Socioeconomic disparities in risk perceptions and precautionary actions against COVID-19 among the working age population aged 18–59 in Japan: a cross-sectional study

Introduction

During the global COVID-19 pandemic, risk perceptions and precautionary actions have been reported to be generally insufficient despite relevant efforts to raise awareness among the general public. A body of literature exists that has examined risk perceptions and precautionary actions against COVID-19. In response to major concerns about social and health gaps following the COVID-19 pandemic, there are substantial studies that assessed the association of socioeconomic position with risk perceptions and precautionary actions, as well as COVID-19 infection and death. In general, existing evidence demonstrates disproportionate patterns of COVID-19 infection and mortality by socioeconomic position, in terms of age, gender, race/ethnicity, education and income.1–11 Considering such disproportionate impacts in terms of both disease and socioeconomic burdens that continue even after the pandemic phase globally, COVID-19-related research on risk perceptions and precautionary actions from socioeconomic perspectives remains an important research agenda.

Evidence of the association of socioeconomic position with risk perceptions and preventions against COVID-19 shows some commonalities, yet mixed results are also reported. Several studies show that being male and having a lower income are significantly and negatively associated with risk perceptions and precautionary actions globally.12–17 For example, women show higher risk perceptions and precautionary actions relative to men. Income is related to various directions of perceptions in the pandemic, such that income is positively associated with precautionary actions such as mask wearing, and negatively related to anxiety about COVID-19 in terms of the disease itself and its socioeconomic consequences.14 17 18 Further, risk perceptions among younger generations are reported to be mixed across the existing evidence.16 19–21

Thus, the association of socioeconomic position with risk perceptions and precautionary actions does not seem to be consistent, underscoring variations across studies and specific measures. Several scholars employ a series of various risk perception measures in their studies, in which the definition, operationalisation and measures of risk perception show some differences, and consequently, the conclusion is not consistent. Besides the majority of the relevant literature that focuses on a specific group of population such as healthcare workers, high-risk groups and students,22–25 there are fewer studies that target the general population and employ several risk perception measures.26–28 These studies have underscored the critical variations and inconsistencies across the different measures related to risk perception. The existing literature though did not focus on and discuss the outstanding distinction between the perceived ‘subjective risk’ including fear of COVID-19 infection, and the perceived ‘objective risk’ including perceived likelihood of being infected with COVID-19 among the general population, as well as the relevant influential factors. The latter ‘objective risk’ was assessed in some COVID-19 studies in varied contexts, in which different definitions and measures were employed to answer respective research questions.29–32

In Asia, a body of literature exists that has assessed risk perceptions and preventive behaviours. Some studies show the association of preventive behaviours with perceptions regarding COVID-19, finding socioeconomic differences among the general public. Generally, being male, younger and having lower education and income are reported to be negatively related to preventive behaviours and risk perceptions in Asia, while being female, older, having higher income and being married are associated with a greater likelihood of taking preventive behaviours.33–37 Relevant evidence from Japan exists,38–47 which generally cautions against such societal gaps, with some variations in disparity patterns across studies. For example, women and lower income individuals are significantly more likely to experience serious psychological distress during the pandemic,40 while the disproportionate pattern of preventive behaviour has been suggested that being female and older age are associated with a higher likelihood.39 42 45 46 Other demographic variables have been also assessed in some studies (eg, income, education), yet the evidence is mixed.39 42 43 45 Such disparities in the perceived likelihood of infection among general Japanese populations—that is operationalised as ‘perceived susceptibility’ from the perspectives of health behaviour theory and understood as ‘objective risk’ perceptions—have been limitedly examined or reported.43 46 The potential disparity in risk perceptions and preventive behaviours by health status has been unclear in these Japanese studies.

As such, evidence gaps still exist towards understanding and promoting COVID-19 prevention among the working age population aged 18–59 who have been affected substantially since the onset of the pandemic, in terms of socioeconomic conditions that are characterised distinctly by subgroup. Although Japanese people, and Asians in general, may seem to accept preventive measures more than others in distinct regions and countries,48 social and generational gaps in risk perceptions and preventive behaviours have been generally underscored in Japan throughout the remarkable transition of pandemic phases. However, inconsistent analytic methods and employed measures in the existing studies result in different conclusions of disparity patterns by subgroup in Japan, with an exception of age and gender patterns that women and the older are commonly reported to be more likely to take preventive behaviours as same as the global evidence.

Therefore, this study aims to examine risk perceptions and precautionary actions against COVID-19 among the working age population aged 18–59 years in Japan, including the association of socioeconomic position with ‘subjective risk’ and ‘objective risk’ perceptions. These two distinct concepts are defined and operationalised respectively in this study, while these differentiations were not critically assessed in the existing literature in Asia that mostly focuses on ‘subjective risk’ perceptions such as an individual’s fear or anxiety. Further, the association between socioeconomic position and precautionary actions is examined, comprising multiple indicators, and the difference by subpopulation group is assessed by health status. It is anticipated that the risk perceptions and precautionary actions differ by socioeconomic position; ‘subjective risk’ and ‘objective risk’ perceptions are distinct risk perception measures, thus their associations with socioeconomic position could be also different.

Materials and methods

Results

Multivariate regression results

Risk perceptions (subjective and objective risk perceptions)

Table 2 shows the multivariate regression results of risk perceptions on socioeconomic position and the self-reported health status, using OLS. In Model 1 (Fear of COVID-19 Scale: FCV-19) of ‘subjective risk perceptions’, risk perceptions are negatively related to being male and unmarried. The higher the self-reported health status, the lower the risk perception. Respondents in their 20s report higher risk perceptions relative to those aged 30–39 at borderline significance. The lowest income quintile shows a positive relationship with risk perceptions, while education, paid employment and household composition do not show a statistically significant association.

Table 2

View inline|View popup

Multivariate ordinary least squares (OLS) regression analyses of subjective and objective risk perceptions (n=2764)

In Model 2 (Likelihood that COVID-19 reaches own community) of ‘objective risk perceptions’, being male is negatively associated with these perceptions, consistent with subjective risk perceptions. Respondents in their 50s are less likely to perceive objective risks than those in their 30s at borderline significance (p<0.10); however, there is no systematic relationship between age and objective risk perceptions. Self-reported health is also negatively related to risk perceptions. Low-income and middle-income quintiles are associated with lower risk perceptions relative to the highest income quintile, with the former reporting approximately 6–9 percentage points lower scores than the latter. However, education, employment, marital status and household composition are not statistically significantly related to risk perceptions.

In Model 3 (Likelihood to encounter somebody infected with COVID-19) of ‘objective risk perceptions’, being male and not being in paid employment, as well as higher self-reported health, are negatively related to these perceptions. For example, men report lower objective risk perceptions by more than 3 percentage points than women, consistent in magnitude with the other objective risk perception measures. Relative to the highest income quintile, the middle-income quintile reports lower objective risk perceptions. Compared with those aged 30–39, those aged 50–59 are also negatively associated with objective risk perceptions. The remaining variables including education, marital status or household composition are not statistically significantly associated with the variable of our interest in this model.

Precautionary actions

Tables 3 and 4 show the multivariate regression results of precautionary actions on socioeconomic position and self-reported health, using OLS and logistic regression. In Model 4 (precautionary action summative scale), men and lower education and income (particularly the second lowest group) are related to the lower precautionary action scores. For example, men show a lower score by approximately three points compared with women. Household composition is also significantly associated, showing that those who live with elderly people report a higher precautionary action score by two points compared with those who live with adults/children but not with elderly people. The better the self-reported health status, the higher the precautionary action score. Being married shows a positive association at borderline significance, while age and employment do not show a statistically significant difference.

Table 3

View inline|View popup

Multivariate ordinary least squares (OLS) regression analyses of precautionary action summative score (n=2764)

Table 4

View inline|View popup

Multivariate logistic regression analysis of key precautionary actions (n=2764)

Further, results of the specific key preventive behaviours are also shown in table 4, figure 1 and online supplemental figures 1 and 2 in terms of predictive probabilities evaluated for the observed data. In Model 5 (wearing a mask), men and people with lower income groups, such as the lowest, lower, and middle-income groups, are less likely to wear a mask, compared with women and the highest income group. Specifically, the odds of strongly agreeing to wear a mask among men are 52.6% lower than that among women (Robust OR=0.474). Besides, those who live alone are also less likely to wear a mask compared with those who live with adults/children but no elderly people. Better self-reported health is positively associated with mask-wearing. No statistically significant difference is shown by age, education, employment, or marital status.

Request permission

Figure 1

Predictive probability of the key precautionary action by age and gender. (A) mask wearing; (B) avoid large gatherings of people; and (C) wash hands more often. Blue line=male, red line=female, and black line=the both genders combined. The interval shows the 95% CIs.

In Model 6 (avoid large gatherings of people), men, teenagers aged 18–19, people in paid employment and those with relatively lower income compared with the highest income group are less likely to avoid large gatherings. In particular, the odds of strongly agreeing to avoid large gatherings among men are 45.2% lower than that among women (Robust OR=0.548). The better the self-reported health status, the more likely people are to avoid large gatherings. There is no such statistically significant relationship with education, household composition, or marital status.

Last, in Model 7 (wash hands more often), men and people with relatively lower income, except the lowest income group, are also less likely to wash hands more often. For example, the odds of strongly agreeing to wash hands more often among men are 38.2% lower than that among women (Robust OR=0.628). Those living with elderly people and the better health status group are more likely to wash hands. No such statistically significant relationship is shown with age, employment, education, and marital status, while living alone is negatively related to hand washing with borderline significance.

Discussion

This study has examined subjective and objective risk perceptions and precautionary actions against COVID-19 among the general working age population aged 18–59 in Japan, focusing on the socioeconomic position reemphasised as a key determinant for health and social welfare following the COVID-19 pandemic. This study has underscored the distinction between subjective and objective risk perceptions, which have been rarely differentiated or investigated in the literature despite their outstanding conceptual distinction and influential socioeconomic factors, respectively. The results from this study confirm the statistically significant relationship of risk perceptions and precautionary actions with several of these key socioeconomic determinants, including gender, income, employment, and household composition. Yet, this Japanese evidence also demonstrates unexpected results, especially among specific subgroups, namely, the younger generation and worse health status groups. This suggests the complex pathways leading to risk perceptions, precautionary actions, and prevention of COVID-19 in Japan and possibly elsewhere among the concerned subpopulation groups, which are not sufficiently articulated in the health behaviour theory, and not critically tested or compared in the existing literature. These disparities also suggest that the government’s COVID-19 prevention and control strategy was likely to be received and transformed into action differently by subgroup in Japan, although the Japanese government’s policy has been recognised as relatively strict against COVID-19 and other infectious diseases as well as other East-Asian countries including China and Korea, after lessons from previous pandemics.56–64

First, this study has shown partially consistent results with the existing evidence in terms of generational differences in COVID-19 prevention, highlighting the higher subjective risk perceptions but relatively limited precautionary actions among the young generation. This result is not consistent with previous studies that showed relatively lower perceived risk of COVID-19 infection among younger generations.15 16 19 Further, it should be noted that teenagers’ behaviours (age 18–19) have shown variations across the key precautionary actions in this study, in particular, less avoidance of large gatherings of people. Existing Japanese evidence of COVID-19 and other infectious disease prevention also generally suggests generational patterns of preventive behaviours with positive relationships – that is, the older the more likely to take preventive behaviours.39 42 45 46 65 This is generally consistent with the existing evidence base of an infectious disease control study showing low mortality risk perceptions, and hence, limited preventive actions among younger generations,66 underscoring the need for further investigation for effective, targeted intervention approaches to encourage precautionary actions for this particular group.

Second, the results from this study have confirmed significant gaps in risk perceptions and precautionary actions by gender and income, consistent with the existing evidence in Asia and elsewhere.1 33–37 Regarding gender gaps, Japanese women perceive higher risks and take more precautionary actions, and their infection rates have been reported to be relatively lower than men at several points in time.67 Further, income gaps are likely to explain and/or influence the gaps in risk perceptions and precautionary actions in Japan, which has not been consistently reported from the existing Japanese studies.39 42–46 The pattern of lower precautionary actions among the lowest income group people would require further investigation, whose higher subjective risk perceptions and lower precautionary actions do not align, as same as the pattern observed among the young generation. In addition, the influence of household member compositions, particularly residence with elderly people, should also receive adequate attention after this evidence, given the suspected, disproportionate socioeconomic consequences of the pandemic for these subgroups and better infectious disease controls through family infection.

Third, evidence from this study cautions against the consistency among subjective and objective risk perceptions and precautionary actions, which may explain the mixed evidence across existing studies as aforementioned. Subjective and objective perceptions do not always align as observed among the young generation and the lowest income group, despite the fact that these groups may be more likely to be exposed to personal contacts in the public and workplace in general (thus being generally recognised as a higher risk group). In consideration of this Japanese evidence and previous literature pointing to the important contrasts between these two risk perception measures in Europe and the USA,30–32 as well as other papers underscoring variations of risk perception measures,26–28 the intuition of their differences, implications, and determinants should be examined further. Investigation is warranted to understand pathways leading to COVID-19-related perception and prevention, in consideration of the multidimensionality of COVID-19-related perceptions.

Fourth, it is particularly concerning that precautionary actions among the worse health population are generally less likely, although their risk perceptions are generally higher. This is contrary to the anticipated pathway that the worse health group perceives higher risk and then takes more precautionary actions as shown in the literature,23 24 and neither pattern was reported from the relevant Japanese studies several of which did not examine this association.39 42–46 This pattern of disconnect and the unexpected opposite direction of the relationship are similar to those among the younger generation in this study population. Among the young generations though, the literature has generally pointed to their risk-taking behaviours and risk preference, which can have adverse consequences for general health (eg, smoking, alcohol, drug use, etc.)68–71 as well as COVID-19.72 Such health risk assessment among young generations may also possibly apply to infectious disease prevention. This concerning pathway in these population groups needs to be further investigated in future analyses.

This study entails several limitations. First, this is a cross-sectional survey; thus, it is not possible to make causal inference on the hypothesised pathway. Second, although the study sample was randomly drawn from the panel of one of the largest online survey companies in Japan, this study was an online survey that targeted towards the selected geographic areas at higher risk of COVID-19 infection to answer this research question in the high-risk context. This was because of the Japanese situation where COVID-19 infection rates varied substantially by prefecture though, consequently, this evidence is not nationally representative. Third, there may have been more explanatory variables that were not observed in this first data collection, while a subsequent survey collected additional variables on socioeconomic position, COVID-19 vaccination and relevant perceptions. Thus, analysis and discussion of COVID-19 vaccination are not a scope of this paper.

Despite these limitations, this is one of the first studies in Japan and Asia that has examined subjective and objective risk perceptions and precautionary actions employing international tools used for COVID-19 and other infectious diseases. Evidence from this study shows a unique case from Japan, being partially consistent with the existing evidence and cautioning against the general expectation of consistency among subjective and objective risk perceptions and precautionary actions in the expected traditional pathway by the health behaviour theory. Specifically, this evidence underscores such uniqueness and complexity among the high-risk and higher risk-taking groups, particularly the young generation and the worse health group, as well as the lowest income group, whose health-related perceptions and behaviours are likely to be influenced by several other factors than risk perceptions. These disparities reflect gaps in perceptional and behavioural responses among the Japanese general public, despite the fact the government implemented relatively strict measures in the early phases of the pandemic, followed by subsequent multiple pandemic phases with remarkably high infection cases in 2022. Distinctions and relevant intuitions about several risk perceptions (eg, subjective vs objective risk, perceived risk of infection vs mortality) seem to have a potential to shed light on the deeper understanding of the key influencing factors and mechanisms surrounding the prevention of COVID-19 and other infectious diseases. Further attention and research would be needed to better understand the complex pathways surrounding infectious disease prevention, as well as the social gaps and disproportionate impacts of the relevant pandemic, in an effort to address health equity concerns globally.