 15–19 years age group [−15.7 percentage points (−22.8, −8.5)]. The UCLA Loneliness Scale score was the highest in the subgroup aged 20–29 years in 2020 [7.01 (6.18, 7.83)], but the score decreased to 6.13 (5.98, 6.28) in 2021. In the other age subgroups, the score increased from 2020 to 2021, particularly among the middle-aged and older populations.

Figures 1, 2 display the observed distribution of social isolation and loneliness across age groups (solid lines). Regarding social isolation, the age distribution in 2020 was characterized by a dip around age 20 in early life; however, in the period of middle and old ages, the distribution was almost stable. In 2021, although the proportion of social isolation decreased among those under the age of 20 compared to 2020, the overall trend after the age of 20 was similar to the 2020 distribution. In contrast, the age distribution of loneliness in 2020 was characterized by a sharp peak around the age of 20 and a downward trend in middle and late adulthood. However, as indicated by the broad confidence band, there was large inter-individual variability in loneliness in this age group. After 1 year, the level of loneliness flattened across young and middle adulthood and declined continuously into old age.

Table 2 represents the weighted prevalence of social isolation and loneliness by demographics and socioeconomic status. Those who were married had a greater proportion of social isolation in 2020, but it significantly decreased in 2021 compared to unmarried people. Loneliness was consistently greater in unmarried individuals than in married individuals in both years. In 2020, people cohabiting had a higher proportion of social isolation than those living alone; however, this disparity was attenuated by 2021. Loneliness was greater among those living alone than among those living with others in both 2020 and 2021.

People with lower education were more likely to be socially isolated but have lower loneliness in 2020, whereas people with lower income tended to be socially isolated and have higher loneliness. These trends continued in 2021. Concerning employment conditions, those with permanent jobs and those who were unemployed (including homemakers) had a higher proportion of social isolation by 2020. These decreased in 2021, but the proportions remained higher than those in the other categories. Social isolation among students remarkably improved in 2021 compared to 2020. The disparity in loneliness by employment condition observed in 2020 was suppressed by 2021. Social isolation among restaurant workers was high in 2020 but decreased in 2021.

Finally, Table 3 illustrates the weighted prevalence of social isolation and loneliness by health conditions and outbreak situation in the residential prefecture. People with each chronic disease were less likely to be socially isolated, but they experienced higher loneliness in 2020. The changing trend of social isolation varied according to the type of chronic disease, and there was a general decrease in loneliness in 2021. Changes in the proportion of social isolation and loneliness scores were smaller in prefectures with greater numbers of COVID-19 infection cases.

This study investigated changes in the prevalence of social isolation and loneliness among Japanese people between the first and second years of the COVID-19 outbreak. The study also examined the differences among subgroups defined by demographic characteristics, socioeconomic status, health conditions, and outbreak situations in residential prefectures. Overall, we found that social isolation decreased from 2020 to 2021, whereas loneliness increased.

In the first year of the pandemic in Japan, most schools were closed, and students received classes online. The extracurricular activity was also assessed. Many companies have introduced teleworking, including, as of April through May 2020, 56.4% of all companies, 83.3% of medium-sized companies, and 50.5% of small-sized companies (30). Additionally, most community activities were forced to refrain. In contrast, in the second year of the pandemic, the situation regarding COVID-19 had changed. For example, COVID-19 vaccination has been widespread, and people have understood and adopted infection-prevention behaviors naturally. Schools and community activities were reopened. An increasing number of workers were going to the office; the proportion of teleworkers was 38.4% in March 2021 (30). Under these circumstances, the COVID-19 outbreak struck everyone's social connections with others and society; however, the influence was relieved in the second year. Therefore, social isolation decreased by September and October of 2021 compared to 2020.

However, there were variations in the degree of change according to participants' characteristics. The decrease was particularly large for those aged 15–19 years, which is likely due to the school's shift to face-to-face classes in 2021. The proportion of social isolation among the older population was higher than that among the younger population as of 2020, but the percentages decreased more sharply among those in the older groups than among those in the younger groups by September and October of 2021. As previous research indicated (7), older people in Japan were not familiar with online communication tools such as the Internet (e.g., Zoom and Skype) and social networking services (SNSs) (31). Another survey showed that the proportion of older people who reported an increase in the frequency of use of online connections was lower in Japan than in Western countries, including the United States, Germany, and Sweden (32). Thus, social isolation tended to be prevalent in the older population during the first year of the pandemic. However, because community activities, the major participants of which were older people (33), have been reopened, they could be less socially isolated in the second year.

People who were married and cohabiting were more likely to be socially isolated than their counterparts in the first year of the pandemic. A previous study conducted before the COVID-19 pandemic reported that older people living with others tended to become more socially isolated than those living alone (21). People who were married or living with someone tended to complete their social interactions within the household without having connections with those outside the household, particularly during the pandemic period. However, these gaps observed in 2020 were attenuated in 2021, which implies that people expanded their daily activities in the second year of the pandemic when restrictions on people's behaviors eased.

The proportion of social isolation has decreased significantly among restaurant workers. In the first year of the outbreak, most prefectural governments in Japan requested shorter hours or restaurant closures. Contrarily, in the second year, prefectures (except those with large numbers of COVID-19 cases) did not. Therefore, a remarkable decline in the proportion of restaurant workers is observed. In the prefectures with fewer COVID-19 infection cases, we observed a greater decrease in social isolation. In such prefectures, municipal governments tend to request fewer behavioral restrictions on people, especially in the second year. People in the prefectures were somewhat free of social activities; thus, they were unlikely to be socially isolated.

In contrast to social isolation, loneliness increased in almost all the strata. This implies that the prolonged pandemic may have caused people to feel overwhelmed and exhausted, which is sometimes called “COVID-19-related psychological fatigue” (34), resulting in an increase in loneliness from 2020 to 2021. By age group, loneliness was higher among younger generations in 2020, especially among those aged 15–29 years, although they were not likely to be socially isolated. This result is consistent with those of a public cross-sectional survey conducted in Japan (35). Social isolation and loneliness are similar concepts but not the same. While the younger generation is good at connecting through the Internet and SNSs, such online connections are less likely to generate support that can address function limitation concerns (36, 37). Therefore, they might feel lonely, even if they have more online connections. In addition, people who were unmarried and living alone had greater loneliness during the study period, although the proportions of social isolation among them were lower in 2020. Earlier research presented a similar trend (38, 39). These similar findings provide evidence to support that social isolation and loneliness are not necessarily the same concept.

As an exception, loneliness was higher among those with chronic illness than their counterparts in the first year, and it decreased in the second year. A study in Denmark reported a similar trend (17). Psychological burdens such as anxiety and worries about COVID-19 may have decreased with the gradual establishment of infection-prevention behaviors in the population and with the spread of COVID-19 vaccination (17). Those with chronic illness and at a higher risk of COVID-19 infection might have carried a greater psychological burden in 2020 due to alienation from others, but this condition may have eased by 2021. Loneliness was higher in prefectures with more infections in the first year. However, in the second year, it increased in all prefectures, particularly within those with fewer infection cases. As a result, the disparity among the prefectures was reduced, suggesting that loneliness was pervasive, regardless of infection status.

Some limitations must be considered. First, because we used data from the web-based survey, the sample did not necessarily reflect the demographic distribution of the general population. This potential selection bias was likely to result in more optimistic estimates than the true levels in the population. Fundamentally, average social isolation and loneliness levels are likely to be worse than those reported in this study. Considering this, a weighting technique using external nationally representative data was applied to reduce bias. However, a residual bias may still exist. For example, people who were less negatively affected by the pandemic tended to answer online surveys. Furthermore, older people with better cognitive function and technological competencies were more likely to participate in the survey than those with worse status and those lacking technological competencies. Second, questions regarding social isolation did not include work-related connections, although we measured the frequency of contact with family members or relatives who were living apart and friends/neighbors. Thus, the prevalence of social isolation, particularly in young and middle-aged individuals, may have been underestimated. Various forms of social connections should be assessed in future surveys to capture social isolation conditions correctly. Third, this study only investigated participants' experiences at two time points during/after the pandemic. The level of social isolation and loneliness could be influenced by social events of the moment (e.g., infection spread, lockdown, stay-at-home order) (17). A time trend study is necessary to capture the status of these indicators accurately. Finally, the results included some response categories with small proportions of participants (e.g., job types and health conditions). We deemed that the prevalence of social isolation and loneliness in these categories could be useful information to understand vulnerable populations that were strongly influenced by the COVID-19 pandemic. However, because a small sample size possibly leads to low statistical power, the results must be interpreted with caution.

Social isolation decreased from the first year (2020) to the second year (2021) of the COVID-19 pandemic, whereas loneliness increased among the Japanese population. However, the trends in these changes differed according to the characteristics of the participants. Furthermore, the assessment of the impact of the COVID-19 pandemic on social isolation and loneliness will lead to an understanding of who was vulnerable during the pandemic. Based on the findings, the government and public health sector can take measures to prioritize those most affected by social isolation and loneliness and establish efficient strategies to prevent mental disorders in future pandemics.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Research Ethics Committee of the Osaka International Cancer Institute. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

HM and TT designed the study and collected data. HM, TSu, IN, and TSh conducted data analyses. HM prepared the manuscript. TSu, IN, TSh, and TT provided critical feedback. All the authors have read and approved the final version of the manuscript.