https://orcid.org/0000-0002-9622-0117
Figures
Abstract
Background
Concerns about increased loneliness during the COVID-19 pandemic have been widely reported. However, long-term trajectories of loneliness remain insufficiently described.
Objective
To examine annual trends and five-year transition patterns of loneliness in Japan from 2020 to 2024 using data from a nationwide online panel.
Methods
We analyzed data from 1,673 adults who participated in all five annual survey waves from 2020 to 2024 and had complete information on the variables used in the analysis. Loneliness was measured using the three-item UCLA Loneliness Scale. Descriptive statistics were used to estimate annual prevalence and five-year transition patterns. Sensitivity analyses were conducted using a stricter loneliness threshold.
Results
The prevalence of loneliness peaked in 2021, declined in 2022, and remained relatively stable from 2022 to 2024. Most respondents demonstrated stable trajectories over five years, either consistently reporting loneliness or consistently not reporting loneliness. A smaller proportion exhibited transient loneliness during the pandemic period. Under the stricter loneliness threshold, the 2021 peak and the predominance of stable over transitional sequences were preserved, although the ranking of specific sequences changed.
Conclusions
Among respondents retained in this five-wave online panel, loneliness prevalence peaked in 2021, declined in 2022, and remained relatively stable from 2022 to 2024. Stable trajectories were more common than transitional trajectories. These findings provide descriptive evidence on long-term loneliness patterns but should not be interpreted as nationally representative prevalence estimates.
Citation: Amarsanaa J, Nguyen TXT, Khan MSR, Kadoya Y (2026) The dynamics of loneliness: A longitudinal exploration of transitions in Japan across the COVID-19 period, 2020–2024. PLoS One 21(9): e0358269. https://doi.org/10.1371/journal.pone.0358269
Editor: Bruna Moretti Luchesi, UFMS: Universidade Federal de Mato Grosso do Sul, BRAZIL
Received: February 26, 2026; Accepted: August 28, 2026; Published: September 11, 2026
Copyright: © 2026 Amarsanaa et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: This work was supported by JSPS KAKENHI through grant numbers JP23K25534 and JP24K21417, awarded to Yoshihiko Kadoya; JP25K16669, awarded to Trinh Xuan Thi Nguyen. The funder had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. There was no additional external funding received for this study.
Competing interests: The authors have declared that no competing interests exist.
1. Introduction
Loneliness has received increasing public health attention in recent years [1–4], particularly during the COVID-19 pandemic [5–9]. Social distancing measures, mobility restrictions, and changes in daily routines raised concerns about potential increases in social isolation and subjective loneliness across populations [1,3]. Numerous cross-sectional studies documented elevated levels of loneliness during the early stages of the pandemic [10–12]. However, less is known about how loneliness evolved over multiple years as societies adapted to changing conditions.
Most existing studies rely on cross-sectional data or short-term longitudinal follow-up periods [13–17]. While such studies provide valuable snapshots, they offer limited insight into longer-term stability and change. Understanding whether loneliness levels returned to pre-pandemic patterns, remained elevated, or exhibited heterogeneous trajectories across individuals requires longitudinal data spanning multiple years.
Japan provides a useful setting for examining long-term changes in loneliness because the same nationwide online survey was conducted annually over five consecutive years spanning February 2020 to March 2024. This repeated measurement makes it possible to examine not only annual prevalence but also individual-level stability and transition patterns across the different phases of the COVID-19 period.
The present study uses data from a nationwide online panel survey conducted annually from 2020 to 2024 to describe changes in loneliness over time. Rather than focusing on causal explanations, we document annual prevalence trends and summarize five-year response sequences to characterize common longitudinal patterns. By providing descriptive evidence on both stability and transition in loneliness, this study contributes to ongoing public health monitoring efforts and offers a longitudinal perspective on experiences among respondents retained in the longitudinal panel during and after the COVID-19 period.
2. Data and methodology
2.1. Data
This study used data from five consecutive waves of the Household Behavioral and Financial Survey conducted by Hiroshima University in 2020, 2021, 2022, 2023, and 2024. The surveys were administered online by Nikkei Research using its own and allied registered monitor panels. Participants were adults aged 20 years or older residing throughout Japan. The survey protocol was reviewed internally by Nikkei Research in accordance with its ethical and survey administration procedures. All participants provided informed consent before completing the questionnaire. The baseline survey was conducted from February 20–25, 2020. The authors requested that the survey company collect approximately 17,000 completed responses. Nikkei Research distributed email invitations to registered panel members until the target sample size was reached, yielding 17,463 effective responses. Because the number of invitations required to reach the target sample was not available to the authors, the response rate for the baseline survey could not be calculated.
Follow-up surveys were conducted from February 19–26, 2021; February 18–28, 2022; February 22 to March 6, 2023; and February 27 to March 11, 2024. The numbers of distributed invitations and effective completed responses were 16,949 and 6,103 in 2021, 6,103 and 4,281 in 2022, 4,221 and 3,410 in 2023, and 3,486 and 2,929 in 2024, respectively. At later waves, some respondents from the preceding wave could not be contacted because they had withdrawn from the survey company’s monitor panel, while some invited panelists did not complete the follow-up survey. The 2020 survey was conducted shortly before the World Health Organization (WHO) declared COVID-19 a pandemic on March 11, 2020 [18]. The 2021–2023 waves were conducted during the pandemic period. The 2024 wave was conducted after the WHO determined on May 5, 2023 that COVID-19 no longer constituted a Public Health Emergency of International Concern [19].
After respondents were matched across all five waves, 2,817 individuals remained in the merged longitudinal dataset. We subsequently excluded observations with missing information on variables required for the analysis, including employment status, financial literacy, household income, and household assets. The final analytic sample consisted of 1,673 respondents. The reduction from the matched five-wave sample to the final analytic sample resulted from complete-case exclusion. Gender, rural residence, university education, and the presence of children were obtained from the 2020 baseline wave. Age and current socioeconomic and psychological characteristics were measured using the 2024 wave. Variables indicating changes in living arrangements, employment, financial conditions, health, future anxiety, financial satisfaction, and depression were constructed using information from 2020 and 2024.
To assess potential selection associated with attrition and complete-case restriction, we compared baseline characteristics between respondents included and not included in the final analytic sample (S1 Appendix). The groups differed in several observed characteristics, including sex, age, university education, financial literacy, and household assets. These differences are considered in the interpretation of the results and discussed as a limitation.
2.2. Variables
Loneliness was measured using the three-item UCLA Loneliness Scale developed by Hughes et al. [20]. Participants were asked how often they felt that they lacked companionship, felt left out, and felt isolated from others. Response options were “hardly ever or never,” “some of the time,” and “often.” Participants were classified as lonely if they answered “some of the time” or “often” to at least one of the three items; those who answered “hardly ever or never” to all three items were classified as not lonely. This coding has been used in previous large-scale analyses of Japanese online panel data employing the UCLA-3 items [1,21].
This is an inclusive threshold that captures any reported experience of loneliness at a moderate or higher frequency. It therefore produces higher prevalence estimates than a stricter definition based only on “often” responses. We consequently conducted a sensitivity analysis using the stricter definition (S2 Appendix). Cronbach’s alpha coefficients for the three items ranged from 0.8444 to 0.8637 across the five waves, indicating good internal consistency (S3 Appendix).
Our study incorporated a range of participants’ demographic, socioeconomic, and psychological characteristics to analyze loneliness patterns in greater depth. These variables align with those used in previous research [1,8,21–23].
Gender, rural residence, university education, and the presence of children were measured at baseline in 2020. Age and current socioeconomic and psychological characteristics were measured in 2024. Variables indicating changes in living arrangements, employment, financial conditions, health, future anxiety, financial satisfaction, and depression were constructed using information from 2020 and 2024. The detailed definitions of the variables are provided in Table 1.
2.3. Methods
We used univariate and bivariate descriptive statistics to examine annual loneliness prevalence and five-year transition patterns. Each respondent was classified as lonely or not lonely in each wave, producing 32 possible five-year response sequences. We report the ten most frequent sequences and summarize broader patterns, including loneliness in all five waves, no loneliness in any wave, onset, and recovery.
To examine associations between the five most frequent loneliness sequences and categorical sociodemographic or psychological characteristics, we created a binary indicator for membership in each sequence and conducted Pearson’s chi-square tests comparing respondents belonging to the sequence with all other respondents.
All descriptive analyses were unweighted. No multivariable regression analysis was performed, and the findings are interpreted descriptively rather than causally.
As a sensitivity analysis, we applied a stricter loneliness threshold, classifying respondents as lonely only if they reported feeling lonely “often” on at least one of the three UCLA items. We then recalculated annual prevalence rates and five-year sequence distributions to assess the sensitivity of the descriptive findings to the operational definition of loneliness.
3. Results
3.1. Loneliness transitions over the years
Table 2 presents annual loneliness prevalence among the 1,673 respondents included in the final analytic sample. As shown in Table 2, loneliness prevalence peaked in 2021 and was lower and relatively stable from 2022 to 2024.
After examining annual loneliness prevalence, we tracked respondents’ loneliness status across the five survey waves. Because loneliness was coded as a binary variable in each wave, 32 possible five-year response sequences were observed. We report the ten most frequent sequences to summarize the predominant patterns in the analytic sample. Table 3 presents the distribution of loneliness, revealing the number and percentage of participants belonging to the most prevalent patterns.
The ten most frequent sequences accounted for 80.3% of the analytic sample. The most common sequence was persistent loneliness across all five waves (11111; 45.1%), followed by no loneliness in any wave (00000; 14.0%). Each of the remaining sequences accounted for 3.4% or less of the sample. Overall, stable sequences were more common than transitional sequences, although several less frequent patterns indicated onset or recovery at different points during the study period.
3.2. Descriptive statistics of sociodemographic variables
Table 4 summarizes the characteristics of the final analytic sample. Of the 1,673 respondents, 71.1% were male, and the median age in 2024 was 57 years (IQR 49–66). Approximately 54.9% lived in rural areas, 64.7% had a university degree, 60.1% had children, and 21.1% lived alone. The male-skewed composition of the sample and its implications for generalizability are addressed in the Discussion.
3.3. Chi-square test results
Table 5 presents the distribution of the top five loneliness patterns by various sociodemographic and psychological variables.
3.3.1. Long-term loneliness.
Membership in the persistent-loneliness sequence (lonely in all five waves) differed significantly by age, having children, and university education. Persistent loneliness was more common among respondents younger than 65 years (50.8%; 602/1,185) than among those aged 65 years or older (31.1%; 152/488; χ² = 53.93, p < 0.01). It was also more common among respondents without children (54.0%; 361/668) than among those with children (39.1%; 393/1,005; χ² = 36.17, p < 0.01). The sequence was slightly more common among respondents without a university degree (48.5%; 286/590) than among those with a university degree (43.2%; 468/1,083; χ² = 4.27, p < 0.05).
3.3.2. No loneliness in any wave.
Membership in the “not lonely in any wave” sequence also differed by age, having children, and university education. The sequence was more common among respondents aged 65 years or older (23.2%; 113/488) than among those younger than 65 years (10.3%; 122/1,185; χ² = 47.35, p < 0.01). It was more common among respondents with children (17.7%; 178/1,005) than among those without children (8.5%; 57/668; χ² = 28.00, p < 0.01), and among respondents with a university degree (15.7%; 170/1,083) than among those without a university degree (11.0%; 65/590; χ² = 6.93, p < 0.01).
3.3.3. Fluctuating but long-term loneliness.
Associations were less consistent for the smaller trajectory groups. Some descriptive differences were observed, including a relatively high proportion of worsening depression in the “not lonely in 2020; lonely from 2021 to 2024” group, but these findings should be interpreted cautiously because of the small number of respondents in each sequence.
3.4. Sensitivity analysis
Under the stricter definition, in which respondents were classified as lonely only if they answered “often” to at least one of the three UCLA items, annual loneliness prevalence was 28.2% in 2020, 32.0% in 2021, 28.1% in 2022, 25.5% in 2023, and 26.7% in 2024. Thus, the stricter threshold substantially reduced the absolute prevalence estimates, although prevalence still peaked in 2021 and was lower in all subsequent waves.
The distribution of specific five-year sequences also differed from that obtained using the primary definition. Under the stricter definition, the most common sequence was no loneliness in any wave (Not lonely in any wave; 50.4%), followed by loneliness in all five waves (Lonely in all five waves; 11.3%). Nevertheless, under both definitions, stable sequences—either loneliness in all five waves or no loneliness in any wave—were more common in combination than the remaining transitional sequences. Full sensitivity-analysis results are presented in S2 Appendix.
4. Discussion
This study provides descriptive evidence on five-year loneliness trajectories in Japan from 2020 to 2024 using a nationwide online panel. Rather than focusing on causal explanations, we documented annual prevalence trends and longitudinal response patterns to better understand how loneliness evolved over the course of the COVID-19 period and its aftermath.
The annual prevalence of loneliness peaked in 2021, declined in 2022, and remained relatively stable from 2022 to 2024. While temporary fluctuations were observed during the early pandemic period, the overall pattern did not suggest a sustained escalation over the five-year span. These descriptive patterns are similar to reports from other settings indicating short-term disruptions followed by partial stabilization [8,17,24].
The predominance of stable sequences indicates that loneliness patterns did not change uniformly across respondents. The two stable sequences—loneliness in all five waves and no loneliness in any wave—together accounted for most respondents, whereas onset and recovery sequences were individually much less common. This heterogeneity cautions against interpreting changes in annual prevalence as evidence that the same individuals uniformly became more or less lonely.
The annual pattern observed in this study is broadly consistent with earlier evidence showing that loneliness worsened during the early phase of the pandemic, although later trajectories varied across studies and populations. In Japan, the JACSIS 2020–2021 study reported an increase in loneliness between the first and second pandemic years [12], while a separate two-year longitudinal study found that loneliness and social isolation remained unresolved across prolonged pandemic phases [17]. In the Swiss Household Panel, increases in loneliness were disproportionately concentrated among younger respondents [16]. Our findings extend this evidence by covering five annual waves and suggest that, among respondents retained in the present panel, the increase observed in 2021 was followed by lower and relatively stable prevalence from 2022 to 2024.
Longitudinal trajectory analysis indicated that most respondents demonstrated stable patterns across the five waves, either consistently reporting loneliness or consistently not reporting loneliness. A smaller proportion exhibited transitional patterns, including temporary increases or decreases during the study period. These descriptive patterns highlight heterogeneity in lived experiences during and after the pandemic.
Descriptive comparisons across sociodemographic and psychological characteristics did not reveal uniform or consistent distributional differences across trajectory groups. While certain characteristics appeared more common in some trajectory categories, no single factor clearly distinguished all patterns. These observations underscore the complexity of loneliness as a multidimensional and context-dependent experience.
By following the same respondents over five annual waves, this study provides a longer-term perspective than a single cross-sectional snapshot. Monitoring stability and transition patterns over time may be important for identifying groups that experience persistent loneliness, while recognizing that many individuals demonstrate stable trajectories.
Several limitations should be acknowledged. First, loneliness was measured using a brief three-item scale and dichotomized using an inclusive threshold. This operationalization captured respondents who endorsed any item at least “some of the time” and consequently produced relatively high prevalence estimates. The sensitivity analysis using a stricter threshold yielded substantially lower prevalence, although the annual peak in 2021 and the predominance of stable over transitional sequences remained.
Second, attrition across the five survey waves was substantial. The baseline comparison indicated that respondents included in the final analytic sample differed from those not included in several observed characteristics. Included respondents were older, more likely to be male and university educated, and had higher financial literacy and household assets. These differences indicate potential selection bias, and unobserved differences may also remain. The findings should therefore be interpreted as descriptive patterns among retained respondents rather than as nationally representative prevalence estimates.
Third, participation required continued access to and engagement with an online survey panel. Older adults who participate repeatedly in online surveys may differ from older adults with limited internet access or lower digital familiarity. They may also differ in health status, socioeconomic resources, and their capacity to participate in repeated surveys. Consequently, the age-related patterns observed in this study may not generalize to the broader older population in Japan.
Fourth, the analytic sample was substantially male-skewed, with men accounting for 71.1% of respondents. Because loneliness patterns may differ by sex, this imbalance may have influenced both prevalence estimates and trajectory distributions.
Fifth, the baseline response rate could not be calculated because the survey company continued recruitment until the target sample size was reached and the total number of invitations was unavailable to the authors. Finally, the analyses were descriptive and did not establish causal relationships between the COVID-19 period, individual characteristics, and loneliness trajectories.
Despite these limitations, the study offers five-year panel evidence on loneliness trajectories in Japan. Future research may build upon these descriptive findings using analytical approaches designed to examine potential mechanisms and causal pathways.
5. Conclusions
This study examined annual prevalence and five-year transition patterns of loneliness among respondents retained in a nationwide Japanese online panel from 2020 to 2024. Loneliness prevalence peaked in 2021, declined in 2022, and remained relatively stable from 2022 to 2024.
Loneliness in all five waves and no loneliness in any wave were the two most common sequences, while onset and recovery patterns were less frequent. These findings indicate substantial heterogeneity in individual loneliness experiences across the study period.
Because of attrition, the online survey mode, and the male-skewed analytic sample, the results should not be interpreted as nationally representative prevalence estimates. Nevertheless, the five-wave data provide descriptive evidence on the stability and transition of loneliness over an extended period spanning the COVID-19 emergency and its aftermath.
Supporting information
S1 Appendix. Baseline characteristics of participants included and not included in the final analytic sample.
https://doi.org/10.1371/journal.pone.0358269.s001
(DOCX)
S3 Appendix. Cronbach’s alpha for the UCLA-3 items.
https://doi.org/10.1371/journal.pone.0358269.s003
(DOCX)
S1 Data. Minimal anonymized dataset necessary to replicate the findings reported in this study.
https://doi.org/10.1371/journal.pone.0358269.s004
(DTA)
References
- 1. Khan MSR, Yuktadatta P, Kadoya Y. Who Became Lonely during the COVID-19 Pandemic? An Investigation of the Socioeconomic Aspects of Loneliness in Japan. Int J Environ Res Public Health. 2022;19(10):6242. pmid:35627778
- 2. Cacioppo JT, Hughes ME, Waite LJ, Hawkley LC, Thisted RA. Loneliness as a specific risk factor for depressive symptoms: cross-sectional and longitudinal analyses. Psychol Aging. 2006;21(1):140–51. pmid:16594799
- 3. Child ST, Lawton L. Loneliness and social isolation among young and late middle-age adults: Associations with personal networks and social participation. Aging Ment Health. 2019;23(2):196–204. pmid:29171764
- 4. Luhmann M, Hawkley LC. Age differences in loneliness from late adolescence to oldest old age. Dev Psychol. 2016;52(6):943–59. pmid:27148782
- 5. Voitsidis P, Nikopoulou VA, Holeva V, Parlapani E, Sereslis K, Tsipropoulou V, et al. The mediating role of fear of COVID-19 in the relationship between intolerance of uncertainty and depression. Psychol Psychother. 2021;94(3):884–93. pmid:33216444
- 6. Stickley A, Ueda M. Loneliness in Japan during the COVID-19 pandemic: Prevalence, correlates and association with mental health. Psychiatry Res. 2022;307:114318. pmid:34896846
- 7. Gruber J, Prinstein MJ, Clark LA, Rottenberg J, Abramowitz JS, Albano AM, et al. Mental health and clinical psychological science in the time of COVID-19: Challenges, opportunities, and a call to action. Am Psychol. 2021;76(3):409–26. pmid:32772538
- 8. Lal S, Nguyen TXT, Sulemana A-S, Yuktadatta P, Khan MSR, Kadoya Y. A Longitudinal Study on Loneliness during the COVID-19 Pandemic in Japan. Int J Environ Res Public Health. 2022;19:11248.
- 9. Palgi Y, Shrira A, Ring L, Bodner E, Avidor S, Bergman Y, et al. The loneliness pandemic: Loneliness and other concomitants of depression, anxiety and their comorbidity during the COVID-19 outbreak. J Affect Disord. 2020;275:109–11. pmid:32658811
- 10. Bu F, Steptoe A, Fancourt D. Who is lonely in lockdown? Cross-cohort analyses of predictors of loneliness before and during the COVID-19 pandemic. Public Health. 2020;186:31–4. pmid:32768621
- 11. Ernst M, Niederer D, Werner AM, Czaja SJ, Mikton C, Ong AD, et al. Loneliness before and during the COVID-19 pandemic: A systematic review with meta-analysis. Am Psychol. 2022;77(5):660–77. pmid:35533109
- 12. Murayama H, Suda T, Nakamoto I, Shinozaki T, Tabuchi T. Changes in social isolation and loneliness prevalence during the COVID-19 pandemic in Japan: The JACSIS 2020-2021 study. Front Public Health. 2023;11:1094340. pmid:36875370
- 13. Sun H-L, Zhang Q, Si TL, Bai W, Chen P, Lam MI, et al. Interactive changes in depression and loneliness symptoms prior to and during the COVID-19 pandemic: A longitudinal network analysis. Psychiatry Res. 2024;333:115744. pmid:38301287
- 14. Raina P, Wolfson C, Griffith L, Kirkland S, McMillan J, Basta N, et al. A longitudinal analysis of the impact of the COVID-19 pandemic on the mental health of middle-aged and older adults from the Canadian Longitudinal Study on Aging. Nat Aging. 2021;1(12):1137–47. pmid:37117519
- 15. Pierce M, Hope H, Ford T, Hatch S, Hotopf M, John A, et al. Mental health before and during the COVID-19 pandemic: a longitudinal probability sample survey of the UK population. Lancet Psychiatry. 2020;7(10):883–92. pmid:32707037
- 16. Köster F, Lipps O. How loneliness increased among different age groups during COVID-19: a longitudinal analysis. Eur J Ageing. 2024;21(1):2. pmid:38170323
- 17. Sugaya N, Yamamoto T, Suzuki N, Uchiumi C. Loneliness and Social Isolation Factors Under the Prolonged COVID-19 Pandemic in Japan: 2-Year Longitudinal Study. JMIR Public Health Surveill. 2024;10:e51653. pmid:39250195
- 18.
World Health Organization. WHO Director-General’s opening remarks at the media briefing on COVID-19 - 11 March 2020. https://www.who.int/news-room/speeches/item/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19---11-march-2020. Accessed 2026 February 22.
- 19.
World Health Organization. Statement on the fifteenth meeting of the IHR (2005) Emergency Committee on the COVID-19 pandemic. https://www.who.int/news/item/05-05-2023-statement-on-the-fifteenth-meeting-of-the-international-health-regulations-(2005)-emergency-committee-regarding-the-coronavirus-disease-(covid-19)-pandemic Accessed 2026 February 22.
- 20. Hughes ME, Waite LJ, Hawkley LC, Cacioppo JT. A Short Scale for Measuring Loneliness in Large Surveys: Results From Two Population-Based Studies. Res Aging. 2004;26(6):655–72. pmid:18504506
- 21. Khan MSR, Kadoya Y. Loneliness during the COVID-19 Pandemic: A Comparison between Older and Younger People. Int J Environ Res Public Health. 2021;18(15):7871. pmid:34360164
- 22. Nabeshima H, Kuramoto Y, Khan MSR, Kadoya Y. Does the Easing of COVID-19 Restrictive Measures Improve Loneliness Conditions? Evidence from Japan. Sustainability. 2023;15(24):16891.
- 23. Sulemana A-S, Nguyen TXT, Lal S, Khan MSR, Kadoya Y. Loneliness during the COVID-19 Pandemic: A Comparison of Urban and Rural Areas. Sustainability. 2023;15(16):12218.
- 24. Barfoot KL, Child J, Dunn R, Colombage RL. Perceived loneliness, anxiety and depression symptomology before, during and after COVID-19 lockdowns in England. Current Psychology. 2026;45:480.