Behavioral factors associated with SARS‐CoV‐2 infection in Japan

Abstract Background The relative burden of COVID‐19 has been less severe in Japan. One reason for this may be the uniquely strict restrictions imposed upon bars/restaurants. To assess if this approach was appropriately targeting high‐risk individuals, we examined behavioral factors associated with SARS‐CoV‐2 infection in the community. Methods This multicenter case–control study involved individuals receiving SARS‐CoV‐2 testing in June–August 2021. Behavioral exposures in the past 2 weeks were collected via questionnaire. SARS‐CoV‐2 PCR‐positive individuals were cases, while PCR‐negative individuals were controls. Results The analysis included 778 individuals (266 [34.2%] positives; median age [interquartile range] 33 [27–43] years). Attending three or more social gatherings was associated with SARS‐CoV‐2 infection (adjusted odds ratio [aOR] 2.00 [95% CI 1.31–3.05]). Attending gatherings with alcohol (aOR 2.29 [1.53–3.42]), at bars/restaurants (aOR 1.55 [1.04–2.30]), outdoors/at parks (aOR 2.87 [1.01–8.13]), at night (aOR 2.07 [1.40–3.04]), five or more people (aOR 1.81 [1.00–3.30]), 2 hours or longer (aOR 1.76 [1.14–2.71]), not wearing a mask during gatherings (aOR 4.18 [2.29–7.64]), and cloth mask use (aOR 1.77 [1.11–2.83]) were associated with infection. Going to karaoke (aOR 2.53 [1.25–5.09]) and to a gym (aOR 1.87 [1.11–3.16]) were also associated with infection. Factors not associated with infection included visiting a cafe with others, ordering takeout, using food delivery services, eating out by oneself, and work/school/travel‐related exposures including teleworking. Conclusions We identified multiple behavioral factors associated with SARS‐CoV‐2 infection, many of which were in line with the policy/risk communication implemented in Japan. Rapid assessment of risk factors can inform decision making.

.64]), and cloth mask use (aOR 1.77 [1.11-2.83]) were associated with infection. Going to karaoke (aOR 2.53 [1.25-5.09]) and to a gym (aOR 1.87 [1.11-3.16]) were also associated with infection. Factors not associated with infection included visiting a cafe with others, ordering takeout, using food delivery services, eating out by oneself, and work/school/travel-related exposures including teleworking.

| INTRODUCTION
Coronavirus disease 2019 , caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has resulted in substantial morbidity and mortality globally. 1 Japan has been no exception, but the relative burden of COVID-19 after 2 years has not been as severe as in many other countries, with fewer cumulative cases and deaths relative to the population despite its aging population. 2 Many factors may have contributed to this, such as the tireless efforts of public health centers in extensive contact tracing including backward tracing (source investigation), high mask-wearing adherence, maintaining greater physical distance, and strict infection prevention and control measures at health-care/long-term care facilities. 3,4 Among these, one intriguing hypothesis is the unique policy with a focused approach targeting restaurants and bars to reduce business hours at night and prohibiting the serving of alcohol. 5 As the Japanese government's response against COVID-19 is based on the Act on Special Measures for Pandemic Influenza and New Infectious Diseases Preparedness and Response, Japan has declared a state of emergency several times during the course of the pandemic. 6 However, unlike many other countries, there were no strict restrictions imposed on individual citizens such as lockdowns and obligatory curfews. Rather, there were voluntary requests to stay at home and engage in basic infection prevention measures such as proper mask wearing/hand hygiene. In comparison, restaurants and bars were ordered to suspend business if they cannot operate without serving alcohol, to stop serving alcohol, and to reduce business hours at night until 8:00 p.m. in prefectures with high transmission. Non-compliant businesses were disclosed publicly. This specific approach towards restaurants and bars was based on individual case data and cluster investigations with a theoretical rationale that dining or drinking alcohol at restaurants and bars with others (i.e., social gatherings involving food or drinks) provides occasion to interact face to face for a prolonged period without masks and that the influence of alcohol can further lead to laxity of infection prevention measures. 7 Also, an increase in the frequency and proportion of cases with no history of close contact 8 made containment through cluster investigation increasingly challenging and highlighted the lack of understanding regarding risk factors for infection at the community level. These circumstances led to the need to confirm through epidemiological data, with inclusion of a control group, whether behaviors such as social gatherings are indeed risk factors for SARS-CoV-2 infection to inform public health policy and provide evidence-based risk communication. Therefore, we initiated a multicenter case-control study to evaluate risk factors associated with SARS-CoV-2 infection, focusing on social gatherings involving food or drinks. We examined various social settings and further explored other behaviors as potential risk factors.

| Study design and setting
Our study, Factors Associated with SARS-CoV-2 Infection And The Effectiveness of COVID-19 vaccines (FASCINATE study), is a multicenter case-control study in health-care facilities in Japan with two objectives: (1) to elucidate risk factors associated with SARS-CoV-2 infection and (2) to estimate the effectiveness of COVID-19 vaccines.
Participating health-care facilities are routinely testing outpatients using polymerase chain reaction (PCR) to diagnose SARS-CoV-2 infection. For this report, data from six health-care facilities in the Kanto region (Tokyo and neighboring metropolitan prefectures) on individuals recruited during June 8-August 1, 2021, were analyzed.

| Inclusion and exclusion criteria
Individuals who were tested for SARS-CoV-2 were included. Exclusion criteria were (1) individuals younger than 20 years (as alcohol drinking is illegal for these individuals), (2) individuals who did or could not consent to participate in the study, (3) individuals who could not complete the questionnaire by themselves, (4) individuals who had already participated in this study, (5) individuals who required immediate treatment, and (6) individuals with history of close contact (because an infection, if confirmed, is most likely due to this specific contact rather than exposures asked about in the questionnaire). For this report, we excluded asymptomatic individuals and individuals vaccinated at least once as COVID-19 vaccination can influence behaviors.

| Questionnaire and classification of cases/ controls
A paper or web-based questionnaire (according to individual preference) was administered before PCR results were available to avoid social desirability bias, where individuals who test positive may be less likely to report potentially high-risk behaviors. The questionnaire was optimized based on a pilot study done at two sites. 9 We defined social gathering as getting together with one or more persons that does not cohabitate with the participant. Cases were defined as PCR-confirmed SARS-CoV-2 positive individuals, while controls were defined as PCRnegative individuals.

| Data analysis
Logistic regression to identify associations between behavioral risk factors and SARS-CoV-2 infection was conducted adjusting for age group, sex, presence of comorbidities, educational attainment, place of residence, past SARS-CoV-2 infection, health-care facility in which SARS-CoV-2 testing was done, and calendar week. These potential confounders were determined a priori. 10,11 Data analyses were performed using STATA version 17.0.

| Participant characteristics
A total of 992 symptomatic individuals were enrolled from six medical facilities during the study period; we excluded 44 due to unknown symptom onset, 16 due to being tested ≥15 days after symptom onset, and 154 due to being vaccinated ( Figure 1 were Japanese nationals and most foreigners were from East Asia.

| Factors related to 3Cs and five situations in the past 2 weeks
Since early in the pandemic, the Japanese government has been promoting avoidance of the "3Cs," representing (1) closed spaces, (2) crowded places, and (3) close-contact settings, which are considered high-risk based on characteristics of early clusters. 12 These "3Cs" were easy for the public to remember and the World Health Organization also began promoting this message. 13 Additionally, since fall 2020, the government started to promote avoidance of "five situations," namely, (1) social gatherings that include alcohol consumption, (2) large group gatherings that involve eating and/or drinking for an extended period of time, (3) conversing without a mask, (4) cohabitation in small living quarters, and (5) relocating to a different area. 12 We first examined factors among the above that could be measured (

| Association between social gatherings with food/drinks in various settings in the past 2 weeks and SARS-CoV-2 infection
We further examined the association between social gatherings that involve eating and/or drinking in various settings and SARS-CoV-2 infection ( Table 2). The odds of infection increased with increased frequency of social gatherings; those who attending social gatherings three or more times had higher odds of infection compared with those who did not (aOR 2.00 [95% CI 1.31-3.05]). We examined this association in detail, specifically by presence of alcohol, location of gathering, and time of day. The odds of infection were substantially higher among individuals who attended social gatherings with alcohol at least once compared with those who did not (

| Association between other behaviors related to food/drinks in the past 2 weeks and SARS-CoV-2 infection
To compare the above findings on social gatherings that involve eating and/or drinking, we examined whether other behaviors related to food/drinks were associated with SARS-CoV-2 infection (

| Association between size, duration, and mask wearing among participants of social gatherings in the past 2 weeks and SARS-CoV-2 infection
The government requested individuals attending social gatherings involving food/drinks to limit these gatherings to five people and to less than 2 hours and to consider use of masks except for when consuming food/drinks. 14 We assessed whether these factors were indeed associated with infection (Table 2). Specifically, those who attended a social gathering involving food/drinks and/or went to a cafe with others in the past 2 weeks were asked how many people attended the gathering, how long the gathering continued at maximum, and when the attendees had their masks on (cafe use was included here as we hypothesized that it also provides occasion to talk face to face for a prolonged period without masks

| Association between type of mask used in the past 2 weeks and SARS-CoV-2 infection
Three types of masks are mainly used among the public in Japan: medical/surgical masks, cloth masks, and polyurethane masks. The Japanese government recommends use of medical/surgical masks rather than cloth or polyurethane masks based on a computer simulation model, 15 but epidemiological data were lacking. Therefore, we examined the association between mask type and infection (

| Association between behaviors related to work/school and travel in the past 2 weeks and SARS-CoV-2 infection
We lastly examined whether behaviors related to work/school were associated with SARS-CoV-2 infection ( Table 4). The odds of infection

| DISCUSSION
In this multicenter case-control study, we investigated the association between various behavioral factors and SARS-CoV-2 infection in the community setting. First, we found that attending social gatherings with food/drinks was associated with SARS-CoV-2 infection. We strengthened our findings and those of previous reports 16-18 by showing the association in a dose-dependent manner, with the odds of infection increasing with increasing frequency of social gatherings in the past 2 weeks. We also investigated the details of specific settings of social gatherings that were associated with infection. First, attending social gatherings with alcohol was associated with infection.
When we examined the location of gatherings, attending gatherings at restaurants or bars was associated with infection. This finding was consistent with previous ecological/modeling and case-control studies where going to restaurants or bars was associated with infection. [16][17][18] As there were strict restrictions imposed upon restaurants and bars in Japan, there was a concern that people may choose to have gatherings at home or out on public streets/parks, with 10% of young people reporting that they had done the latter. 19 Indeed, individuals who had social gatherings exclusively at home had higher odds of infection, and attending gatherings outdoors or at parks was associated with infection. Attending a gathering at night was also associated with infection. The reason may be that social gatherings at night tend to be longer in duration, and individuals may become more intoxicated and care less about infection prevention measures. In contrast to the findings on social gatherings, ordering takeout, using food delivery services, and eating out by oneself were not associated with infection. This is expected as these behaviors would not substantially increase contact with others; our findings provide opportunity for the food industry to sustain its business. Details about how these gatherings took place also mattered; attending a gathering with five or more people or gathering lasting 2 hours or longer was associated with infection. Not wearing a mask or taking it off when seated at the gathering was also associated with infection, supporting the idea of "mask-dining" (taking off the mask only when putting food in the mouth or sipping drinks and keeping the mask on while talking, while waiting for food/drinks to be served, and after finishing meals), which has been recommended by the government when at restaurants and bars. On a related note, regular use of cloth or polyurethane masks was associated with infection, specifically among individuals who attended social gatherings or visited a cafe with others. In addition to source control, the association here suggested the protective effect of medical/ surgical masks, 20 and individuals who engage in high-risk behaviors may benefit from wearing medical/surgical masks. We could not evaluate associations with any mask use, as the mask-wearing adherence was high.
We identified some factors unrelated to social gatherings, namely, karaoke and gym use, which are also known to be hotspots for clusters. 7  Finally, identified risk factors may be country/region/culture-specific and time-dependent due to changes in COVID-19-related policies and behaviors, as well as emergence of SARS-CoV-2 variants.
In conclusion, we identified multiple behavioral factors associated with SARS-CoV-2 infection, particularly in various settings of social gatherings. These factors may be country/culture-specific and timedependent due to changes in COVID-19-related policies and behaviors, so continuous monitoring in various settings is important to inform decision making.

ACKNOWLEDGMENTS
This work was supported in part by a grant from the Japan Agency for

DATA AVAILABILITY STATEMENT
Individual-level data of patients included in this manuscript after deidentification are considered sensitive and will not be shared. The study methods and statistical analyses are all described in detail in Section 2 and throughout the manuscript.