---
title: "Loneliness trajectories in Japan during COVID-19: Five-year longitudinal patterns (2020–2024)"
id: "plos-one-7-the-dynamics-of-loneliness-a-longitudinal-exploration-of-transitions-in-japan"
canonical_url: "https://medichelpline.com/clinical-feed/plos-one-7-the-dynamics-of-loneliness-a-longitudinal-exploration-of-transitions-in-japan"
content_type: "clinical_feed_article"
specialty: "General"
source_name: "PLOS ONE (Medicine)"
source_url: "https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0358269"
published_at: "2026-09-11T14:00:00.000Z"
evidence_level: "Journal Feed"
license: "CC-BY-NC-4.0 / Informational Use"
---
# Loneliness trajectories in Japan during COVID-19: Five-year longitudinal patterns (2020–2024)
## Provenance & Clinical Metadata
- **Canonical URL:** https://medichelpline.com/clinical-feed/plos-one-7-the-dynamics-of-loneliness-a-longitudinal-exploration-of-transitions-in-japan
- **Specialty:** [General](https://medichelpline.com/clinical-feed/general.md)
- **Primary Source:** PLOS ONE (Medicine)
- **Source URL:** [Original Journal Publication](https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0358269)
- **Published At:** 2026-09-11T14:00:00.000Z
- **Evidence Rating:** Journal Feed
## Executive GIST (TL;DR)
- This study used five consecutive annual waves (February 2020–March 2024) of a nationwide online panel in **Japan** to describe longitudinal patterns in **loneliness** during and after the COVID-19 period. - Surveys were fielded in 2020, 2021, 2022, 2023, and 2024 through Nikkei Research; the baseline wave targeted ~17,000 responses (17,463 effective responses) and later waves had smaller effective samples. - After matching respondents across all five waves, 2,817 individuals remained; the final analytic sample was 1,673 adults with complete data on required variables after excluding incomplete cases. - **Loneliness** was measured using the three-item **UCLA Loneliness Scale** (UCLA-3); participants were classified as lonely if they reported “some of the time” or “often” on any item (inclusive threshold). Cronbach’s alpha across waves ranged ~0.844–0.864, indicating good internal consistency. - Main descriptive findings: loneliness prevalence peaked in 2021, declined in 2022, and remained relatively stable from 2022 through 2024 among respondents retained in the five-wave panel. - Individual-level sequences over five years were dominated by stable trajectories—either consistently lonely or consistently not lonely—while transient or switching sequences comprised a smaller share. - A sensitivity analysis using a stricter loneliness threshold (only “often” responses) preserved the 2021 peak and the predominance of stable over transitional sequences, though the order of specific sequences changed. - The analysis reports a range of demographic, socioeconomic, and psychological covariates measured at baseline (2020) and 2024; variables indicating change were constructed from 2020 and 2024 data. - Limitations noted: the analytic sample reflects respondents retained in a longitudinal online panel and complete-case exclusion reduced the sample; baseline comparisons show differences between included and excluded respondents on observed characteristics, so results should not be interpreted as nationally representative prevalence estimates. - The study provides descriptive longitudinal evidence on stability and transition in loneliness across the pandemic period but does not attempt causal inference.
## Clinical Analysis & Structured Key Points
The dynamics of loneliness: A longitudinal exploration of transitions in Japan across the COVID-19 period, 2020–2024 | PLOS One Browse Subject Areas ? Click through the PLOS taxonomy to find articles in your field. For more information about PLOS Subject Areas, click here . Article Authors Metrics Comments Media Coverage Peer Review Reader Comments Figures 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 . Download: PNG larger image TIFF original image Table 1. Definitions of variables. https://doi.org/10.1371/journal.pone.0358269.t001 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. Download: PNG larger image TIFF original image Table 2. Annual prevalence of loneliness from 2020 to 2024. https://doi.org/10.1371/journal.pone.0358269.t002 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. Download: PNG larger image TIFF original image Table 3. Top ten most common patterns of loneliness (2020–2024). https://doi.org/10.1371/journal.pone.0358269.t003 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. Download: PNG larger image TIFF original image Table 4. Characteristics of the final analytic sample (N = 1,673). https://doi.org/10.1371/journal.pone.0358269.t004 3.3. Chi-square test results Table 5 presents the distribution of the top five loneliness patterns by various sociodemographic and psychological variables. Download: PNG larger image TIFF original image Table 5. Chi-square analysis of sociodemographic and psychological characteristics across the top five loneliness patterns. https://doi.org/10.1371/journal.pone.0358269.t005 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 gr
## Related Clinical Research

- [Rising Demand for Plastic Surgery Among Young Men Fueled by Influencers and Looksmaxxing](https://medichelpline.com/clinical-feed/stat-news-1-opinion-more-young-men-are-coming-to-me-for-plastic-surgery-the-way-i-think.md)
- [Medicare expands ACCESS pilot to cover more chronic conditions and lists 39 providers](https://medichelpline.com/clinical-feed/stat-news-1-stat-medicare-to-expand-its-pilot-that-pays-for-technology-to-manage-chronic.md)
- [Social isolation and health behaviors across European generations: findings from ESS Round 11](https://medichelpline.com/clinical-feed/plos-one-15-social-isolation-and-health-risks-do-the-younger-generations-behave-differently.md)
- [Fear of COVID-19 and Coping Strategies in Female University Students: Role of Physical Activity an](https://medichelpline.com/clinical-feed/plos-one-20-factors-associated-with-fear-and-coping-strategies-during-pandemic-in-female.md)
- [Adverse nurse working hours and increased sickness absence: longitudinal e‑roster analysis](https://medichelpline.com/clinical-feed/bmj-open-16-associations-between-adverse-working-hours-and-nurses-sickness-absence-a.md)

## Navigation
- [← Back to General Feed](https://medichelpline.com/clinical-feed/general.md)
- [← All Clinical Specialties](https://medichelpline.com/clinical-feed.md)
## Medical & Regulatory Disclaimer

> [!CAUTION]
> MedicHelpline content is structured for research, educational, and professional discovery purposes. It does not constitute individual medical advice, clinical diagnosis, or treatment recommendations.
> Always verify dosing, contraindications, and regulatory alerts against official product labeling and primary regulatory sources before clinical decision-making.