This study used primary data from a web-based survey conducted among residents of Japan. The survey was administered in a questionnaire format through Cross Marketing Inc, a Japanese internet research company. Cross Marketing Inc and its partner companies maintain an active panel of over 5 million people, who have registered their sociodemographic information in the company’s database and responded to at least one survey within the past year [34].

Data collection was conducted from June to July 2020 using Cross Marketing Inc’s web-based response system. The company provided respondents with a survey URL via email requesting their participation. First, individuals who agreed to participate underwent a screening survey regarding their demographic attributes, such as gender and age (n=19,301). Next, stratified sampling was performed, evenly allocating 250 men and 250 women from each age group (20s, 30s, 40s, 50s, 60s, and 70s), resulting in a total sample size of 3000 participants. This nonprobabilistic sampling was used to minimize data shortages for specific age and gender groups; however, it has limitations in ensuring representativeness. Respondents were selected in order of earliest response, and the survey continued until the target number of participants was reached, excluding inconsistent or invalid responses. The sample size was determined primarily based on feasibility and financial constraints.

In this study, economic deterioration was considered a key explanatory variable. In Japan, many individuals in their 20s to 50s work full-time and earn income, making them more susceptible to economic issues. Therefore, samples from individuals in their 60s and 70s were excluded, leaving a dataset of 2000 respondents.

Economic deterioration was analyzed based on changes in personal income, which required focusing on income earners. Respondents who had been unemployed before the pandemic and remained unemployed at the time of the survey (including full-time homemakers) were excluded. Those who were employed before the pandemic but unemployed during the survey period were also excluded due to their small number (n=17). These exclusions reduced the sample to 1513 respondents.

Following listwise deletion for missing values, the final dataset included 1238 respondents, with most exclusions resulting from non-responses to the household income question (n=254). In total, 18.2% (275/1513; 275 excluded) of the sample was excluded in this process.

The K6 scale was used to assess mental health, serving as the dependent variable. This scale measures psychological distress over the past 30 days with 6 items rated on a 5-point scale, producing total scores from 0 to 24, with higher scores indicating greater distress [35]. Furukawa et al [36] validated the reliability of the K6 scale in the Japanese population. Following their methodology, a cutoff score of ≥9 was used to screen for mood and anxiety disorders.

The primary independent variables for hypothesis testing were economic deterioration due to COVID-19, social support, and negative interactions. Economic deterioration was measured by asking, "How has your income (total pre-tax income from work and non-work sources) changed as of April compared to before the COVID-19 pandemic? Please select the option that best describes your situation. If your salary is yet to be confirmed (eg, not yet deposited), please answer based on your expectations.” The response options were: 1=increased (expected to increase), 2=slightly increased (expected to slightly increase), 3=almost unchanged, 4=slightly decreased (expected to slightly decrease), and 5=Decreased (expected to decrease). The response distribution was: 1 (n=21), 2 (n=37), 3 (n=735), 4 (n=197), and 5 (n=248). Categories 1 and 2 were combined with Category 3 and recorded as 0, while Categories 4 and 5 were combined and recorded as 1, resulting in binary variables. The correlation between economic deterioration and household income (a continuous variable) was weak (r=−0.15).

Social support was assessed using a 4-item, 4-point scale, with total scores ranging from 4 to 16, with higher scores indicating greater social support. The items were: “How much do your friends really care about you?” “How much do they understand your feelings?” “How much can you rely on them for help if you have a serious problem?” and “How much can you open up to them if you need to talk about your worries?” Response options were: 1=not at all, 2=a little, 3=some, and 4=a lot [37,38]. The Cronbach α was 0.90.

Negative interactions were measured using a 4-item, 4-point scale, with total scores ranging from 4 to 16, where higher scores indicated more negative interactions. The items were: “How often do your friends make too many demands on you?” “How often do they criticize you?” “How often do they let you down when you are counting on them?” and “How often do they get on your nerves?” Response options were: 1=never, 2=rarely, 3=sometimes, and 4=often [37,38]. The Cronbach α was 0.84.

The control variables used in this study were sex (1=female and 0=male), age, marital status (1=married, 0=single, divorced, or widowed), education level (with 3 categories: university or graduate school; junior college, technical college, or specialized training school; and high school or middle school), last year’s household income (log-transformed for analysis), employment conditions (regular employment, nonstandard employment, and self-employment), presence of chronic illness (1=no and 0=yes), and residential area (1=urban and 0=rural).

This study was approved by the ethics committees of the Kyoto University of Advanced Science (KUAS 20‐3) and the National Institutes of Biomedical Innovation, Health, and Nutrition (NIBIOHN 202).

Informed consent was obtained on the research company Cross Marketing Inc’s website before participants completed the survey. Participants were presented with an information sheet on the web-based survey landing page, and only those who confirmed that they had read the information sheet and agreed to participate were included in the study.

Before transferring the data to researchers, the research company removed participants’ names, addresses, contact information, and any other details that could potentially be used to identify individuals.

Subjects registered with the research company could receive point rewards for participating in the survey (Cross Marketing Inc does not publicly disclose the number of points respondents receive for completing surveys).

A logistic regression model was used to evaluate the relationship between economic deterioration, social support, negative interactions, and mental health. Control variables included sex, age, marital status, education level, last year’s household income, employment conditions, presence of chronic illness, and residential area. In addition, interaction terms between economic deterioration and social support, as well as between economic deterioration and negative interactions, were added sequentially. A final model with a 3-way interaction term among economic deterioration, social support, and the female dummy variable was also examined. Statistical analyses were conducted using R (version 4.2.2; R Foundation for Statistical Computing). Listwise deletion was applied to data with missing values, yielding a final sample size of 1238 individuals.

The primary finding of this study is that the negative impact of economic deterioration due to COVID-19 on mental health is amplified when social support is low, supporting hypothesis 1.

Previous studies have established that economic deterioration due to COVID-19 and social support affect mental health [2,7,14-17,30], but have examined the main effects of each variable. These analyses do not identify protective factors that mitigate the negative impact of economic deterioration. By contrast, this study suggests that social support functions as a protective factor. While social support has been widely recognized for its protective effects across various contexts [25-29], this study further confirms its role in mitigating the impact of economic deterioration during the pandemic.

The main effect of negative interactions was statistically significant in all models. Unlike previous studies [19,20], this study also analyzed the interaction between economic deterioration and negative interactions. However, this interaction term was not statistically significant, indicating that negative interactions do not exacerbate the negative impact of economic deterioration. Thus, hypothesis 2 was rejected. Nevertheless, the main effect of negative interactions remained consistently significant, aligning with previous studies [19,20] on the impact of negative interactions during the pandemic. Even before the pandemic, the influence of negative interactions was considered more significant than social support [31]. This suggests that the impact of negative interactions on mental health is consistently strong and should be recognized as a potent stressor, irrespective of the pandemic context.

In addition, building on previous studies that reported the economic and psychological vulnerability of female workers in Japan during the second wave of the COVID-19 pandemic [8,32,33], this study focused on women during this period. Specifically, drawing on the conservation of resources theory [21-23], hypothesis 3 was proposed to examine whether women with greater social support were better able to cope with this stressful event and benefit from its buffering effects. However, the 3-way interaction term required to test this hypothesis was not statistically significant, leading to its rejection.

By contrast, the persistent significance of the interaction between economic deterioration and social support suggests that the buffering effect of social support was effective regardless of gender. This result likely reflects the broader reality that, during this period, both men and women faced economic difficulties and social pressures, emphasizing the importance of social support for the mental health of both genders. In Japan, traditional gender roles, which position men as breadwinners and women as homemakers or caregivers, remain deeply ingrained [39]. These norms have long linked economic difficulties to suicide among working men in Japan, even before the outbreak [40,41]. During the pandemic, these issues continued to contribute to suicide rates [18], indicating that men faced significant pressures from both economic burdens and societal expectations tied to gender roles. Consequently, while women’s mental health deteriorated significantly during the second wave [8,32,33], men also faced the risk of losing critical resources, such as income. Therefore, social support likely served as an essential buffer for maintaining the mental health of both men and women.

This study has several limitations. First, its cross-sectional design restricted the ability to establish causal relationships. Individuals with preexisting poor mental health may struggle with work and interpersonal relationships, increasing their susceptibility to economic difficulties. As such, a bidirectional relationship, rather than a unidirectional causal link, should be considered.

Second, although the sample was drawn based on age and gender group allocations, it was limited to individuals willing to participate in a web-based survey, which may have introduced sampling bias. It is likely that individuals with severely poor health, less motivation to participate in web-based surveys, or heavy workloads and limited free time were underrepresented.

Third, this analysis excluded individuals who were employed before the pandemic but unemployed at the time of the survey, due to the small number of such individuals (n=17), which made meaningful analysis difficult. Therefore, people who lost their jobs due to the pandemic were not considered. Possible reasons for this exclusion include a lack of time or resources among the unemployed to participate or limited access to the internet. This study focused on individuals employed both before the pandemic and at the time of the survey, allowing it to demonstrate the severity of the mental health challenges faced by workers experiencing economic hardship.

No previous studies have specifically addressed the factors that mitigate or amplify the negative impact of COVID-19-induced economic deterioration on mental health or explored gender differences in these effects. This paper analyzed data from a Japanese survey conducted during the second wave of the pandemic, a time when women’s mental health was believed to be in decline. The results revealed that the impact of economic deterioration was more pronounced when social support was limited, regardless of gender, and that negative interactions consistently showed significant main effects. These results suggest the ongoing need for targeted support for individuals with limited social support during societal crises such as a pandemic, irrespective of gender.