Introduction

In response to the spread of coronavirus disease (COVID-19), measures to prevent infection, such as avoidance of the three Cs (3Cs, that is, areas with poor ventilation, crowded areas, and settings where close-range conversations may occur) have been promoted in Japan and have become widely known (Hayasaki, 2020). After lifting the state of emergency at the end of May 2020, the country entered a transitional period in which restrictions on traveling and the use of facilities were gradually alleviated. The continued implementation of the aforementioned measures to prevent infectious diseases has become important in preventing the re-occurrence of infection. In other words, the degree to which each citizen thoroughly implements preventive behaviors against infectious diseases will greatly influence the future progression of the COVID-19 pandemic (Amsalem et al., 2021; Yonemitsu et al., 2020). In Japan, the Ministry of Health, Labour and Welfare of Japan (2020a, b) has proposed “The New Lifestyle,” which includes the following prevention behaviors against COVID-19 that individuals can practice in daily life, such as mask-wearing, handwashing, and social distancing. The habitual practice of these infection-preventive behaviors is one of the most important measures for preventing the spread of infection and is being used as reminders and recommendations.

Such infection-prevention behaviors (IPBs) against COVID-19 are considered one of the main topics of studies on COVID-19 in the field of psychology. Brouard et al. (2020) suggested that psychological factors, such as anxiety about COVID-19 and perceived political ideology, promote IPBs, such as mask-wearing. In Japan, Yamagata et al. (2021) found that during the COVID-19 pandemic, there are more opportunities to routinely take IPBs different from normal conditions, and that such an increase in IPBs was associated not only with anxiety and risk perception toward COVID-19 but also with the fear of and aversion to physical or spiritual impurity with a gender difference. Furthermore, a review and meta-analysis of studies on the current state of COVID-19 related IPBs in various countries indicated that knowledge about COVID-19 promotes COVID-19 related IPBs, while fatigue and illness, anxiety, and stress, might inhibit COVID-19 related IPBs (Saadatjoo et al., 2021). Thus, identifying the psychosocial variables associated with IPBs, similar to the abovementioned studies, is a notable attempt at providing important insights when considering measures for promoting COVID-19 related IPBs. One of the important factors in determining an individual’s health-related behavior, including infection-prevention behavior, is psychological characteristics or personality traits. For example, previous scholars found a relationship between the “Big Five” personality traits and COVID-19 related IPBs among the general population; conscientiousness positively related to individual and social behaviors related to health and infection prevention, whereas extraversion negatively related to IPBs (e.g., Brouard et al., 2020; Carvalho et al., 2020). Other scholars report that conscientiousness is positively related to compliance with COVID-19 protection protocols, whereas extraversion is negatively related among the general population (Brouard et al., 2020). Thus, predicting COVID-19 related IPBs from the perspective of personality traits would render the identification of individuals and groups at high or low risk of COVID-19 possible. Therefore, examining the relationship between psychological characteristics and IPBs, specifically in Japan, is important.

The sense of coherence (SOC) is defined as an individual’s perception and the sense that their experiences in the world are coherent, comprehensible, consistent, and meaningful; those with high levels of SOC can cope effectively and flexibly with stressors in the theory of salutogenesis with SOC as a core concept (Antonovsky, 1987). SOC consists of three components: comprehensibility, manageability, and meaningfulness (Antonovsky, 1987). Comprehensibility means that the stimuli encountered in the future will be predictable (Antonovsky, 1987). Manageability refers to the extent to which a person perceives that their resources are adequate to meet the demands (Antonovsky, 1987). Meaningfulness refers to the extent that stimuli are considered worth investing energy in, worthy of commitment, and challenges rather than burdens (Antonovsky, 1987). Moreover, SOC is a strong predictor of mental health or distress among personality traits (Flensborg-Madsen et al., 2005; Grevenstein & Bluemke, 2015) and is associated with the maintenance of mental health even under difficult circumstances (Hochwälder & Forsell, 2011). Simultaneously, it is one of the psychological characteristics reported to be associated with health-related behaviors. For example, individuals with high levels of SOC were more likely to engage in daily physical activities among students and adults (Kuuppelomäki & Utriainen, 2003; Suominen et al., 2005), brush their teeth frequently to maintain oral care conditions among the adults and older adults (Bernabé et al., 2012; Savolainen et al., 2005), and experience less alcohol-related problems among general population (Midanik et al., 1992; Antonovsky, 1987), who proposed the concept of SOC, explained that individuals with high levels of SOC do not simply prefer health-related behaviors; instead, they opt for health-related behaviors as the most effective coping resource (general resistance resources) because of an accurate understanding of their health problems. Given these findings, the study proposes that SOC would be negatively associated with distress during the COVID-19 pandemic. Furthermore, individuals with high levels of SOC would be more likely to engage in COVID-19 related IPBs, one of the coping resources for health problems during the COVID-19 pandemic.

Furthermore, resilience is a factor with a theoretical similarity to SOC. Resilience is defined as a psychological trait enabling adaptation despite challenging or threatening circumstances that help people recover from distressFootnote 1 (Masten et al., 1990; Oshio et al., 2002). Resilience has been conceptualized and assessed somewhat differently in different studies. According to the definition by Oshio et al. (2002, 2003), those with high resilience are characterized by a high preference for new challenges, emotional control, and a positive outlook on the future. These elements are considered to constitute resilience as factors referred to as “novelty seeking,” “emotional regulation,” and “positive future orientation” (Oshio et al., 2002). The presence of the trait helps relieve the stress associated with negative events and promotes normal adaptive behavior, which, thus, has been associated with the prevention of distress (Fergus & Zimmerman, 2005; Kukihara et al., 2014; Oshio et al., 2002). Such an orientation is included as comprehensibility in SOC (Antonovsky, 1987) and novelty seeking in resilience (Oshio et al., 2003) and is considered a characteristic that positively affects stress coping. Quantitative studies with the general population have shown that SOC and resilience have overlapping components, while, simultaneously, the two traits feature relatively different associations with external indicators (e.g., physical and distress, strategies for coping with stress, and well-being), where SOC was reported to exhibit better predictive utility, especially regarding health-related indicators (Grevenstein et al., 2016; Izydorczyk et al., 2019; Yoneda et al., 2018). The reason for these differences is that SOC includes psychological traits that promote perceptions and appraisals such as meaningfulness and manageability, which are motivations to find meaning in dealing with specific issues and actively engage in them, and thus may be more strongly related to health-related behaviors than it is to resilience, which focuses on adaptation to the current situation. Because of their conceptual similarity and because they are both characteristics associated with health-related behaviors, they are expected to be associated with COVID-19 related IPBs; however, due to the aforementioned conceptual differences, their associations with external indicators may be different. Accordingly, both factors should be considered in the context of their relationship with distress and COVID-19 related IPBs, and their similarities and differences should be elucidated.

The extant research highlights the importance of predicting individual prevention behaviors against infectious diseases, where resilience and SOC are candidate predictors. Because SOC and resilience include orientations that can be changed by social and environmental factors, they are both considered acquired and transformable traits, although relatively stable (Delbar & Benour, 2001; Ueno et al., 2019); thus, scholars have investigated interventions to improve resilience and SOC (e.g., Ando et al., 2011; Robertson et al., 2015). Therefore, if associations exist between SOC, resilience, and COVID-19 related IPBs then targeted interventions could be developed to address these psychological factors and thereby promote infection-prevention behavior while maintaining and improving distress. Accordingly, this study investigates the relationship between SOC, resilience, and COVID-19 related IPBs. In addition to these major factors, the study assesses the several control variables potentially related to COVID-19 related IPBs and psychological factors such as social attributes (gender, age, and residential area) and perceived vulnerability to disease (Duncan et al., 2009). Against this background, this study presents the following hypotheses:

  • H1: SOC and resilience can be used to predict lower distress during the COVID-19 pandemic.

  • H2: SOC and resilience are positively associated with COVID-19 related IPBs.

Methods

Participants

An online survey was conducted in June 2020. A total of 1,900 Japanese men and women (aged in their 30 to 50 s and members of the iBRIDGE Ltd. research registry) participated anonymously in the survey through Freeasy, a web research service provided by iBRIDGE Ltd. Data from 1,484 participants (male: 686; female: 798; mean age: 45.1 years, SD = 8.3 years) who correctly answered the Directed Questions Scale (DQS; Maniaci & Rogge 2014) were analyzed. The DQS was added to these questions to detect and screen out the participants who are not paying sufficient attention to answer the questions. Specifically, the item “Select ‘probably no’ for this item” was added to the questionnaire. Using DQS, participants who exhibit a satisfactory answering style in which a respondent does not intend to allocate attentional resources to the investigation, such as those who answer without carefully reading the question, were excluded (Maniaci & Rogge, 2014). Of the analyzed, 814 were university graduates. Additionally, there were 522 living in the specific alert prefectures.Footnote 2

Ethical considerations

Written informed consent was obtained from the participants through an online survey. Data collection was conducted anonymously. The participants were assured that non-participation would not lead to any disadvantages. These ethical considerations were explained to potential participants via a text explanation at the interface of the survey. This study was conducted with approval from the ethics committee at the previous affiliation of the first author (approval number: 20–29).

Measurements

Measurement of psychological variables

SOC was measured using the Leipzig Short Scale of SOC (SOC-L9; Lin et al., 2019; Japanese version: Kase & Endo, 2020; Togari et al., 2015). SOC-L9 is composed of nine items (e.g., Do you have the feeling that you are in an unfamiliar situation and don’t know what to do?), which are rated using a seven-point Likert-type scale (e.g., 1 = very often; 7 = very seldom or never). High scores of SOC-L9 indicate high SOC levels. The SOC-L9 has been verified to have high validity, as its item fit and association with criterion-related indicators such as distress, anxiety, and depression have been verified (Lin et al., 2019; Kase & Endo, 2020).

Resilience was measured using the Adolescent Resilience Scale (Oshio et al., 2002, 2003). It has been reported that ARS is reliable and valid in quantitative studies on adults (Ueno et al., 2019). ARS is composed of 21 items (e.g., I am sure that good things will happen in the future), which are rated using a five-point Likert-type scale. High ARS scores indicate high levels of resilience. Although the ARS was developed based on data obtained from adolescents, its measurement follows the general definition of resilience (Nancy et al., 2006), and because it is a scale whose reliability and validity have been confirmed in multiple studies with college students (e.g., Nakaya et al., 2006; Oshio et al., 2003), we considered it suitable for use in this study.

As an outcome, distress was measured using the Kessler Psychological Distress Scale (K10; Kessler et al., 2002; Japanese version: Furukawa et al., 2008). K10 has ten items (e.g., During the past four weeks, how often did you feel tired out for no good reason?), which are rated using a five-point Likert-type scale. Although K10 has a cutoff value for determining the presence or absence of mental disorders, whether this cutoff value functions relatively well under other conditions is unclear because this study was conducted under the specific stress conditions imposed by the COVID-19 pandemic (Okubo, 2021). Therefore, we decided to use the scores for K10 as a continuous variable for analysis. High K10 scores highly indicate distress. The K10 has been associated with measures of mental health and distress, such as the University Personality Inventory and the General Health Questionnaire, and it is validated (Sakai & Noguchi, 2015).

In addition to these measurement scales, the study used the Perceived Vulnerability to Disease Scale (PVD: Duncan et al., 2009; Japanese version: Fukukawa et al., 2014) to measure psychological fear and anxiety about infectious diseases in general (as opposed to COVID-19 specifically) as the control variable. PVD considers two theoretical factors, namely, perceived infectability (Generally, I am very susceptible to colds, flu, and other infectious diseases; seven items) and germ aversion (I prefer to wash my hands soon after shaking someone’s hand; eight items). It is composed of 15 items rated using a seven-point Likert-type scale. High PVD scores indicate high fear and anxiety regarding infectious diseases. PVD has theoretically valid correlates with psychosomatic tendencies and personality traits (Fukukawa et al., 2014).

Measurement of COVID-19 related IPBs

Based mainly on “The New Lifestyle” practices presented by the Ministry of Health, Labour and Welfare (2020a, b), the study collected data on COVID-19 related IPBs for measurement concerning information made available to the public for June 2020. In May 2020, the Ministry of Health, Labour, and Welfare (2020a, b) proposed a new, unified preventive action plan for COVID-19 in Japan under the title of The New Lifestyle, and the following June, the expert panel further organized and published the information. Thus, the survey was conducted during this period because it was deemed the time at which a common understanding of preventive behavior was formed in Japan.

After repeated discussions among the researchers to ensure the absence of bias or overlap in the content, 14 types of behaviors were selected for measurement, and questions were prepared (Table 1). The survey is composed of 14 items rated using an 11-point scale from 0 (not at all applicable) to 10 (very applicable) based on the response method used by Brouard et al. (2020). In this manner, the study aimed to determine the extent to which the participants performed COVID-19 related IPBs in the previous three months. Additionally, concerning Brouard et al. (2020) and Firouzbakht et al. (2021), we used dummy variables that indicate values above or below the median value of each item (0: a lower level of practicing the COVID-19 related IPBs; 1: a higher level of practicing COVID-19 related IPBs) in the data analysis, to compare the degree of association of independent variables to levels of practicing by odds ratios.

Table 1 Questions on COVID-19 related infection prevention of behavior
Full size table

Confounding variables

Demographic variables consisted of gender (female = 1; male = 0), age, residential area (specific alert prefectures = 1, others = 0), and level of education (university graduates = 1, undergraduates = 0). For the residential area, we asked the respondents to specify their place of residence among the 47 prefectures in Japan. Thereafter, a dummy variable was used to discriminate whether the respondents lived in the 13 areas referred to as “specific alert prefectures” by the Ministry of Health, Labor and Welfare (2020a, b), where the spread of COVID-19 was significant, and special behavioral restrictions were requested in May 2020.

Data analysis

First, basic analysis was performed, such as the calculation of descriptive statistics, the reliability coefficients for each scale, and the correlation coefficients among variables. Next, multiple regression analysis was conducted using the scores for K10 as the dependent variable; scores for SOC and resilience as the independent variables; and sociodemographic variables and scores for PVD as the control variables. Afterward, logistic regression analysis was conducted using the scores for COVID-19 related IPBs, which were made into dummy variables, as the dependent variables, scores for SOC and resilience as the independent variables, and sociodemographic variables and scores for PVD as the control variables. In the series of logistic analyses, each COVID-19 related IPB item was analyzed separately. Significance levels were adjusted for the number of multiple tests per analysis. Adjusted critical p-values were p < .01 for the analysis for K10 scores and p < .004 for IPBs scores.

Statistical analyses were conducted using HAD version 17.204 (Shimizu, 2016) and R version 4.1.0 (R Development Core Team, 2022).

Results

Basic analysis

Table 2 provides the mean value, standard deviation, and reliability coefficients of each scale. For the IPBs scores, the frequency distribution was skewed in the negative direction, and based on the test of normality, it could be determined that the normality did not hold up to the parametric test; hence, it was decided to make it a dummy variable as planned. Reliability coefficients were above.70 for all scales, which indicates the absence of problems with internal consistency per scale. Additionally, a significant positive correlation coefficient was obtained between the scores for SOC-L9 and ARS (r = .75, p = .00). These had significant negative correlation coefficients with K10 or the scores for the two sub-scales of PVD (r = − .07 to − 0.68, ps = 0.00 to 0.04), respectively. Additionally, descriptive statistics of IPBs were calculated, and the distribution of dummy variables was confirmed.

Table 2 Descriptive statistics of each variable (N = 1,484)
Full size table

Relationship with distress

Hierarchical multiple regression analysis using the scores for K10 as the dependent variable indicated that R2 and ΔR2 were significant in Step 3 with all control and independent variables included (Table 3; R2 = 0.48, p = .00; ΔR2 = 0.17, p = .00). The standardized partial regression coefficient of scores from ARS to K10, which was significant in Step 2 (β = −0.46, p = 00), was no longer significant in Step 3 (β = 0.00; p = .90). Instead, a significant standardized partial regression coefficient from the score for SOC-L9 was found (β = −0.64, p = .00).

Table 3 Hierarchical multiple linear regression with K10 scores as the dependent variable (N = 1484)
Full size table

Relationship with IPBs

Logistic regression analysis using the dummy variables of the COVID-19 related IPBs as the dependent variables illustrated that resilience was significantly associated with 12 of the 14 items (OR = 1.41 to 2.21, ps = 0.00 to 0.01. see Table 4). SOC was non-significantly associated. The variance inflation factors are all less than 3.

Table 4 Odds ratios with a 95% confidential interval of the final step of logistic regression analysis for associations of relevant variables with COVID-19 related infection behaviors (N = 1,484)
Full size table

Contrary to the hypothesis, no significant association was noted between SOC to most of the COVID-19 related IPBs. Thus, we conducted exploratory analyses excluding resilience from the model predicting COVID-19 related IPBs. The result pointed to significant associations of SOC with all items, except for Q2 and Q8 (OR = 1.14 and 1.40, ps = 0.00 to 0.08; McKelvey and Zavoina Pseudo R2 = 0.05 to.17).

Discussion

General discussion

This study examined the association of SOC and resilience with distress and the COVID-19 related IPBs. First, the results of multiple regression analysis indicated that SOC predicted distress during the COVID-19 pandemic even after controlling for resilience. The results supported those of previous studies that reported the greater predictive utility of SOC than that of resilience for health-related indicators such as psychological well-being and depression tendency (e.g., Izydorczyk et al., 2019; Yoneda et al., 2018). The value of the correlation coefficient between SOC and resilience was.75 (p < .01), which indicates a high positive correlation between them. Previous studies on SOC and resilience observed that the values of correlation coefficients ranged from 0.35 (p < .01; Nygren et al., 2005) to 0.69 (p < .01; Keil et al., 2017). Many scales can be used to measure resilience (e.g., Connor & Davidson, 2003; Friborg et al., 2003; Wagnild & Young, 1993), and they can be used differently according to the theoretical definition of resilience. The current study viewed resilience as an intra-individual factor and used the ARS, which is a suitable scale of measurement for this definition. Therefore, a possibility exists that some aspects of the construct of resilience overlap with some aspects of SOC. However, the value of the incremental explained variance in Step 3 of hierarchical multiple regression analysis (ΔR2 = 0.17, p = .00) may indicate discrimination between SOC and resilience in their relationship with distress.

In contrast, resilience was positively associated with many COVID-19 related IPBs. Simply put, the study suggests that individuals with high levels of resilience tend to be more likely to report practicing COVID-19 related IPBs. For SOC, the results indicated no significant association when resilience was considered. Health-related behaviors that were associated with SOC in previous studies were mainly those that maintain and improve health as daily habits, such as physical activity (Kuuppelomäki & Utriainen, 2003; Suominen et al., 2005) and oral care (Bernabé et al., 2012; Savolainen et al., 2005). Compared to such health-related behaviors, COVID-19 related IPBs are construed as their facet as socially required by government agencies and workplaces out of consideration for others and can be regarded to possess characteristics that differ from those of habitual behavior voluntarily performed toward maintaining and improving health (Nakayauchi et al., 2021; Sakakibara & Ozono, 2021). According to the prototypes of personality, which is a classification of personality characteristics based on the Big Five personality traits, individuals classified as Resilients (those with the highest levels of resilience) are characterized by their tendency to engage in prosocial behavior (Asendorpf et al., 2001; Herzberg & Roth, 2006; Elliott et al., 2017). Furthermore, scholars suggested that resilience is positively correlated with social desirability in terms of measurement indices (Wagnild, 2009). Based on these facts, we infer that persons with high levels of resilience are more likely to practice IPBs against the COVID-19 pandemic as desirable actions required of and by society. Against these tendencies, resilience was not significantly associated with the behaviors of going out and avoiding eating out. Resilience features an aspect called novelty seeking, which refers to an interest in new events and willingness to take on new challenges (Oshio et al., 2003), and was reported to be positively correlated with the Big Five personality traits of extraversion and openness (Nakaya et al., 2006). Going out and eating out tend to satisfy novelty seeking; the study inferred that they are not associated with the avoidance of these behaviors. Alternatively, the only behavior found related to SOC was whether health information was collected through social networking services. The result indicates that those with high levels of SOC tended to be less likely to engage in such behavior. A characteristic of people with high levels of SOC is an attitude of systematically selecting and organizing information (Antonovsky, 1987; Kase et al., 2016). Thus, we inferred that they do not use SNS as an information resource on COVID-19 due to this attitude and the large amount of information that exists in an unregulated manner, which renders determining their authenticity difficult.

Limitations and directions for future research

In summary, the results suggest that SOC and resilience have been negatively associated with distress during the early phase of COVID-19 pandemic; those with higher SOC and resilience tended to have lower distress. Moreover, resilience was positively related with COVID-19 related IPBs, and individuals with higher resilience tended to engage in IPBs. In addition, there is no association with SOC and IPBs in relationship with resilience. This study only examined the relationship between psychological variables and behavioral indicators using cross-sectional data based on self-assessment; however, it does not support a causal interpretation regarding the relationships among these variables. This study also focuses on the early pandemic experience and does not provide information on changes in distress or IPBs during the COVID-19 pandemic; therefore, the results may indicate different trends from the present situation. Furthermore, in this study, the survey was conducted to coincide with the release of The New Lifestyle, a set of standards for infection-prevention behavior in Japan, and by asking survey participants to recall their behavior during the three months from March to June 2020, we could obtain a self-assessment of their average commitment to preventive behavior immediately after the outbreak. The survey was conducted to collect data regarding these three months. However, the assumption of three months may lead to a recall bias in the self-evaluation of efforts exerted toward preventive behavior, depending on factors, such as whether they had experienced infection and whether they had close relatives who had been infected. In addition, responses for the survey may have been affected by unaware changes in preventive behaviors over the course of the months. In the future, it may be possible to collect less-biased data, by having the participants describe the most recent activities, such as one week or asking them to respond to the frequency of the action using an experience sampling method. Additionally, identifying the temporal relationship between COVID-19 related IPBs and psychological characteristics using a longitudinal study that considers perceived COVID-19 risk, comorbidities, level of disruption caused by the pandemic, and changes in the state of epidemics of infectious diseases is necessary to elucidate the effects of SOC and resilience on COVID-19 related IPBs.

Furthermore, differences in the association of both factors with distress and COVID-19 related IPBs may indicate discrimination between SOC and resilience, which include similar concepts. However, additional research is required to further examine the discriminant validity of SOC and resilience, and their differential associations with other variables such as social desirability and prosocial behaviors. Thus, elaborating on statistical and theoretical similarities and discrimination between SOC and resilience will be necessary by considering prosocial behavior, social desirability, and variables considered overlapping with SOC and resilience (e.g., emotional well-being) in future studies. In this study, we used a scale for adolescents to measure resilience; however, there is also a need to use a validated scale for adults in the future to increase the interpretability of the results.