Lockdown and mandatory quarantine are commonly used and effective measures that are implemented by governments to contain the transmission of respiratory infectious diseases, including the COVID-19 disease [1,2]. Lockdown refers to general and widespread restrictions on movement, work, and travel for all people in a city, region, or country. Lockdown measures include travel restrictions, the mandatory closure of schools, and bans on nonessential commercial and social activities. Given the global spread of COVID-19 to 216 countries/regions, which has resulted in over 30.6 million infections and 950,000 deaths as of September 20, 2020 [3], more than 100 countries (eg, the United States, France, Australia, Thailand, and South Africa) have adopted various forms of lockdown measures to control the pandemic [4]. In mainland China, more than 80 cities in around 20 provinces and municipalities were put in lockdown [5], and over 780 million people were under certain travel restrictions [6].

Mandatory quarantine is a form of isolating people who are not ill, but may have been exposed to a disease that is dangerous to society [7]. People who had close contact with individuals who were confirmed/suspected to have COVID-19, and people with a history of travel were quarantined for 14 days in designated facilities (eg, governmental facilities and hotels) or home settings [8]. Globally, more than 140 countries (eg, the United States, the United Kingdom, Spain, Italy, and Singapore) have adopted mandatory quarantine measures for disease control [9-12].

Although lockdown and mandatory quarantine have played crucial roles in the sharp decrease of the number of newly confirmed/suspected COVID-19 cases [13-15], concerns have been raised over the threat that these measures pose to mental health, as these unprecedented measures have restricted daily routines and increased social isolation [16,17]. Empirical studies on the impact of lockdown/mandatory quarantine on mental health have yielded inconsistent results. According to the Office of National Statistics, more than 25 million people in the United Kingdom have experienced high levels of anxiety in late March 2020, which is when the lockdown was announced [17]. In a study conducted by Sibley et al [18], participants from New Zealand reported a slight increase in psychological distress, but less fatigue and no significant changes in rumination, feelings of belongingness, perceived social support, satisfaction with life, standards of living, future security, personal relationships, and health during the early phase of the nationwide lockdown compared to those during the pre-COVID-19 period. A study that was conducted in southern China during early February 2020 also found that people who were quarantined had a greater prevalence of anxiety and depression than those who were not affected by quarantine [19]. Furthermore, in mid-February 2020, Li et al [20] reported a positive association between perceived inconvenience to daily life caused by home quarantine and depression/anxiety among the general adult population in China. However, Zhu et al [21] reported that there was no significant differences in depression and anxiety between Chinese people who were and were not subjected to mandatory quarantine, and they concluded that although these mental health problems were not related to quarantine control measures, these measures did impact daily life. Another study in China even found a significantly lower prevalence of depression and anxiety among people under mandatory quarantine than among those who were infected by SARS-CoV-2 or the general public [22]. These inconsistent results highlight the importance of exploring potential underlying mechanisms (eg, the impacts of lockdown and quarantine on individuals) that may explain the relationship between lockdown and quarantine measures and mental health problems. We however did not identify such studies.

Pregnant women could be more susceptible to lockdown and quarantine measures and SARS-CoV-2 infection than the general population, due to their great need for social support and ongoing prenatal care services, concerns over fetal safety, immunocompromised status, and physiological and psychosocial adaptive changes during pregnancy [23]. Most studies on the mental health of pregnant women have small sample sizes (ie, 70 participants to around 560 participants), and these studies have only reported the prevalence of mental health problems during the COVID-19 pandemic [24-37]. We found 3 studies on the mental health of pregnant women that had larger sample sizes, with about ≥1000 respondents (ie, 946 participants to around 2421 participants) [24,25,30], and only 1 study (N=260) that investigated the impact of social isolation on mental health by simply asking pregnant women whether they believed that social isolation due to the pandemic affected their psychological well-being [27]. More studies that include large sample sizes and investigate the impact of lockdown/mandatory quarantine on the mental health of pregnant women during the COVID-19 pandemic are needed. Early detection and intervention can prevent the adverse impact that mental problems (eg, prenatal mental disorders) have on both mothers and children in the long term.

Social and cognitive mechanisms may play critical roles in mediating the relationships between lockdown/mandatory quarantine and mental health. First, lockdown and quarantine are isolation measures that, by their nature, may induce social isolation and reduce social resources, such as social support, and reductions in such social resources are a risk factor of mental health problems [16]. We found 2 studies with a sample size that ranged between 308 participants to around 403 participants. These studies investigated the association between social support and mental health problems among pregnant women during the COVID-19 pandemic, and both studies reported negative associations [26,28]. The mediation effects of social support are supported by the conservation of resources theory, which predicts that resource loss (eg, losses of social resources like social support) is the principal factor in the stress process and the cause of mental disorder development [38].

Second, the governmental implementation of unprecedented measures for disease control may be a stressful event for the public that induces maladaptive cognitive responses. According to the response styles theory, both depression and anxiety are related to faulty cognitive responses to stressors and negative emotions [39,40]. Rumination (ie, the repetition of the same feelings and thoughts) and catastrophizing (ie, having thoughts that explicitly emphasize the terror of what one has experienced) are common maladaptive cognitions, and responses to maladaptive cognitions emerge when individuals experience threatening and uncertain events [41]. When compared to the different kinds of maladaptive cognitive responses, anxiety has been found to be more related to catastrophizing, which focuses on future threats [42], and depression has been found to be more related to ruminative thinking, which concentrates on past negative experiences and emotions [43,44]. Empirical studies have also supported the mediation roles of rumination and catastrophizing between threatening events (eg, daily hassles) and depression/anxiety [41,45]. No study has assessed the mediation effects of rumination and catastrophizing in the context of COVID-19 or pregnant women.

eHealth refers to information and communications technologies in health care and the community. eHealth can be an optimal communication modality for people under stay-at-home orders, especially for those with time-sensitive health conditions, such as pregnancy [46]. eHealth and telemedicine services enable pregnant women to maintain their regular prenatal visit schedule and avoid the unnecessary risk of COVID-19 exposure [47]. Such online services may significantly affect individuals’ coping resources, stress appraisals, and perceived or actual social support [48,49], and may reduce the influence of external and environmental stress on individuals’ mental health during the COVID-19 pandemic. Thus, we hypothesized that the effects of control measures on interpersonal resources (eg, social support), cognitive status (eg, maladaptive cognition), and mental health might vary between pregnant women who use eHealth resources, such as using social media to obtain health information and online services to make appointments with doctors for prenatal care services during the COVID-19 pandemic, and those who did not use such resources. In addition, such extra resources may reduce the adverse effects of maladaptive cognition and enhance the protective effects of social support on mental health. We did not find any research that investigated the prevalence of eHealth resource use among pregnant women and the impact of using eHealth resources on the mental health of pregnant women.

This study aimed to investigate the prevalence of mental health problems (ie, depression and anxiety) among a large sample of pregnant women recruited from multiple regions in mainland China. We also assessed the direct and indirect effects of lockdown and mandatory quarantine on mental health problems through perceived social support and maladaptive cognition. We hypothesized that lockdown (ie, model path H1) and mandatory quarantine (ie, model path H2) would be positively associated with mental health problems. We further hypothesized that lockdown and mandatory quarantine would be indirectly associated with mental health problems through reduced perceived social support (ie, model path H3) and increased maladaptive cognition (ie, model path H4). Moreover, this study aimed to assess the moderation effects of eHealth-related variables, including using social media to obtain health information, using online prenatal care services, and making appointments with doctors during the COVID-19 outbreak, on pregnant women for each model path.

In this study, the inclusion criteria for the sample were (1) female sex, (2) age ≥18 years, (3) the ability to speak Chinese, (4) current pregnancy, and (5) the use of maternal health care services provided by the Maternal and Child Health Hospitals of the Chinese Preventive Medicine Association. Pregnant women who planned to terminate their pregnancy were excluded from this study. An online cross-sectional survey was conducted from February 24 to March 10, 2020. Eligible participants were identified from the records of Maternal and Child Health Hospitals from multiple regions of China and were invited to take part in the online survey by doctors through WeChat (Tencent Inc), which is the most widely used social media platform in Chinese populations. Interested participants visited the online survey through a link or quick response code and read the informed consent form before starting the survey. They were informed that clicking the “starting the survey” button implied informed consent. They were also informed that the study was anonymous and confidential, and that refusal to take part in the survey would not affect any services they would obtain. The survey took about 15 minutes to complete. No incentive was provided. A total of 19,515 pregnant women from all 34 provincial-level administrative regions in China (eg, 2127 pregnant women from Beijing, 4015 from Shandong, 3659 from Zhejiang, 1886 from Guangdong, 1250 from Hunan, 3178 from Shanxi, etc) completed the survey, with a valid response rate of 87.7%. This study was approved by the Survey and Behavioural Research Ethics Committee of the Chinese University of Hong Kong (Number SBRE-19-395).

Sociodemographic and maternal information, including age, education level, occupation, the area of residence, gestational duration, the number of children born, complication during pregnancy, the major means of using prenatal care service during the COVID-19 pandemic (ie, using online services, making appointments with doctors, and going to a hospital as usual), and the frequency of using social media to obtain health information in the past week, were obtained from the survey. Participants also reported on their lockdown and mandatory quarantine status by answering the following questions: (1) “Had the city, town, or county where you currently live been put under lockdown by the local government because of the COVID-19 epidemic” (response score: 0=no and 1=yes); and (2) “Had you been under mandatory quarantine (e.g., governmental facilities-, hotel- or home-quarantine) because of the COVID-19 epidemic” (response score: 0=no and 1=yes)?

Perceived social support was measured by 2 dimensions (ie, general social support and perceived change in social support during the COVID-19 pandemic compared to those during the pre-COVID-19 period). The following 2 survey items were used to assess this: (1) “Overall, to what extent did you receive social support from families, friends, and others during the COVID-19 epidemic” (response scale: 1=very poor to 10=very good); and (2) “To what extent did your social support become poorer or better during COVID-19, compared to that before the outbreak of COVID-19” (response scale: 1=much poorer to 5=much better)? Similar survey items have been used in previous studies [50].

Maladaptive cognition related to COVID-19 was measured with the short-form Cognitive Emotion Regulation Questionnaire (CERQ) [51]. The CERQ was developed to evaluate the cognitive aspects of emotion regulation when one experiences stressful or unpleasant events. Sample items from the CERQ include “I am preoccupied with what I think and feel about what I have experienced” and “I keep thinking about how terrible it is what I have experienced.” These items are rated on Likert scales (ie, 1=almost never to 5=almost always). The subscales of rumination and catastrophizing were used in this study. The Chinese version has been validated in previous studies [52]. The reliability of the 2 subscales was acceptable for our sample (Cronbach α=.66; Cronbach α=.84, respectively).

Depression was measured using the Patient Health Questionnaire-9 (PHQ-9) [53]. Respondents evaluated the presence (PHQ-9 response scale: 0=none to 3=almost every day) of 9 criteria for a depressive episode that occurred in the past 2 weeks, in accordance with the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) (eg, “Feeling bad about yourself—or that you are a failure or have let yourself or your family down?”). The cutoff points for mild, moderate, moderately severe, and severe depression were total PHQ-9 scores of 5, 10, 15, and 20, respectively. The Chinese version has been used in previous studies [54], and it had good internal consistency (Cronbach α=.83).

Anxiety was measured by the Generalized Anxiety Disorder 7-item (GAD-7) scale [55]. It is based on DSM-IV criteria and is used to measure the severity of generalized anxiety disorder based on the past 2 weeks. Participants respond according to a 4-point Likert-type scale (ie, 0=none to 3=almost every day). The cutoff points for mild, moderate, and severe anxiety were total GAD-7 scores of 5, 10, and 15 respectively. The Chinese version has been validated in previous studies [56]. It had a Cronbach α of .90 with our sample.

Descriptive statistics, including frequency, means, and standard deviations, were computed for participants’ sociodemographic characteristics. Differences in depression scores and anxiety scores based on sociodemographic characteristics were compared by either an independent 2-tailed t test or analysis of variance. The effect size (ie, Cohen d or Cohen f) was reported. Bivariate correlations between the key variables were presented. The effect size was considered low if the value of r varied around .10, medium if r varied around .30, and large if r varied by more than .50 [57]. Multiple indicators multiple causes (MIMIC) modeling was conducted to test the proposed mediation model. The goodness of fit was tested, and standardized path coefficients (ie, β) were reported. The mediation hypotheses were tested by bootstrapping analyses. The 95% confidence intervals of the indirect effects were obtained from 5000 bootstrap samples. The effect size (ie, proportion of mediation [PM]) was reported. Multigroup analyses were conducted to test the proposed moderators; P<.05 in the Chi-square difference test (Δχ²/Δdf) would suggest a significant moderation effect. The missing data rate was below 5%, and all missing values were replaced by using multiple imputation. The level of statistical significance was .05, and SPSS version 21.0 (IBM Corp) and Amos Version 26 (IBM Corp) were used for data analyses.

Tables 1-3 present the sociodemographic and maternal characteristics of the participants. Of 19,515 participants, 13,885 (71.1%) were aged 26-35 years, 11,627 (59.6%) had an education level of college or above, 10,741 (55%) did not have a baby before this pregnancy, and 17,856 (91.5%) did not experience any complications or comorbidities during pregnancy. Additionally, 5394 (27.6%) participants were unemployed/housewives (27.6%) and 5193 (26.6) were technical/administrative personnel (26.6%). The distribution of participants’ areas of residence was approximately even (municipality/provincial capitals: n=5930, 30.4%; general cities: n=6855, 35.1%; counties: n=6730, 34.5%). Participants’ average gestational duration was 25.4 weeks (SD 9.8 weeks).

With regard to the means of using prenatal care services during the COVID-19 epidemic, 10,189 (52.2%) participants went to the hospital for prenatal care as usual, 7568 (38.8%) made appointments with doctors, 640 (3.3%) used online services, and 1118 (5.7%) were uncertain. Most participants (n=17,734, 90.9%) used social media to obtain health information in the week before the survey.

Of the 19,515 participants, 12,209 (62.6%) participants reported lockdown in their areas of residence; 737 (3.8%) were subjected to mandatory quarantine; 8712 (44.6%) had probable mild to severe depression; 1442 (7.4%) had suicidal ideations, as measured by question 9 in the PHQ-9; and 5696 (29.2%) had probable mild to severe anxiety.

As seen in Tables 4 and 5, factors that positively significantly associated with both depressive and anxiety symptoms included young age, being a student, residing in counties, being in the early or final stages of pregnancy, using means of prenatal care other than making appointments with doctors, using social media to obtain health information, experiencing lockdown in the areas of residence, and being quarantined. Other factors that were significantly associated with greater anxiety included low education levels (P=.01) and complications during pregnancy (P<.001). With regard to the number of children born, participants who were giving birth for the first time reported significantly more depressive symptoms than those who were not giving birth for the first time (P<.001). They also reported more anxiety symptoms than those who had given birth to 1 child before this pregnancy (P=.002).

As seen in Table 6, living in an area under lockdown had significant positive associations with all the key psychological variables, including perceived general social support (P<.001), social support change (P<.001), rumination (P<.001), catastrophizing (P<.001), depression (P=.01), and anxiety (P=.03). Being quarantined was significantly and negatively associated with perceived general social support (P<.001) and social support change (P=.01), while it was significantly and positively associated with catastrophizing (P<.001), depression (P<.001), and anxiety (P<.001). However, all the correlations associated with lockdown and quarantine had small effect sizes. Both perceived general social support and social support change had small to moderate negative correlations with depression (r=−0.17, P<.001; r=−0.16, P<.001, respectively) and anxiety (r=−0.17, P<.001; r=−0.15, P<.001, respectively). Both rumination and catastrophizing had moderate positive correlations with depression (r=0.25, P<.001; r=0.28, P<.001, respectively) and anxiety (r=0.26, P<.001; r=0.31, P<.001, respectively).

MIMIC modeling showed that both the measurement model (χ²₁₂=411.75; comparative fit index [CFI]=.99; nonnormed fit index [NNFI]=.98; root mean square of error approximation [RMSEA]=.04, 90% CI 0.038-0.045) and structural model (χ²₁₄=495.21; CFI=.99; NNFI=.98; RMSEA=.04, 90% CI 0.038-0.045) (Figure 1) fit the data well. Lockdown was significantly and positively associated with perceived social support (B=.13; β=.04; P<.001) and maladaptive cognition (B=.12; β=.09; P<.001), but not with mental health problems (B=.01; β=.01; P=.33). Mandatory quarantine was associated with low perceived social support (B=−.27; β=−.03; P<.001), great maladaptive cognition (B=.07, β=.02, P=.01), and great mental health problems (B=.09; β=.04; P<.001). Perceived social support was negatively associated with mental health problems (B=−.07; β=−.23; P<.001), while maladaptive cognition was positively associated with mental health problems (B=.29; β=.39; P<.001). Detailed results regarding regression weights are provided in Multimedia Appendix 1.

The indirect effects of lockdown (B=.03; β=.03, 95% CI 0.02-0.03; P=.001; PM=78.8%) and quarantine (B=.04; β=.02, 95% CI 0.01-0.02; P=.001; PM=30.1%) on mental health problems were statistically significant. Specifically, the indirect effect of lockdown on mental health problems through perceived social support was negative (B=−.01; β=−.01; P<.001), and the indirect effect of lockdown on mental health problems through maladaptive cognition was positive (B=.03, β=.04, P<.001). The indirect effects of quarantine on mental health problems through perceived social support (B=.02; β=.01; P<.001) and maladaptive cognition (B=.02; β=.01; P<.001) were positive. The total effects of lockdown (B=.03; β=.03; P=.001) and quarantine (B=.13; β=.05; P=.001) on mental health problems were statistically significant.

The use of social media for obtaining health information during the COVID-19 pandemic significantly moderated the association between perceived social support and mental health problems (Δχ²=18.58, Δdf=2) (Multimedia Appendix 2). Specifically, the negative associations between perceived social support and mental health problems became stronger with the increased frequency of using social media (never: B=−.05; β=−.18; P<.001; sometimes: B=−.06; β=−.22; P<.001; always: B=−.08; β=−.27; P<.001).

The means of using prenatal care services during the COVID-19 pandemic significantly moderated the 3 model paths (Multimedia Appendix 3), including the paths from lockdown to maladaptive cognition (Δχ²=20.10, Δdf=3; went to hospital as usual: B=.16; β=.12; P<.001; made appointment with doctor: B=.07; β=.05; P<.001; online prenatal care: B=.13; β=.09; P=.03; uncertain: B=.05; β=.04; P=.25), the paths from mandatory quarantine to perceived social support (Δχ²=8.12, Δdf=3; went to hospital as usual: B=−.30; β=−.03; P=.004; made appointment with doctor: B=−.07; β=−.10; P=.504; online prenatal care: B=−.78; β=−.12; P=.01; uncertain: B=−.43; β=−.06; P=.07), and the paths from maladaptive cognition to mental health problems (Δχ²=11.48, Δdf=3; went to hospital as usual: B=.30; β=.41; P<.001; made appointment with doctor: B=.26; β=.37; P<.001; online prenatal care: B=.35; β=.42; P<.001; uncertain: B=.32; β=.38; P<.001).