Gender differences in health protective behaviours and its implications for COVID-19 pandemic in Taiwan: a population-based study

Questionnaire was developed based on existing questionnaire used in studies on risk perception and precautionary behaviors of the general public during outbreaks of SARS and Avian Influenza [21, 22]. The questionnaire was based on an integrated model to explain health behaviors, including constructs from the Protection Motivation Theory [23] and the Health Belief Model [24]. Risk perception is specified as a combination of perceived severity (a person’s belief on how serious contracting the illness would be for him/her) and perceived vulnerability (a person’s perception of the chance that he/she will contract the disease) [24]. The Protection Motivation Theory includes another two key constructs, namely response efficacy (a person’s belief in the effectiveness of the preventive measure) and self-efficacy (a person’s level of confidence in his/her ability to perform the preventive measure). Therefore, participants were asked about preventive measures against the new influenza to measure their health protective behaviors.

The questionnaire went through meetings formed by experts, cognitive interviews, and multiple pretests. All pretests employed the same sampling (stratified three-stage probability proportional to size sampling) method and involved 300 participants. The quality of questionnaires was assessed by face and content validity, length, and comprehensibility. Cronbach’s α was used to assess the internal reliability of the questionnaire. The last internal consistency of pre-tests had a coefficient α of 0.96.

All items of the questionnaire (Table 1) were rated on a 5-point Likert scale, ranging from 1 (definitely no) to 5 (definitely yes). The total sample size was 2,005, with a response rate of 53%. This study included 1,990 respondents, with no missing data for any of the study variables. Written informed consent was obtained from each respondent. The ethics committees/institutional review boards of the Academia Sinica, Taiwan approved this study and the consent procedure.

To compare participants’ characteristics by gender, we used chi-square test (Fisher’s exact test when appropriate) for categorical variables and two tailed t test for continuous variables. The preliminary internal structure of questionnaire was explored by using exploratory factor analysis (EFA), a statistical technique to detect common factors of multiple items [25]. The EFA was done to determine the construct validity of the attitude and practice domains of the questionnaire due to their ordinal responses [26]. Bartlett’s test of sphericity and the Kaiser-Meyer-Olkin measure of sampling adequacy were used to examine whether the correlations of nine items were suitable for exploratory factor analysis. EFA with oblimax rotation was executed, and numbers of factors were decided according to eigenvalues (> 1) and Cattell’s scree test [25].

Variables were then grouped into different dimensions according to the factors found through analysis. Each dimension was treated as a dependent variable for multivariate analysis to examine the difference between men and women. Different adjusted models of socioeconomic demographic factors were derived to examine the interaction effect between these dimensions and health behaviors (i.e., marital status, living condition such as living with parents or/and children, self-related health condition including status of happiness, satisfaction of life, physical health, and education level). Gender stratified analysis was also performed separately with socioeconomic demographic factors adjusted. All model-based results are presented with 95% confidence intervals.

This study highlights gender disparities in demographic characteristics, health status, as well as adherence to risk perception, and health protective behaviors in a national sample from Taiwan.

The first finding, is that women exhibited a higher perception the of pandemic risk than men. Women were also more likely to adopt health protective behaviors, such as wearing face masks, washing hands, avoiding public places, and practicing sanitization compared to men. Our results are consistent with previous studies reporting that gender influences health behaviors, with women exhibiting higher tendency to adopt new health protective behaviors [28, 29]. Studies from Hong Kong reported that women declared a higher frequency of face mask wearing than men in compulsory situations, such as visiting clinics during flu seasons, and when presenting respiratory symptoms [30, 31]. Published reports during the COVID-19 pandemic also found that age, income, education, and especially gender affect mask-wearing behavior [32‐34]. Women wore masks more often than men, [33] probably because masks were perceived as a sign of weakness among some men, as suggested by previous work in the United States [35]. Our findings indicate that women displayed higher level of risk perception and knowledge of preventive measures than men [34]. Women may be more likely to protect themselves and others by wearing a mask because they handle the majority of caregiving within families [36], or because of awareness of the preexisting gender inequalities in access to health care that have been further amplified due to the pandemic [37].

The second finding is that, surprisingly, despite being more likely to adopt health protective behaviors, women, and individual living with children (who were equally distributed across genders) exhibited increased hesitancy to receive vaccination. This observation is consistent with the previous observation that women are more likely to exhibit COVID-19 vaccine hesitancy compared to men [38, 39]. One among predominant concerns of vaccine hesitancy found among women having children was around safety of vaccines [40] Another possible explanation is that women in our study had lower household income than men. Thus, the vaccine acceptability may be lower among these women, if they had perceived the vaccination as an additional expense [41, 42]. Further, women reported lower intention to disclose contact-tracing and to practice home quarantine than men. This may be explained by women’s fear of being discriminated towards infectious disease, which could prevent them from seeking help and medical care [43].

In subgroup analysis, we observed that living with household members was a predictor of health protective behaviors in men only. However, the positive relationship between living with household members and intention to receive vaccination was observed in both men and women. Respondents bearing family caring responsibilities may explicitly link their fears of the pandemic and their obligations to their household members, and in turn show greater adherence to health protective behaviors [7].

Our findings may contribute to public health strategies in several ways. First, our research highlights characteristics that may predict compliance to health preventive measures, allowing risk communications to be targeted. In men, interventions should focus on increasing mask-wearing behavior and the importance of adopting nonpharmaceutical preventive measures to protect family members [44]. In women, individual with children, individuals with lower education level, and individuals who are single or living alone, strategies should focus on decreasing vaccine hesitancy. Finally, women, the elderly, and those living alone, should benefit from education on the importance of contact-tracing and home quarantine that have proven efficient in controlling COVID-19 transmission [45].

The findings of this study should be considered in the context of certain limitations. First, the study population only included participants from Taiwan, and the findings may not be generalizable to other populations. However, the three stage random sampling procedures, face-to-face interviews and validation of TSCS provided a valuable insight into the whole population residing in Taiwan [46] s, the measurement of household members relied on participants’ characteristics and the questionnaire did not measure the relationship between gender and family, such as by asking men and women separately about whether they felt worried about the effect of pandemic on their household members. Further, this study measured the gender difference using a correlation analysis rather than causal analysis. Thus, we could not obtain contributing factors behind the gender difference. Finally, this study received only more than half responses among the population surveyed (53%), which might limit the generalization of the results. For the past decades, the response rates for survey research in many countries have been declining [47]. Besides, Taiwan Social Change Survey has estimated that about 20–25% of the non-response rates were due to the discrepancy of registered addresses and actual living addresses [48]. Studies have suggested through updating addresses from other supplemental sampling frames, such as postal addresses or other commercially available databases, might remedy the problem of low-response rate in a household survey [49].

Our study reveals gender differences in health protective behaviors and vaccine hesitancy. An appreciation of how socio-economic background and gender are influencing health protective behavior could have important implications for public health management and mitigation strategies for future viral pandemics.