Data Resource Profile: The Japan COVID-19 and Society Internet Survey (JACSIS)

The Japan COVID-19 and Society Internet Survey (JACSIS) was established in 2020 to investigate the multifaceted impacts of the COVID-19 pandemic, enabling the comparison of pre- and post-pandemic data through linkage with the previous Japan “Society and New Tobacco” Internet Survey (JASTIS) studies (6494 participants from pre-pandemic period; 7351 from pandemic onset). The study commenced in August 2020 in Japan with 28 000 initial participants (25 482 after excluding satisficers) comprising Japanese adults aged 16–79 years in Wave 1, expanded to include those aged ≥80 years in Wave 8, with approximately equal gender distribution. Follow-up surveys have been conducted twice a year, maintaining ∼30 000 participants per wave through replacement sampling (response rates: ∼60% for previous participants, ∼30% for new panelists); the dataset includes 74 003 unique individuals due to the re-recruitment of previous participants. Data collection encompasses demographics, socioeconomic factors, validated health scales such as Kessler-6, COVID-19 assessments (fear of COVID-19, vaccine attitudes, etc.), lifestyle behaviors (tobacco/alcohol use, gambling), social measures (Lubben Scale, childhood experiences), and pandemic impacts on healthcare utilization and daily activities (hospital visits, influenza vaccination, work from home). The dataset is available to external researchers through collaborative research frameworks; contact the corresponding author at [email protected] for the application procedures. To address critical research needs about the societal impacts of COVID-19, we established the Japan COVID-19 and Society Internet Survey (JACSIS) cohort in August 2020. This internet-based study aims to investigate the multifaceted impacts of the COVID-19 pandemic through four primary objectives: (i) conducting rigorous, data-driven scientific analyses of how the pandemic has affected socioeconomic factors, health outcomes, and overall quality of life across diverse population segments; (ii) disseminating accurate and timely information to the public; (iii) supporting evidence-based policymaking; (iv) contributing to the global body of evidence on COVID-19. A key strength of JACSIS is its ability to link data with the pre-pandemic Japan “Society and New Tobacco” Internet Survey (JASTIS) studies conducted between 2015 and 2020 [1]. The study is funded by the Japan Society for the Promotion of Science (JSPS), Health Labour Sciences Research Grants, Japan Agency for Medical Research and Development (AMED), Japan Science and Technology Agency (JST), and various other governmental and institutional sources. Participant recruitment was conducted through the commercial research agency Rakuten Insight, whose panel comprises ∼2.2 million individuals from the Japanese population. The cohort consists of Japanese adults aged 16–79 years in Wave 1, expanded to include those aged ≥80 years in Wave 8, with approximately equal gender distribution. The JACSIS cohort participants were followed through questionnaire surveys conducted twice a year from August 2020. As of February 2024, eight waves have been completed. Survey waves were distributed as follows: Wave 1 (August and September 2020), Wave 3 (September and October 2021), Wave 5 (September and October 2022), and Wave 7 (September–November 2023), while participants were also invited to participate in JASTIS surveys conducted in Wave 2 (February 2021), Wave 4 (February 2022), Wave 6 (February 2023), and Wave 8 (January and February 2024). Each wave maintained ∼30 000 participants through replacement sampling, with retention rates varying from 65% to 81% between consecutive waves and the cumulative retention from Wave 1 declining to 24% by Wave 8 (Table 1). However, the seemingly low retention rate reflects participants who responded to every single wave consecutively. Our recruitment strategy allows previous participants to rejoin after missing waves, resulting in 11 749 participants providing valid responses to both Wave 1 and Wave 8, and a total of 74 003 unique individuals participating across all waves. The response rates were ∼60% for participants from previous waves and 5%–10% for newly recruited panelists from Rakuten Insight. Despite the dynamic nature of the cohort composition across the waves, the distribution of demographic variables and socioeconomic statuses in our cohort is similar to those of nationally representative samples (Table 2 and Supplementary Table S1). Number of eligible responses, response rate, dropout, and follow-up rate of each wave. Number of retained participants are calculated only among participants who provided eligible responses in each wave. The retention rate represents the proportion of participants who provided eligible responses in both the previous wave and the current wave. The sum of retained and newly added participants does not equal the total number of eligible responses due to participants who did not provide eligible responses in previous waves but did so in the current wave. We did not recruit new participants from Rakuten Insight panelists because we fulfilled the target sample size by recruiting participants from previous waves including the Japan Society and New Tobacco Internet Survey 2015–2020. The recruitment period was similar to those of other waves. Characteristics of the cohorts in Wave 1 and Wave 8, compared with the Japanese population. Median (Q1, Q3); n (%). Data of Japanese population came from National census 2020 (age, sex) and Comprehensive Survey of Living Conditions 2019 (academic attainment, marital status, household equivalent income, type of employment). Low indicates income 0–200 million Japanese Yen (JPY), middle 200–400 million JPY, high >400 million JPY. The JACSIS cohort has secured research funding through to 2028, ensuring the continuation of data collection at least through to Wave 13. Wave 9 has already been completed and is currently in the data-cleaning phase, while preparations for Wave 10 are underway, with the research team finalizing the survey items. The investigators anticipate securing additional funding for 2029 and beyond. For participants aged 16–17 years, informed consent was obtained directly from them without parental consent in accordance with the Ethical Guidelines for Medical and Biological Research Involving Human Subjects implemented by Japan’s Ministry of Education, Culture, Sports, Science and Technology and Japan’s Ministry of Health, Labour and Welfare [2]. Ethics approval information is provided below. The JACSIS dataset is a primary data-collection initiative using a repeat panel design. Data collection was conducted via Rakuten Insight’s secure web platform, with participants receiving compensation through a credit point system known as “E-points,” which could be used for internet shopping and cash conversion. In determining the cohort size, we established a final target sample size of 28 000 participants in Wave 1. This figure was determined based on budgetary constraints while ensuring adequate sample sizes within each sex and age stratum for reliable statistical analyses, including propensity score modeling and event-rate estimation. As a multipurpose study designed to address diverse research questions, formal power calculations for specific hypotheses were not conducted. The JACSIS dataset is produced through internet-based questionnaires administered to participants from Rakuten Insight’s panel of ∼2.2 million Japanese individuals. For initial recruitment in Wave 1, a stratified sampling strategy was employed based on sex, age, and geographic region to match the intended representative distribution. After conducting a preliminary response rate assessment with 28 000 randomly selected panelists, Rakuten Insight adjusted the number of invitations to achieve the target sample size, resulting in a final participation rate of 12.5% (28 000/224 389). For subsequent waves, our recruitment strategy has followed a two-step process. First, we invite all individuals who have participated in any previous wave (including JASTIS 2015–2019) to rejoin the study. If this approach does not yield sufficient participants to meet our target sample size (which may vary slightly based on budget constraints and other factors), we then recruit new participants from Rakuten Insight’s panel by using the same stratified sampling method as employed in Wave 1 (stratified by sex, age, and geographic region). Therefore, while a part of the cohort changes at each wave, the cohort does not change entirely, as many participants return after missing one or more waves, maintaining similar characteristics to the national cohort. Of note, we did not recruit completely new participants from Rakuten Insight panelists in Wave 8 because the target sample size was fulfilled by recruiting only from participants in past waves. Comparisons between replacement participants (those who did not respond to the previous wave but did participate in the current wave) and dropouts (those who responded to the previous wave but not the current wave) show that, while replacement participants tend to be slightly younger, their demographic and clinical profiles are otherwise similar (Supplementary Table S2A–G). Data-quality-assurance methods can be broadly divided into two categories: those implemented by Rakuten Insight as part of their standard panel management and those specifically designed by us for this study. Rakuten Insight employs comprehensive quality-control procedures at multiple stages: during recruitment through various channels, at registration with automated fraud-prevention algorithms, and in ongoing participation via introductory surveys, regular qualitative checks, and demographic verification. Additional details on Rakuten Insight’s panel-quality-assurance processes are available in their published documentation [3]. We implemented additional study-specific quality measures by identifying and excluding “satisficers” through the algorithmic detection of inconsistent or unreasonable response patterns and attention checks throughout the survey. The algorithm to detect satisficers was modified for each wave based on the specific items and available data (such as response time) in that wave. Details of the algorithms used in each wave are presented in Table 3. Algorithms to identify satisficers in each survey. The question for the attention check was: “Select the second option from the bottom.” Participants were asked: For those who selected “Use almost every day” or “Sometimes” for the question “Are you currently consuming or using alcohol or drugs?” please respond for each of the following items: alcohol (such as beer, sake, shochu, wine, or whiskey), sleeping pills or anti-anxiety medications, narcotics such as morphine (prescribed by a doctor for cancer pain), narcotics such as morphine (prescribed by a doctor for pain other than cancer), narcotics such as morphine (obtained by methods not prescribed by a doctor), inhalation of organic solvents such as thinner or toluene (excluding appropriate use for work purposes), dangerous drugs (such as legal highs or magic mushrooms), cannabis (marijuana), and stimulants, cocaine, or heroin. Participants were asked: For those who responded “Yes” to all the following items of the question “Do you currently have any chronic illnesses?” please answer each item: 1. Hypertension, 2. Diabetes, 3. Asthma, 4. Bronchitis or Pneumonia, 5. Atopic Dermatitis, 6. Periodontal Disease, 7. Dental Cavities, 8. Otitis Media, 9. Angina, 10. Myocardial Infarction, 11. Stroke (Cerebral Infarction or Cerebral Hemorrhage), 12. COPD (Chronic Obstructive Pulmonary Disease), 13. Cancer or Malignant Tumors, 14. Chronic Pain (such as back pain or headaches lasting >3 months), 15. Depression, 16. Other Mental Disorders besides Depression. The JACSIS dataset comprises a comprehensive collection of self-reported data gathered through online questionnaires administered twice a year (Table 4). Survey items were determined through multiple iterative discussions among the research team, with careful selection of the most appropriate scales for each construct of interest, balancing participant burden with measurement validity and reliability. Survey items include: demographic information (age, sex, gender identity, sexual orientation, residence); socioeconomic indicators (education, income, marital status, employment status); anthropometric measurements (self-reported height and weight); health-related factors (physical symptoms, medical conditions, treatment status of chronic diseases such as hypertension and diabetes, hospitalization experiences, healthcare utilization); validated psychometric scales [Kessler-6 for depression screening [4], EQ5D-5L for health-related quality of life [5], COPD-Q for respiratory symptoms [6]; lifestyle behaviors (tobacco use with Fagerstrom Test [7], alcohol use with AUDIT [8] and CAGE questionnaires [9], gambling behaviors); COVID-19-specific measures (infection history, COVID-19-like symptoms, preventive behaviors, vaccination status, vaccine attitudes using the Fear of COVID-19 Scale) [10]; and social measures (Lubben Social Network Scale [11], UCLA loneliness scale version3 [12], adverse and positive childhood experiences [13, 14])]. Additionally, parallel information about partners, spouses, and children was collected for relevant items, enabling a thorough analysis of family dynamics and their influence on health and well-being. The dataset includes both cross-sectional and longitudinal data components, with some variables collected at every wave and others at specific intervals based on the research priorities. All measures are obtained via self-report rather than clinical assessments or biological sampling. The first survey wave included ∼600 questions; however, due to branching logic in the questionnaire design, each respondent typically answered ∼100 questions rather than the entire set. In Wave 7, in which the completion times were first measured systematically, the mean survey completion time was 49.4 minutes (standard deviation = 28.5 minutes). List of collected variables that were utilized in published studies. COPD-Q, chronic obstructive pulmonary disease knowledge questionnaire; EQ5D-5L, EuroQol 5 dimensions 5-level; HPV, human papilloma virus; JACSIS, the Japan COVID-19 and Society Internet Survey; VDT, visual display terminal. Note: All variables were assessed by using self-administered, web-based questionnaires conducted via Rakuten Insight’s secure web-based platform. Variables without a specific scale name were measured by using original items developed by the JACSIS study team. Detailed survey questionnaires are available in Japanese upon request. The JACSIS dataset is linked with the pre-pandemic JASTIS studies conducted between 2015 and 2020. This linkage enables pre- and post-pandemic comparisons, with 6494 participants from JASTIS 2019 (pre-pandemic) and 7351 from JASTIS 2020 (pandemic onset) also participating in JACSIS 2020. Each research participant is assigned a unique numerical identifier (ID) that is independent of their personal information, with these IDs assigned consecutively from JASTIS 2015 and new IDs generated only for new participants who have never previously participated in any wave of JACSIS or JASTIS. When data are received from Rakuten Insight, survey responses are linked to these IDs, which are then used to merge data across different waves for longitudinal analyses. While these IDs are linked to Rakuten Insight login credentials, the correspondence table between IDs and login information is managed exclusively by Rakuten Insight, with researchers having no access to this identifying information. JACSIS has revealed significant socioeconomic disparities in pandemic impacts, healthcare utilization, and adherence to preventive measures, while also highlighting the role of factors such as trust in government and information sources in shaping public responses to the pandemic. The research has provided valuable insights into the effects of changing work environments, the prevalence of mental health issues, and the complexities surrounding COVID-19 vaccine uptake and hesitancy in Japan, with these findings disseminated through 117 published papers by 30 November 2024. The full list of papers is available on the JACSIS website (https://jacsis-study.jp/output/index.html). Our main findings include the following: We examined the association between participation in Japan’s subsidy program for domestic travel (“Go To Travel” campaign) and the incidence of COVID-19-like symptoms by using Wave 1 data [15]. We found that participants in the subsidy program (12.9% of respondents) had higher odds of experiencing COVID-19-like symptoms compared with non-participants. For example, participants had 1.83 [95% confidence interval (CI), 1.34–2.48] times higher odds of experiencing high fever and times higher odds of and this the Japanese government to the To in 2020. We the association between socioeconomic due to the COVID-19 pandemic and pain in Japan by using Wave 1 data We found that of the participants pain and those who household income work and were more to pain odds and The JACSIS dataset has a diverse of variables COVID-19-specific measures, enabling researchers to on a of health behaviors and to the pandemic. ongoing data analyses from the JACSIS dataset include into the between symptoms and behaviors such as alcohol and use (including the of as as studies and the between use, health and analyses include into the burden of and Additionally, the research team is to new survey items into waves, including screening for symptoms of sexual and scales and intended This study significant a sample size of ∼30 000 participants for each survey wave, providing statistical power and analyses. The longitudinal follow-up enables the of and changes our of strength is the of variables a comprehensive of behaviors, and socioeconomic the study from the of researchers from multiple across Japan, diverse from various including public clinical and social A unique of this study is its ability to data from and after the COVID-19 pandemic, as a of participants had already part in JASTIS studies conducted between 2015 and 2020 (January to This allows valuable insights into the of the pandemic on various of Additionally, the of and information, collected via online method validated by a for these significant to the enabling research on Despite its this study has The primary is the for sampling in internet-based As we information about individuals who are not of the Rakuten Insight our sample may not be representative of the population. However, we can to this through the application of The details of are provided in the Supplementary is that our data provide measures due to the self-reported nature of online While our questionnaires to both experiences (such as of COVID-19, and mental health and indicators (such as employment status and of government the validity of measures is by the of linkage with data or public For research questions clinical or investigators appropriate medical or governmental data sources rather than self-reported survey However, this does not the of our for changes in experiences, and behaviors during the COVID-19 pandemic. the of the data collection As JACSIS recruitment in August 6 after the COVID-19 pandemic in Japan, the characteristics may already However, our study this through questionnaire items life changes and data linkage with pre-pandemic JASTIS studies. as with all self-reported is a of or in responses, which could the of some to the data is by the study investigators and in using the data a research that is with and To for data access or to more information about the please contact the can provide details about the application available and The JACSIS study received initial approval from the of the Cancer on 2020 the institutional the study also obtained approval from the Ethics of of and October The JASTIS study was by the Research Ethics of the Cancer For both all participants provided informed consent part in the survey. The consent was conducted online and participants were to to the of the study could with the survey For the consent of participants aged 16–17 years, details in the The of this cohort was by the and of Ethics the study and the the formal and funding for the and conducted the and developed the and were for and provided overall for the and the the first of the All to and the and have and to the published of the Supplementary is available at was by the Japan Society for the Promotion of Science Health Labour Sciences Research the Japan Agency for Medical Research and Development the Japan Science and Technology Agency the and Agency number the Research to the of the to COVID-19 of the Research on for COVID-19 period the of the National for the for Research Promotion of from for Health Promotion and the research program on Health to and the of Health conducted at the for The findings and of this are the of the and not the of the research The data this be on to the corresponding to approval by the study investigators and in accordance with and to access the data a research and be to the and were used to the and of this were not used for data collection or All was and by the who full for the