Sleep quality is a key indicator of health. For community workers, good sleep quality can not only help them better treat the public but also maintain the best state of their immune system to prevent infection [10]. Mental health and sleep are also affected by many social and cultural factors [9]. Social support refers to the care and support from others [11]. Social support can be classified into two types: one is objective, practical or visible support, including direct material support and social networks; the other is subjective, experienced or emotional support, which refers to individual feelings, such as emotional experiences and/or degree of satisfaction with being respected, supported, and understood in society [10]. However, objective social support and subjectively perceived support from others have different effects on health [3,4]. This research focuses on the subjectively perceived social support, that means the effective emotional support and help from relatives, friends, leaders, and colleagues in life, and the impact on mental and physical health.

Previous studies have found that adequate social support has a positive impact on mental health and sleep function [12]. During the COVID-19 infection epidemic, the level of social support of medical staff was positively correlated with sleep quality [9]. It is found that for women, the social support of family members is significantly related to their sleep quality [9]. Research on infertile women found that social support plays a moderating role between sleep quality and anxiety and depression [10]. Studies on sleep quality of the elderly have found that social support from being close to others can significantly improve sleep quality [13]. There is a significant positive correlation between sleep quality and social support in the elderly [14]. Based on this, we propose hypothesis 1: There is a significant correlation between social support and sleep quality.

The research found that there is a significant correlation between social support and psychological resilience [15–22]. Psychological resilience (PR) is an evaluation of personal stress coping ability. High psychological resilience can make people better recover from stress events and improve their mental health under pressure. Individuals with a high sense of social support are more likely to recover from negative events when facing stressful events because they feel more support and experience more emotional support. This may be the reason why social support is positively correlated with psychological resilience. Everyone’s psychological resilience is constantly changing under the influence of external and internal factors [23]. Effective social support can help people cope with the crisis and enhance their adaptability to change [24]. The results show that psychological resilience is positively correlated with perceived social support [15,16,22].

Similarly, the research also found a significant positive correlation between sleep quality and psychological resilience [25]. Studies have found that improving psychological resilience indirectly improves sleep quality [26]. Training emotional psychological resilience can improve sleep quality [27,28].

According to the stress-buffering theory of social support. Protective effects of good social support foster resilience; meanwhile, resilience also enhances people’s ability to cope with stress [29]. Therefore, we propose hypothesis 2: Psychological resilience plays a mediating role between social support and sleep quality.

The COVID-19 infection epidemic is an acute stress event. Studies have found that stress leads to abnormal decreases in hormones such as cortisol and progesterone. Corticosteroids may play a negative role in reducing the negative emotions of patients through corticosteroids [30]. The results showed that the degree of psychological distress of medical staff was great, including the symptoms of post-traumatic stress disorder (PTSD) [31]. Residents living in Ebola virus outbreak areas reported more somatization and anxiety symptoms, dysfunction, and/or Ebola-related concerns [2].

It is found that social support is negatively correlated with perceived stress [32–35]. Social support was negatively correlated with post-traumatic stress disorder (PTSD) [36,37]. The quality of perceived social support, rather than quantity, reduces blood pressure’s response to stress [38]. Perceived stress can lead to a decline in health and well-being. Conversely, health problems and unhappiness can also lead people to think that environmental stress is extremely high [39].

Also, the research found that there was a significant correlation between perceived stress and sleep quality [40–45]. There was a significant negative correlation between perceived stress and sleep quality [46–48]. There was a significant positive correlation between stress perception and fatigue [49].

On the contrary, other studies have found that sleep quality is significantly positively correlated with stress perception [48,50], for example, occupational stress positively predicts sleep quality [51]. The level of stress management was positively correlated with sleep quality [52]. Mindfulness-based decompression significantly improves sleep quality [53,54]. Based on this, we propose hypothesis 3: Perceived stress mediates social support and sleep quality.

Previous studies have found that psychological resilience is negatively correlated with perceived stress [34,55–59]. Psychological resilience significantly negatively predicted perceived stress [39,60,61]. Perceived stress can regulate the protective effect of psychological resilience on suicide [62].

It is found that psychological resilience is related to stressful life events. When more stress events affect individuals, psychological resilience can cushion perceived stress; but when stress is not chronic, psychological resilience can’t cushion stress [63]. It is found that acute stress is significantly correlated with cognitive emotion regulation, social support, and resilience. Social support and resilience have multiple mediating effects on acute stress response [64].

Primarily, According to the stress-buffering theory of social support [24,29]. Protective effects of good social support foster resilience; meanwhile, resilience also enhances people’s ability to cope with stress [65]. The level of perceived stress is tightly related to sleep quality since the stress system serves as effectors on the wake-sleep centers of the brain [66]; therefore, sleep quality is added to the model, supposed to be affected by the level of perceived stress. Secondly, according to the main-effect theory of social support [29], social support is hypothesized to be directly associated with sleep quality in the model. Therefore, this research further proposes Hypothesis 4: Psychological resilience and perceived stress play a chain mediating role between social support and sleep quality.

Researchers have used different methods to measure psychological resilience and found that men have stronger psychological resilience than women [67]. On the contrary, the research of the elderly found that the psychological resilience level of elderly women was higher than that of elderly men [68]. In addition, the research found that the sleep quality of adult women is worse than that of adult men [69]. Based on this, this research will further examine the gender moderating effect on mediating role after testing the mediating effect. To sum up, by combing and analyzing the existing literature, this paper discusses the relationship between social support and sleep quality, gender differences, and its mechanism to improve the sleep quality of front-line community workers in Wuhan During the COVID-19 infection epidemic.

The four variables, social support, psychological resilience, perceived stress, and sleep quality, and their interaction can be analyzed by structural equation modeling (SEM), which is a multivariate analysis method to determine the structural relationship between the measured variables. Therefore, this observational research aims to determine the impact of social support on the sleep quality of front-line community workers in Wuhan by using SEM. This research mainly analyzes the direct impact of social support on sleep quality, and the mediating role of different psychological states in the impact of social support on sleep quality, and builds a hypothetical model as shown in Fig. 1.

The study was designed in accordance with the tenets of the Declaration of Helsinki. Approval from the ethical authority of School of Educational Science, Huazhong University of Science and Technology was granted. Confidentiality was ensured by placing anonymous codes on the self-report questionnaires, and subjects signed a statement providing informed consent.

Using the Questionnaire Star platform (Wenjuanxing, http://www.wjx.cn), there are 500 front-line community workers in Wuhan, including 322 males and 178 females with an average age of 38.7 ± 10.3 years. After completing the questionnaire, some interviews were conducted with these community workers to understand the nature and content of their work. We learned that their tasks included door-to-door investigation of febrile patients, reporting of the epidemic situation, sending confirmed and suspected patients to hospitals, and solving the household chores of isolated residents, such as delivering vegetables, buying medicine, and disinfecting public places in the community. As a matter of fact, one person took many positions.

The perceived social support scale (PSSS) [17]. It contains 12 items and includes three subscales: family support, friend support, and other support. For example: “I can get emotional help and support from my family when I need it” (family support); “my friends can really help me” (friend support); some people (leaders, relatives, colleagues) will support me when I encounter problems. The higher the score, the higher the level of social support (1 = strongly opposed, 7 = strongly agreed). In this research, the Cronbach α coefficient of the scale was 0.963.

The Connor-Davidson Resilience Scale (CD-RISC) was originally developed by Connor and Davidson [15,70]. After that, Campbell sills and Stein (2007) revised the CD-RISC. The revised version contained 10 items, with 5 grades of score: 1 for “never,” 5 for “always.” The score of 10 items was added up to get the score of psychological resilience. In this research, the Cronbach α coefficient of the scale was 0.943.

The perceived stress scale (PSS) was used to evaluate the perceived stress [40,71]. In order to measure the perceived stress more comprehensively, 10 items were randomly selected from the perceived stress scale. In this research, the Cronbach α coefficient of the scale was 0.878.

The Pittsburgh sleep quality index (PSQI) was compiled by Buysse et al. [13]. There were 18 self-rated items on the scale. Seven components of sleep quality were measured: subjective sleep quality, sleep time, the amount of sleep, sleep efficiency, sleep disorders, hypnotics, and daytime dysfunction. The higher the score, the worse the sleep quality. In this research, the Cronbach α coefficient of the scale was 0.917.

SPSS 25.0 and Process Syntax were used to perform the regression analysis and construct the predicted moderated mediation model. All data were expressed as the mean ± standard deviation (SD), median (interquartile range), or number (frequency). The data of social support, resilience, perceived stress and PSQI showed a non-normal distribution, so we normalized the data using Cumulative Relative Frequency Distribution method and standardization method. After normalization, the data of PSQI showed the normal distribution, p > 0.05, skewness = 0.089 ± 0.109, kurtosis = −0.237 ± 0.218; the data of perceived stress showed the normal distribution, p > 0.05, skewness = −0.080 ± 0.109, kurtosis = −0.257 ± 0.218. However, the data of social support still did not obey the normal distribution, p < 0.05, skewness = −0.242 ± 0.109, kurtosis = −0.498 ± 0.218; the data of psychological resilience also did not obey the normal distribution, p < 0.05, skewness = −0.326 ± 0.109, kurtosis = −0.545 ± 0.218. Therefore, we use the method of nonparametric test in the follow-up statistics. Non-parametric test was used to compare the psychological resilience and social support of different gender groups separately. Pearson correlation was used to analyze the correlation among social support, psychological resilience, perceived stress and PSQI. All the tests were bilateral, and the significance level was α = 0.05.

The possible multicollinearity was quantified using the variance infiltration factor (VIF). This parameter indicates how much the variance of a given covariable is increased compared to complete absence of correlation (orthogonality) and is generally agreed to indicate potential model instability if higher than five. In order to detect potentially influential multicollinearity between social support, psychological resilience and perceived stress in the constructed models we calculated the variance inflation factor (VIF). We used linear regression analysis for collinearity diagnosis, with PSQI as the dependent variable and social support, resilience and persistent stress as independent variables. The results showed that the VIF of social support = 1.617; the VIF of persistent stress = 1.568; the VIF of psychological resistance = 1.927. It shows that there is no serious multicollinearity problem in our data.

In this research, Haman single factor method was used to test the common method bias. All items of social support, resilience, perceived stress and sleep quality index were analyzed without rotation to determine the minimum number of factors necessary for variable variation. The results showed that there were 18 factors with eigenvalues greater than 1, and the variance explained by the first factor was 34.87%, less than the critical standard of 40%, indicating that there was no serious common method deviation in this research.

The results of the correlation analysis are shown in Tab. 1. The sleep quality index (PSQI) of front-line social workers was significantly negatively correlated with social support, psychological resilience, and perceived stress (p < 0.001); there was a significant positive correlation between social support, psychological resilience, and perceived stress (p < 0.001). In this research, the relationship between the variables is relatively close, and the absolute value of the correlation coefficient is between 0.260 and 0.597, which is suitable for further analysis.

Two independent samples nonparametric test was conducted on the psychological resilience of subjects of different genders, and Mann-Whitney U test was used. The results showed that the level of psychological resilience of males was significantly higher than that of females (p < 0.001).

Similarly, two independent samples nonparametric test was conducted on the social support of subjects of different genders, and Mann-Whitney U test was used. The results showed that the level of social support of males was also significantly higher than that of females (p = 0.031).

Then, we conducted a one-way ANOVA analysis on the relationship between perceived stress of front-line community workers and gender and contact history. Before the analysis of variance, the homogeneity test of variance is carried out. In Levene’s test, F_{(1, 498)} = 0.982, p = 0.322, indicating homogeneity of variance and analysis of variance can be carried out.

Taking the perceived stress of community workers as dependent variables, the perceived stress scores of different genders and contact history were analyzed by one-way ANOVA. The results showed that the main effect of contact history on perceived stress was significant, F_{(1, 498)} = 16.859, p = 0.004, η²_p = 0.033; the main effect of gender on perceived stress was significant, F_{(1, 498)} = 8.550, p = 0.650, η²_p = 0.017; the interaction between gender and contact history was significant F_{(1, 498)} = 2.774, p = 0.096, η²_p = .006. Further simple effect analysis showed that for male front-line community workers, the stress they perceived when they had contact history (M = 30.950, SD = 0.795) was significantly lower than that of those without contact history (M = 33.872, SD = 0.457), F_{(1, 130)} = 10.162, p = 0.002; for female front-line community workers, the stress they felt when they had contact history (M = 8.325, SD = 0.454) was also significantly lower than that of the people without contact history (M = 6.541, SD = 0.261), F_{(1, 130)} = 11.576, p < 0.001. This shows that, for both male and female front-line community workers, compared with those with contact history, the front-line community workers without contact history feel a lower stress level. See Tab. 2.

Next, we analyzed the relationship between sleep quality index (PSQI) and gender and contact history of front-line community workers by one-way ANOVA. Before the analysis of variance, the homogeneity test of variance is carried out. In Levene’s test, F_{(1, 498)} = 0.510, p = 0.476, indicating homogeneity of variance and analysis of variance can be carried out.

Taking the PSQI of front-line community workers as a dependent variable, the single factor analysis of variance was conducted on the PSQI of different gender and contact history. The results showed that the main effect of contact history on PSQI was significant, F_{(1, 498)} = 8.238, p = 0.004, η²_p = 0.016; however, the main effect of gender on PSQI was not significant, F_{(1, 498)} = 0.207, p = 0.650, η²_p < .001; the interaction between gender and contact history on PSQI was not significant, F_{(1, 498)}= 0.071, p = 0.791, η²_p < .001. Further simple effect analysis showed that for male first-line community workers, PSQI (M = 8.325, SD = 0.454) with contact history was significantly higher than that without contact history (M = 6.541, SD = 0.261), F_{(1, 130)} = 11.576, p < 0.001; while for female community workers, there was no significant difference in their sleep quality. This shows that for male front-line community workers, contact history has an impact on sleep quality, compared with community workers without contact history, front-line community workers with contact history have poorer sleep quality, while women do not have this difference. See Tab. 3.

In order to effectively control the measurement error, the structural equation modeling method is used to test the multiple mediating effects. Firstly, after controlling gender, age, and working years, the total effect of social support on sleep quality index was tested, and the path coefficient was significant (β = −0.260, t = −6.004, SE = 0.043, p < 0.001). Secondly, the mediating variables, psychological resilience, and perceived stress are added to the model to obtain the mediating model shown in Fig. 2. The results show that (see Tab. 4), social support can significantly positively predict psychological resilience; when social support and psychological resilience predict perceived stress at the same time, social support and psychological resilience have a significant positive predictive effect; when the three predict the sleep quality of individuals at the same time, perceived stress can significantly predict individual sleep quality, while social support and psychological resilience have no significant predictive effect on sleep quality. This indicates that psychological resilience and perceived stress play a completely mediating role in the influence of social support on sleep quality.

Next, the results of the direct test of mediating effect (see Tab. 5) showed that the bootstrap 95% confidence interval of the total indirect effects of psychological resilience and perceived stress did not contain 0 value, indicating that the two mediators had a significant mediating effect between social support and sleep quality. The mediating effect is composed of three indirect effects: First, the indirect effect produced by the path of social support → psychological resilience → sleep quality, with a confidence interval of 0, indicating that the indirect effect produced by this pathway is not significant; second, the confidence interval of the indirect effect produced by the path of social support → perceived stress → sleep quality does not contain 0 value, indicating the perceived Stress has a significant indirect effect (−0.043, 55.48%) between social support and sleep quality; third, the confidence interval of indirect effect 3 produced by the path of social support → psychological resilience → perceived stress → sleep quality does not contain 0 value, indicating that the indirect effect produced by this pathway (−0.026, 34.26%) also reaches a significant level.

According to the results of previous studies, there are significant differences in perceived stress between males and females. Based on this, this research examined whether the mediating effects of psychological resilience and perceived stress have cross-gender consistency. In order to reduce multicollinearity, all variables are standardized. First, we use perceived stress as a dependent variable to test whether gender has a moderating effect on the relationship between social support and perceived stress. The results showed that social support and gender had no significant predictive effect on perceived stress (β = 0.129, t = 1.484, SE = 0.086, p = 0.138). There was no interaction effect between gender and social support.

Secondly, the moderating role of gender is shown in Tab. 6. After adding gender into the model, the multiplicative term of social support and gender had a significant positive predictive effect on perceived stress (β = 0.163, t = 2.082, p = 0.038). It shows that gender plays a moderating role in the relationship between social support and perceived stress. The moderating effect is shown in Tab. 6 and Fig. 3. Further simple slope analysis showed that: For male subjects, social support had a significant positive predictive effect on perceived stress, simple slope = 0.122, t = 2.262, SE = 0.052, p = 0.019; for female subjects, social support had a significant positive predictive effect on perceived stress, simple slope = 0.285, t = 4.091, SE = 0.070, p < 0.001. Compared with male subjects, female subjects’ social support has a stronger positive predictive effect on perceived stress, and the predictive effect of social support on perceived stress shows an upward trend. See Tab. 7 and Fig. 4.

According to the results of previous studies, there are significant differences in psychological resilience between males and females. Based on this, this research examined whether the mediating effects of psychological resilience and Pittsburgh sleep Quality Index (PSQI) have cross-gender consistency. Firstly, we use PSQI as a dependent variable to test whether gender has a moderating effect on the relationship between psychological resilience and PSQI. The results showed that psychological resilience and gender had no significant predictive effect on PSQI (β = −0.186, t = −2.001, SE = 0.093, p = 0.046). There was a significant interaction effect between gender and psychological resilience.

Secondly, the moderating role of gender is shown in Tab. 8. After adding gender into the model, the multiplicative term of psychological resilience and gender had a significant positive predictive effect on PSQI (β = −0.233, t = −1.979, p = 0.048). It shows that gender plays a moderating role in the relationship between psychological resilience and PSQI. The moderating effect is shown in Tab. 8 and Fig. 5. Further simple slope analysis showed that: for male subjects, psychological resilience had no significant predictive effect on PSQI, simple slope = 0.034, t = 0.549, SE = 0.062, p = 0.583; for female subjects, psychological resilience had a significant negative predictive effect on Pittsburgh sleep Quality Index (PSQI), simple slope = −0.163, t = −1.870, SE = 0.087, p = 0.042. Compared with male subjects, female subjects’ psychological resilience has a stronger negative predictive effect on PSQI. Moreover, the relationship between psychological resilience and PSQI shows a downward trend. See Tab. 9 and Fig. 6.

The purpose of this research was to explore the relationship between social support and sleep quality in Wuhan during the epidemic period. The results showed that there was a significant positive correlation between social support and sleep quality, which verified hypothesis 1. This result is consistent with the existing research results [9]. It can be seen that physical and mental health and security, and then have a higher quality of sleep. Therefore, for the physical and mental health of front-line community workers, especially their sleep quality under the epidemic situation, social support plays an important role; the support from family, friends, and leaders, as well as the respect, support and, care from families, friends, and colleagues may promote the improvement of sleep quality to a certain extent. The previous study found that positive attitudes and strong social and family support may help to resist acute stress [72]. This is consistent with the stress-buffering theory of social support [24,29].

This research found that the mediating role of psychological resilience between social support and sleep quality is not significant, hypothesis 2 has not been verified. However, the results of correlation analysis show that there is a significant positive correlation between psychological resilience and social support, which may be due to the fact that individuals will have more psychological resources to deal with stressful events when they feel more social support and get more respect and support. Therefore, they are more likely to recover from the stress events faster and show higher psychological resilience [15–22]. The results of correlation analysis also found that there is a significant positive correlation between psychological resilience and sleep quality, which may be due to the fact that individuals with high psychological resilience can better face the stressful events in life and better deal with the impact of negative emotions. Therefore, the mental health level is relatively high, and the sleep quality may be improved, which is consistent with the existing research [15–22]. The community work has certain risks, and even directly contacts with the confirmed patients or suspected patients. Compared with the general work, the possibility of exposure to the risk is high. The results showed that the degree of psychological distress of medical staff was great, including the symptoms of post-traumatic stress disorder (PTSD) [31]. In this special situation, social support, as one of the important sources of strength, will inevitably have an impact on the psychological and physical health of front-line community workers. Therefore, the improvement of psychological resilience is helpful to improve their sleep quality and promote their physical and mental health.

This research also found that perceived stress played a mediating role between social support and sleep quality, which verified hypothesis 3. This research explored the relationship between these three variables and confirmed that social support is an important factor affecting sleep quality, and it also increases the mediating factor of social support on sleep quality. However, this is not consistent with the research results of high social support promoting perceived stress [32–35]. But it is consistent with the research that perceived stress has a positive effect on sleep quality [48,50]. The reason for the inconsistency with the existing research results may be due to the particularity of the COVID-19 infection epidemic. During the outbreak of the COVID-19 infection epidemic, the risk of illness is much higher than that before the outbreak of the COVID-19 infection epidemic. Especially for Wuhan residents, people are in danger. Their mental health and pressure may not be consistent with those before the outbreak of the COVID-19 infection epidemic, and the perceived stress baseline is higher than that before the COVID-19 infection epidemic. Moreover, due to the sudden outbreak of the epidemic, the pressure generated is severe.

In addition, this research examined the chain mediating role of psychological resilience and perceived stress between social support and sleep quality of community workers during the COVID-19 infection epidemic. The perceived social support of community workers can indirectly predict sleep quality through the chain mediation of psychological resilience perceived stress, which verifies Hypothesis 4. This is inconsistent with the previous results that psychological resilience helps to reduce perceived stress. Previous studies have found that psychological resilience is significantly negatively correlated with perceived stress [34,55–59]. However, this research found a significant positive correlation between psychological resilience and perceived stress. The reason may be that in a special period such as the COVID-19 infection epidemic, the perceived stress level of residents is higher than that before the outbreak of the COVID-19 infection epidemic, especially the front-line community workers in Wuhan have a high chance to be exposed to confirmed patients or suspected patients and face a great risk of infection. Previous studies can explain this phenomenon. Studies have found that psychological resilience is related to stressful life events. When more stress events affect individuals, psychological resilience can cushion perceived stress; but when stress is not chronic stress, psychological resilience can not cushion stress [63]. However, the COVID-19 infection epidemic is an acute stress event. Studies have found that stress leads to abnormal decreases in hormones such as cortisol and progesterone. Corticosteroids may play a negative role in reducing the negative emotions of patients through corticosteroids [30].

The results showed that the level of psychological resilience of males was significantly higher than that of females. Similarly, the results showed that the level of perceived stress of males was significantly higher than that of females.

In addition, this study found that there is an interaction between gender and contact history in terms of sleep quality. Specifically, for male front-line community workers, their sleep quality will be affected by contact history. Male front-line community workers with contact history have worse sleep quality, while female front-line community workers have no such difference. Therefore, we should pay more attention to the male front-line social workers who have contact history in the later stage of psychological guidance or counseling for such groups. In addition, the results showed that the level of psychological resilience of males was significantly higher than that of females. Similarly, the results showed that the level of perceived stress of males was significantly higher than that of females.

In the model of this study, gender moderates the relationship between social support and perceived stress, and the influence of social support on perceived stress of women is higher than that of men. In addition, gender moderates the relationship between psychological resilience and PSQI, and only women’s psychological resilience had a negatively predictive effect on PSQI, while men did not, which means that psychological resilience of female frontline community workers can positively predict sleep quality. The mediating effect of gender differences may be due to the lower resilience of women than men [67], for example, women are less likely to recover from stressful events than men. This is consistent with the results of previous studies [67], and the same results were obtained in this research.

In conclusion, during the COVID-19 infection epidemic, there are two ways to influence the sense of social support on the physical health of front-line community workers in Wuhan city. Social support is an environmental variable to improve sleep quality, and social support acts on sleep quality through the chain mediation of psychological resilience and perceived pressure. This research reveals that under the COVID-19 infection epidemic, for the community workers working in the front line, we should pay attention to improve their social support, improve their psychological resilience, reduce their perceived pressure so as to improve their sleep quality and promote their mental and physical health.

This research has the following limitations. Firstly, this survey focuses on the relationship between social support and sleep quality during the period of the COVID-19 infection epidemic, as well as the mediating role of psychological resilience and perceived stress, without considering the long-term impact of social support before the occurrence of the COVID-19 infection epidemic; secondly, in the analysis of perceived stress and sleep quality, various sources of stress are not taken into account; thirdly, the relationship between perceived stress and sleep quality is analyzed, which is cross-sectional research, only using the form of a questionnaire survey, so long-term follow-up experiments and more field studies are needed to verify the experimental results.