The novel coronavirus disease (COVID-19) has become not only a public health crisis but also a serious threat to social and economic development (1, 2). According to the World Bank, COVID-19 has triggered the deepest global recession in decades, led to a 5.2% contraction in the global economy, and left lasting scars on productivity (1). Within China, COVID-19 was first identified in December 2019 and declared a Chinese national emergency on January 29, 2020. As of August 2021, there were more than 93,100 confirmed cases in China (3). Evidence indicated that the control measures to reduce the spread of COVID-19 resulted in a substantial productivity loss that amounted to over US$382 billion in China (4).

In response to the COVID-19 outbreak, the government worldwide adopted a series of prevention and control measures, including early reporting and situation monitoring, large-scale contact tracing, and health communication campaigns (5, 6). International evidence from USA, Iran, UK, and China showed that people in resource-limited communities are generally more likely to have worse health behaviors during COVID-19 (7–11). In the USA, only 20% of rural participants wore a mask, compared to 47% of urban participants (7). In Iran, 45.7% of participants washed their hands regularly, and handwashing was less common among rural than urban residents (8). Additionally, one UK study indicated that better practice in mask-wearing was significantly associated with living in an urban environment (12). In particular, a study in China indicated that 72.22% of rural residents and 85.70% of urban residents wore a mask when they were outside (10). Compared with urban residents, rural residents are less aware of disease prevention and control measures because of their remote residence and poorer economic conditions (9, 11). In addition, 60% of rural residents are migrant workers in China. Their mobility could have increased rural communities' risk of infection (11). The rural area has become an important battlefield for epidemic prevention and control (13).

Evidence also indicated that gender-responsive policies are required to avoid worsening health and social inequities associated with the COVID-19 pandemic. Specifically, women on average report more days of poor physical and mental health than men despite utilizing more preventive care services (14). These health inequities are larger in women with intersecting identities, such as those living in rural areas (14, 15), potentially due to limited access to healthcare resources and inequitable gender norms that may lead to delayed or forgone healthcare and worse health outcomes (16–18). In addition, as the main caregivers in the family, women often have to spend more time on unpaid domestic work due to the lockdown measures in many countries, including China. Previous studies showed that women often make key dietary choices for their families and act as role models for their children regarding healthy behaviors (19, 20). However, despite their greater burden, lack of support, and potential of influencing other family members, few studies have looked at how to support behavioral changes to promote health and reduce disease risks among women living in rural areas of western China (21–23).

Previous studies underscored the associations between individuals' socioeconomic status, exposure to health information, psychological responses, and behavioral responses during COVID-19. A study in Germany suggests that educational background was positively associated with protective behaviors among the general public (24) while a study from Switzerland suggests the opposite association among young people (25). In Bangladesh, information on the proper use of protective measures from the government were identified as the drivers of COVID-19 protective behaviors (26). Findings from Nigeria and the USA showed that individuals who perceived greater risks of COVID-19 infection were more likely to adopt protective behaviors (27, 28). In China, a study conducted in primary schools showed that educational background of the mother was associated with handwashing and mask-wearing practices of the children (29). Another study in China indicated that perceived risk and severity of the participants were associated with excess protective behavior (30). One study indicated that perceived susceptibility to COVID-19 was associated with consistent mask-wearing among Chinese pregnant women (31). However, no studies have examined the direct and indirect associations among these variables, especially among women in rural areas of western China.

To develop the conceptual framework for our analysis, we turned to the Extended Parallel Process Model (EPPM) that examines how media exposure influences perceptions and behaviors of people in the midst of public health emergencies (32–34). The main components of EPPM include exposure to message communication, perceived efficacy, perceived risk, fear, and behaviors. According to the EPPM, when individuals are exposed to a health message, they would make a cognitive appraisal of the message, including the appraisal of risk and the appraisal of efficacy, which are two key factors that influence the health behavior in response to the health risk. The EPPM also suggests that exposure to health communication messages leads to the belief that individuals are able to reduce the risk (higher perceived efficacy), which in turn prompts individuals to adopt protective behaviors. On the other hand, if individuals do not have enough confidence in their ability to act (lower perceived efficacy), they would not adopt protective behaviors (32–34).

Informed by the evidence from the literature and the EPPM, our study examines the relations among socioeconomic status, exposure to COVID-19-related information, psychological response to COVID-19, and self-protective behaviors among women in rural western China. We hypothesized that there are direct and indirect effects among these variables.

Women in rural areas of China are disproportionally affected by COVID-19 but little is known about how they cope with the pandemic and whether they could adopt self-protective behaviors to reduce their risks of infection. This study reached a relatively neglected and vulnerable sample during the peak period of the COVID-19 outbreak, which found that not all women were able to adopt self-protective behaviors, such as mask-wearing and handwashing during the COVID-19 pandemic in western China. By examining the associations between exposure of individuals to COVID-19-related information, psychological response to COVID-19, and self-protective behaviors, we found that perceived efficacy and socioeconomic backgrounds were important factors directly associated with the adoption of self-protective behaviors during the COVID-19 pandemic among women in rural western China. Moreover, exposure to information of public health guidance was associated with self-protective behaviors indirectly via increased perceived efficacy. Such findings may help in tailoring efficacious interventions for improving the COVID-19 pandemic response among women in rural western China.

In the ongoing COVID-19 pandemic, non-pharmaceutical protective measures, such as mask-wearing and handwashing, remain crucial. In our study, most women were able to adopt these protective behaviors (80.1% in mask-wearing and 60.0% in handwashing), which is higher compared to another study in Nigeria where 37.7% of women from the rural areas reported always wearing a face mask when going out and 46.8% responded always washing hands (46), also higher than another study in African where 22.5% of women from the rural areas reported good practice of preventive measures against COVID-19 infection, including hand washing, wearing a mask, and maintain social distance (47). But our behavior results were lower than those reported in an Ecuadorian study where the local women had a higher engagement in mask-wearing (91.9%) and handwashing (96.7%) (48) and still lower than an Iranian study where the local women had a high performance in protective behaviors (97.3%) (49). We additionally found a different trend between the behavior results from ours and the international study. Our study found that mask-wearing was more common than handwashing, which differs from the findings of the above Nigeria and Ecuadorians studies (46, 48), also differs from the studies in Ethiopia and the USA. A study in Ethiopia found that half of the women (51.61%) would wear masks compared with that 90% of wash hands (50). Similarly, in the USA, handwashing (87.2%) was common than mask-wearing (23.1%) among women (51). This difference could be explained by the fact that the public in China is more open to mask-wearing due to the experience of previous outbreaks, such as the severe acute respiratory syndrome (52), early recognition of asymptomatic transmission, government's effort to increase public awareness of the importance of mask-wearing, and strict enforcement of the mask mandate (53).

We found that greater perceived efficacy in coping with COVID-19 is associated with increased self-protective behaviors, which echoes the findings from a growing number of studies that demonstrate the link between self-protective behavior and efficacy (31, 34, 54, 55). EPPM proposed that when faced with health risks, people would perceive that they themselves are able to perform recommended self-protective behaviors and that these behaviors are effective in responding to the threat. This perception would in turn lead to more self-protective behaviors. On the other hand, this result offers new insight into strategies to promote health behaviors through enhancing response efficacy. The reason may be that response efficacy was a type of action perspective to remove the risk (54, 56), individuals are more likely to adopt self-protective behaviors if they believe they can easily, feasibly, and effectively prevent a health threat with serious consequences. Greater perceived response efficacy may indicate stronger confidence that guideline-recommended measures are effective in reducing the risk of COVID-19 infection, which in turn would increase individuals' protective behaviors. More research is needed to identify effective strategies to increase perceived efficacy, especially response efficacy, to promote self-protective behaviors against COVID-19 and other infectious disease crises.

Furthermore, according to EPPM, the behavioral response is directly affected by perceived efficacy of people, which can be targeted by health messages (54, 56). We examined this potential association and found that exposure to public health guidance, particularly the government-endorsed individual precaution measures, was associated with greater perceived efficacy. Previous evidence has shown that official reports by the government and health education campaigns that promote self-protective behaviors were critical in slowing the spread of disease (57, 58). Most women in rural areas are not well-educated and have limited household assets (59, 60). Thus, it might be difficult for them to access health information from a variety of sources (39). To them, the public health guidance from the government could be the most reliable and accessible information source during the COVID-19 pandemic (61). In addition, clear and specific guidance on individual protective measures, such as how to properly and effectively wear masks and wash hands, may have enhanced individuals' confidence in engaging in self-protective behaviors. To reach a broad audience, the government should use specific and actionable messages in their health communication campaigns against COVID-19.

Although communication channel, risk perception, and fear were direct or indirect predictors of behavioral outcomes, they only had modest effects on self-protective behaviors in this study. As explained by EPPM (32, 33), response of an individual to a risk-based message involves two distinct cognitive appraisals. The first appraisal is related to the degree to which the message is perceived as threatening, i.e., perceived susceptibility and severity. If the individual perceives that they are personally vulnerable and that the risk is severe, the second appraisal occurs whereby the individual considers whether the message provides effective and useful strategies to reduce the risk. In other words, perceived susceptibility and severity alone are not sufficient and appraisal of effective actions is needed to prompt actions, as suggested by our primary finding that perceived efficacy was the main precursor of self-protective behaviors.

Consistent with previous studies conducted in China, Brazil, and the United States (62–64), women with higher SES, such as educational and economic levels, were more likely to adopt self-protective behaviors in our study. This might be because women with higher levels of SES had better access to public health information and understand better the government-recommended precautious measures (61, 62). To better promote self-protective behaviors against COVID-19 in the rural areas, health communication interventions should be tailored to accommodate the needs of individuals with low health literacy to reach women from disadvantaged backgrounds.

Taken together, the findings will shed light on direct and indirect factors contributing to the self-protective behaviors during the COVID-19 pandemic among women living rural areas of western China based on EPPM. Within a risk situation, participants may perceive high levels of risk, they may need more information to enhance their efficacy of responding to the risk. The communication message provided therefore should aim at enhancing the efficacy of women from the rural areas by strengthening the government-endorsed individual precaution measures and giving clear guidelines on how the self-protective behaviors can be undertaken, which might be increased their engagement in the self-protective behaviors. SES was also the important precursor of self-protective behaviors, those women with a lower level of SES should be paid more attention during the COVID-19 pandemic. These findings implied to us, that efforts to foster the high efficacious message and strengthen the disadvantaged populations during and in the aftermath of COVID-19 may mitigate the inequitable risks posed by pandemics and other times of healthcare stress.”

The findings should be interpreted in the context of a few limitations. First, data in the study were collected after the lockdown was over, we did not collect any data at the beginning of COVID-19, and we were unable to rule out the possibility of recall bias. However, we emphasized in our survey questions that we were asking about experience of the participants “during the lockdown” and trained enumerators to follow a standard survey manual when asking these questions. Second, we collected data only via structured interviews to measure how dynamic individual characteristics, such as exposure to health information and psychological response, affect health behaviors. Future research could incorporate qualitative approaches to develop a more in-depth understanding of the interplay between these factors and the way they affect health behaviors. Third, the analysis sample was constructed from two surveys conducted 3 months apart at the peak of COVID-19 in our study. Despite similarities between these two sub-samples in background characteristics of the participants, the time difference might have introduced biases from unmeasured co-founders. We attempted to analyze the two sub-samples separately, but the model was under-fitting. To get a larger sample size for model fitting, we thus combined the two sub-samples in the analysis. Fourth, our study sample consists of women from rural areas of Sichuan Province, which is the province with the largest population in western China. However, our finding might not be generalizable to other areas of western China due to differences in population characteristics, cultural customs, social norms, and physical environments across various regions of western China (65, 66). Since the majority of previous studies in China were conducted in urban areas (6, 67, 68), future research on women from other resource-limited settings is needed to understand their health behaviors and develop more targeted health communication strategies. Fifth, for women from rural areas with limited resources, access to personal protective materials, such as masks and soap, maybe other environmental determinants of self-protective behavior. A study in China has shown that an adequate supply of masks and soap was associated with higher odds of mask-wearing and handwashing among the general population (68). Unfortunately, we did not measure access or supply in our study.

This study examined how individual characteristics, exposure to COVID-19-related information, and psychological response to COVID-19 affected self-protective behaviors of women in rural areas of western China. To the best of our knowledge, this is the first study that focused on this often-overlooked vulnerable group and their health behaviors during the peak of the COVID-19 outbreak. Our findings suggested that targeted messages and group-specific risk communication strategies may encourage self-protective behaviors among women from the rural areas of western China. More broadly, to promote behavioral changes in response to public health crises, the government should develop clear and actionable guidelines and adopt targeted health communication strategies to reach the most disadvantaged groups of the society.

The datasets analyzed for the current study are not publicly available due to ethical restrictions related to the consent given by participants at the time of study commencement. An ethically compliant dataset may be made available by the corresponding author on reasonable request and upon approval by the Sichuan University Medical Ethical Review Board. Requests to access the datasets should be directed to zhouhuan@scu.edu.cn.

The studies involving human participants were reviewed and approved by Sichuan University Medical Ethical Review Board (approval number of studies: KS2020246). The participants provided their written or oral informed consent to participate in this study.

RY and YW drafted the paper plan, developed the data analysis plan, and led the writing of the manuscript. RY, YW, CS, QW, and HZ designed the study. YM, WC, and HZ critically reviewed, discussed, and modified the manuscript. All authors contributed to the article and approved the submitted version.

This study was funded by Stanford University Research Foundation (Grant Number: 0040405502159) and National Natural Science Foundation of China (Grant Number: 71874114).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.