The effect of home confinement on parental and child mental health is of great concern. Epidemiological studies have shown that children exposed to pandemics are particularly vulnerable to behavioral problems, including hyperactivity, conduct disorders, externalizing problems, and general psychological distress (1, 6, 7). Significant disruptions in family lifestyle during the home confinement, combined with heightened stress and anxiety can lead to psychological difficulties in preschoolers.

In the meantime, just like their children, adults face all kinds of stress. In particular, parents have to cope with social distancing and changes in their daily routine, such as working remotely or facing unemployment, combined with additional caring for their children during school hours. In some cases, the changes brought about by the confinement were accompanied by reduced income and new responsibilities, which likely exacerbated the pre-existing difficulties and pressures (8, 9). Yet, despite a growing focus on the impact of COVID-19 on adults' and children's mental health, few studies have explored the association between parental well-being index and child mental health problems in the context of the pandemic. Therefore, based on previous evidence, we propose the following hypothesis in the pandemic:

Parenting style refers to a series of practices that create an emotional environment and influence child development and well-being (10). Previous studies have reported that parents' mental health likely affects their parenting behaviors, with harsh parenting style, in turn, leading to child mental health problems (11). In the context of the COVID-19 pandemic, parents experiencing elevated levels of cumulative stress are even more likely to display more rigid and abusive parenting behaviors (12). Ineffective parenting practices are positively associated with child behavioral problems (13). In particular, parental verbal aggression alone as separate and distinct from physical punishment contributes to lower self-esteem in children (14). Concerning parenting strategies, the American Academy of Pediatrics recommends against physical and verbal punishment of children in favor of more effective disciplines for raising healthy children (15). Therefore, increased attention should be paid to investigating the mediating role of parenting style within the parent-child mental health association. Thus, the following hypothesis is developed in relation to the pandemic:

Sleep disturbances frequently occur in preschool children and are associated with a range of adverse health outcomes (16). During the pandemic, the confined preschoolers were reported to demonstrate changes in sleep patterns characterized by later bedtimes and wake times, longer nocturnal, and shorter nap sleep durations (17). Increasing research has indicated the impact of parental psychological functioning on sleep quality in young children. For example, parental depression and stress were linked to child sleep disturbances (18). Moderate/severe maternal depression symptoms were associated with increased odds of children aged 4–5 years sleeping <10 h/day (19). Meanwhile, it has been well-documented that sleep disturbances are associated with poor mental health in children. For example, children with sleep disturbances show more maladaptive emotional generation and regulation processes, more inattention, aggressive and hyperactivity-related problems, more peer problems and school readiness problems (20). Based on these findings, it can be inferred that sleep problems may increase child mental health problems. Thus, we propose the following:

In summary, harsh parenting and child sleep problems are important links that connect parental well-being index and child mental health problems. Environmental and lifestyle factors can affect sleep quality and quantity and lead to sleep disorders (21). Children who live in families with harsh parenting are significantly associated with lower verbal skills and increased behavioral problems, including internalization, externalization, and sleep problems (22). A prospective study demonstrated that hostile parenting predicts child sleep problems (23). Thus, harsh parenting style may be positively contributed to child sleep disturbances. Based on the above, parents with a low well-being index may implement rigid parenting behaviors, which further increase the odds of child sleep disturbances and enhance their mental health problems, a serial link demanding a systematic investigation. Thus, the following hypothesis is proposed:

The current study used data from a population-based online survey that included parents (including other caregivers) of children aged 3–6 years from 28 provinces across mainland China. Specifically, 25.7% of the questionnaires came from East China (Shanghai, Zhejiang, Jiangsu, and so on), 73.3% from southwest China (Chongqing, Sichuan, Guizhou, and so on), and 1% from other regions (Guangdong, Shanxi, Henan, and so on). The survey was conducted through the WeChat-based survey program, Questionnaire Star (https://www.wjx.cn/), a frequently used online survey platform in China, from March 15 to 29, 2020 when the families were on mandatory home confinement while filling the survey (24). The survey was administered through a combination of non-probabilistic convenience and snowball sampling. The scan code and link for the survey were posted to the public. The parents who agreed to participate were encouraged to forward the survey link to other parents. Participants endorsed their consent before moving on to the survey questions. To promote participation and data quality, a free online workshop entitled “how to help your child get sound sleep during COVID-19?” was delivered to the participants by a psychologist and sleep specialist (GW). We also used a series of logical algorisms in the program to maximize the accuracy and quality of the online survey, such as restricting responses range and de-identity. The participant could submit the questionnaire only after all items were completed to reduce the possibility of accidentally skipping items. According to the logged-in WeChat account, each participant was only allowed to participate once. Since the self-report questionnaires might be affected by participants' response bias, all participants remained anonymous and participated voluntarily.

Among 25,162 initial contacts, parents of 24,143 children consented to participate, yielding a 96.0% response rate (Figure 1). As we aimed to recruit a general and representative sample of the general population for assessment of child mental health problems during the pandemic, we did not screen and exclude those with suspected COVID-19 infection. We excluded 241 questionnaires with a completion time outside three standard deviations (M = 18.87 min, SD = 10.56 min) and 7,504 questionnaires outside the age range of 3–6 years old.

The survey consisted of a set of sociodemographic questions investigating children's age, gender, presence of siblings, family type (e.g., nuclear family, extended family, etc.), primary caregivers, and parental education level. The primary caregiver was defined as the child's main caretaker and was dichotomized into parental care vs. non-parental care. Parental education level was self-reported on four levels (undergraduate and above, junior college, high school or technical secondary school, and middle school and below). Following this, a series of standardized measures on child mental health, parental well-being index, and sleep problems were presented. Measures on physical activity and screen exposure were also included.

Child mental health problems were evaluated using the Strengths and Difficulties Questionnaire (SDQ). Parents were asked to rate their child's behaviors over the past 2 months on 25 items, compositing five subscales: peer problems, conduct disorders, hyperactivity, emotional problems, and prosocial behavior. We used the SDQ total difficulty score (excluding the prosocial behavior items), with a score of ≥14 indicating mental health problems (25). The Cronbach's α for SDQ was 0.73 and the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy was 0.89 (p < 0.001, Bartlett's test of sphericity) in the current sample.

The World Health Organization-Five Well-Being Index (WHO-5) was used to assess the parental well-being index over the past 2 weeks (26). On five mental health-related statements, parents were asked to rate the frequency on a six-point Likert scale (0 = Never to 5 = All the time). A total score was calculated, with a score of <13 indicating mental health problems (27). The Cronbach's α was 0.92 and the KMO measure of sampling adequacy was 0.89 (p < 0.001, Bartlett's test of sphericity) in the current sample.

Harsh parenting was evaluated by two questions: (1) to discipline and regulate a child's behavior, how many times have you physically punished your child (spanking or hitting) without hurting him/her or leaving bruises or marks? (2) To discipline and regulate a child's behavior, how many times have you scolded your child (yelling, shouting, or using words to humiliate him or her)? Parents were required to report the frequency of each statement on a 5-point scale (never, 1 or 2 times per week, 3 or 4 times per week, 5 or 6 times per week, almost every day).

The Children's Sleep Habits Questionnaire (CSHQ), which was used to assess child sleep disturbances over the past week (28), is a standardized and internationally recognized instrument consisting of 33 items covering eight domains: bedtime resistance, sleep anxiety, sleep onset delay, sleep duration, night waking, parasomnia, daytime sleepiness, and sleep-disordered breathing. Parents rated the frequency of each item occurring in their children over the past week on a 3-point scale (usually, sometimes, rarely). A total score >41 indicates global sleep disturbances. The Cronbach's α was 0.73 and the KMO measure of sampling adequacy was 0.84 (p < 0.001, Bartlett's test of sphericity) in the current sample.

Daily time spent on physical activity over the past week was reported as <30 min, 30–60 min, >60 min. Daily exposure to media over the past month was reported on weekdays and weekends. The average daily screen exposure time = (screen exposure time on weekdays^*5 + screen exposure time on weekends^*2)/7. As previous studies have linked physical activity and screen exposure time to child mental health problems (29–31), they were considered as covariates in the current study.

Missing data for all key variables were <1% (n = 154) and were handled by listwise deletion. Descriptive analyses were used for sociodemographic data. The multicollinearity test was conducted to exclude multicollinearity problems among the variables. We used Harman's single-factor test to verify the presence of common method bias. We also used Spearman correlations to investigate associations between parental well-being index, harsh parenting, child sleep problems, and child mental health problems. We calculated the Cronbach's alpha to show the reliability of the measures and used the Kaiser-Meyer-Olkin (KMO) test to show its validity. Data analysis was conducted using SPSS 26.0. Finally, we employed PROCESS Model 6 in SPSS 26.0 to examine Hypotheses 1 to 4. In addition, a 95% bias-corrected confidence interval with 5,000 bootstrap samples was applied to determine the significance of the mediational effect.

Descriptive data are summarized in Table 1. A total of 16,398 children aged 3–6 years (M = 4.69 years, SD = 0.75 years) were finally included in the study. The demographic survey results showed that participants comprised 8,512 boys (51.9%) and 7,886 girls (48.1%). Of the children, 58.0% had at least one sibling. Family type distribution was 42.4% with a nuclear family, 54.5% with extended family, 2% with single-parent family, and 1.1% with other family types. The primary caregiver of the children was mostly the parent (71.1%), while for the remaining 28.9% of children, it was a grandparent or others. There were 36.7% parents with the highest level of undergraduate education and above, 22.3% with junior college, 26.7% with high school or technical secondary school, and 24.3% with the lowest level of middle school or below. In 56.5% of the preschoolers, physical activity was <30 min/day, in 35.8%, it was 30–60 min/day, and in 7.7%, it was more than 60 min/day. The average screen time among preschoolers was 2.05 ± 1.87 h/day.

The multicollinearity test revealed that the tolerance of each variable was between 0.90 and 0.92, and the variance expansion factor ranged between 1.09 and 1.11. These findings revealed no multicollinearity problem among the variables.

The results of Harman's single-factor examination showed that eight factors had eigenvalues >1, and the first factor explained 11.89% of the total variance. This result did not exceed the critical value of 40% (32). The above methods indicate an absence of serious common method bias in the current study.

The Spearman correlations (Table 2) showed that significant but moderate positive relationships between child mental health problems with harsh parenting (r = 0.27, p < 0.01) and child sleep problems (r = 0.34, p < 0.01), as well as between harsh parenting and child sleep problems (r = 0.20, p < 0.01). There were significant but moderate negative relationships between parental well-being index and harsh parenting (r = −0.25, p < 0.01), child sleep problems (r = −0.26, p < 0.01), and child mental health problems (r = −0.23, p < 0.01).

Table 3 and Figure 2 show that the direct effect of parental well-being index on child mental health problems was significant and negative (β = −0.10, p < 0.001). In the path of “parental well-being index → harsh parenting → child mental health problems,” parental well-being index had a significant negative impact on harsh parenting (β = −0.08, p < 0.001), which in turn had a significant positive impact on child mental health problems (β = 0.49, p < 0.001). In other words, the parental well-being index enhanced the children's psychological health by reducing harsh parenting. In the path of “parental well-being index → child sleep problems → child mental health problems,” parental well-being index had a significant negative impact on child sleep problems (β = −0.24, p < 0.001), which in turn had a significant positive impact on child mental health problems (β = 0.20, p < 0.001). Thus, the parental well-being index improved child psychological health problems by weakening child sleep problems. Finally, serial mediation analysis revealed that in the path of “parental well-being index → harsh parenting → child sleep problems → child mental health problems,” harsh parenting had a significant positive impact on child sleep problems (β = 0.52, p < 0.001). This indicated that the parental well-being index reduced child sleep problems by alleviating the harsh parenting, which in turn improved child mental health problems.

Table 4 shows that the direct and total indirect effect of parental well-being on child mental health was −0.10 (95% CI −0.11, −0.08; p < 0.001) and −0.09 (95% CI −0.10, −0.09; p < 0.001), respectively, with a total effect of −0.19. The mediating effect of parental well-being index on harsh parenting was −0.04, p < 0.001, with an effect size of 19.76 percent. The mediating effect of parental well-being index on child mental health problems through sleep problems was −0.05, p < 0.001, with an effect size of 24.84 percent. The mediating effect of parental well-being index on child mental health problems through harsh parenting and sleep problems was −0.01, p < 0.001, with an effect size of 4.13 percent.