This was a prospective cohort study conducted in a tertiary maternal and child health hospital in Beijing, China. The hospital was responsible for the prenatal care of all pregnant women living in Tongzhou district of Beijing. The primary aim of this cohort study is to investigate the short- and long-term health effects of prenatal exposures (eg, poor sleep quality) on mothers and their children. Baseline recruitment was conducted in January and February 2020, and pregnant women who visited the outpatient clinic for the first prenatal examination at Tongzhou Maternal and Child Health Hospital were recruited when they met the following inclusion criteria: <14 gestational weeks, singleton pregnancy, plan to have antenatal care and delivery in Tongzhou Maternal and Child Health Hospital, and resided in Tongzhou during the past half year and have no plan to move out after delivery. The study was approved by the institutional review boards at Peking University (IRB00001052–18003), and all participants gave written informed consent at the enrollment.

Baseline information was collected at the first prenatal visit by trained medical workers through a standardized questionnaire, such as sociodemographic characteristics (age, educational level, region, and family income), history of previous health (cesarean section, preterm birth, miscarriage, and fist pregnancy or not), prepregnancy weight, smoking status, physical activities, dietary diversity, and sleep quality. There were 504 pregnant women that met the inclusion criteria and were recruited at baseline. By August 2020, 11 women moved out of Tongzhou, 27 women were lost to follow-up, and 10 women transferred to other hospitals. Finally, the remaining 456 participants were included in this study.

We collected the information on media use about COVID-19 by the following question: “How long did you spent on reading COVID-19 news every day from social media (official or unofficial)?” Participants were divided into five COVID-19–specific information exposure groups (<0.5 hours, 0.5 hours-1 hour, 1 hour-2 hours, 2-3 hours, and >3 hours) by their answers regarding the time spent on reading COVID-19 news. Because of the similar prevalence of miscarriage in the 0.5 hours-1 hour and 1 hour-2 hours groups in this study, we combined the two groups into one (0.5-2 hours) group as the reference group in the final analysis.

The follow-up of the pregnancy outcomes was conducted by local medical workers. Pregnant women took regular antenatal care and delivered in the hospital. The information on pregnancy outcomes was obtained though the medical electronic information system in the hospital, which automatically recorded information during each antenatal care and delivery. Miscarriage is defined as a pregnancy loss that occurs before 20 completed weeks of gestational age [10]. The prevalence of miscarriage was defined as the proportion of participants who had a miscarriage to all participants.

Covariates were collected at the first prenatal visit, including age, educational level, region, family income, history of cesarean section, history of preterm birth, history of miscarriage, gravidity, prepregnancy weight, smoking status, physical activities, dietary diversity, and sleep quality. Prepregnancy BMI was calculated using weight (in kilograms) divided by the square of height (in meters). We assessed dietary diversity using nine food groups, as reported in previous studies [16]. The participants reported their consumption frequencies of various food groups, including meat, vegetables, fish, eggs, fruits, legumes, milk, rice, and nuts. The dietary diversity score (DDS) was calculated, with scores ranging from 0 to 9. Inadequate dietary diversity was defined as DDS<7 [16]. Sleep quality was evaluated by the Pittsburgh Sleep Quality Index (PSQI) [17]. The PSQI is the gold standard questionnaire for assessing subjective sleep quality and is framed in a 4-point Likert scale (0-3) analyzing seven factors, including subjective sleep quality, sleep duration, sleep latency, habitual sleep efficiency, sleep disturbances, use of sleeping medications, and daytime dysfunction [17]. The scores from each component were added to give a sum score, also called a total score (range 0-21). Poor sleep quality was defined as a PSQI≥5 in the pregnant women [18].

Mean (SD) values and proportions of baseline characteristics were calculated. We calculated the mean (SD) for age. We used proportions to describe baseline characteristics of pregnant women, such as age group, region, and educational levels, and the chi-square test or Fisher exact probability test was used to compare the distributions of baseline characteristics according to time spent on reading COVID-19 news. Prevalence of miscarriage and its 95% CI was calculated. Miscarriage prevalence in women with different characteristics were also compared using chi-square test or Fisher exact probability test.

Multivariable log-binomial regression models were used to estimate the adjusted risk ratios (RRs) and their 95% CIs of miscarriage for women with different exposures of COVID-19–specific information. Women were divided into four exposure groups (<0.5 hours, 0.5-2 hours, 2-3 hours, and >3 hours) by their answers regarding the time spent on reading COVID-19 news. Women who spent 0.5-2 hours on reading COVID-19 news per day were set as the reference group in the final analysis. In the multivariable model, we additionally adjusted for other potential risk factors, including age group (<35 years or ≥35 years), educational level (high school or below, or college or above), region (rural or urban), family income (<¥5000 [US $764], ¥5000-¥10,000 [US $764-$1528], or >¥10,000 [US $1528]), history of cesarean section (no or yes), history of preterm birth (no or yes), history of miscarriage (no or yes), first pregnancy (no or yes), prepregnancy BMI (underweight, normal weight, overweight, or obese), smoking (nonsmoker, previous smoker, or current smoker), physical activities (never, sometimes, usually, or every day), inadequate dietary diversity (no or yes), and poor sleep quality (no or yes) by backward methods. To examine the robustness of our findings, we did sensitivity analyses by adjusted covariates in the multivariable models as continuous variables for several variables (age and DDS), instead of categorical variables. In the subgroup analysis, we divided women into different subgroups by baseline characteristics (region, age group, and history of miscarriage). Among these baseline subgroups, we examined the associations between time spent on reading COVID-19 news and the risk of miscarriage after adjusting for other potential risk factors. All the analyses were done with SAS software, version 9.4 (SAS Institute). Two-sided P values less than .05 were regarded as statistically significant.

Among all the 456 pregnant women included, 84.2% (n=384) were younger than 35 years, 54.4% (n=248) were living in an urban area, 9.2% (n=42) had a history of miscarriage, 56.6% (n=258) were having their first pregnancy, 28.1% (n=128) were overweight or obese, 7.9% (n=36) were a previous or current smoker, 18.0% (n=82) did no physical activity, 18.0% (n=82) had inadequate dietary diversity, and 38.2% (n=174) had poor sleep quality. The mean age at baseline was 30.0 (SD 4.2) years.

The mean time spent on reading COVID-19 news was 1.8 (SD 0.9) hours per day. Of the 456 pregnant women, only 23 (5.0%) spent less than 0.5 hours on reading COVID-19 news per day, whereas 247 (54.2%) women spent 0.5-2 hours and 54 (11.8%) women spent >3 hours on reading COVID-19 news per day. Women with excessive media use (>3 hours) were more likely to be previously pregnant (P=.03), have no physical activity (P=.003), have inadequate dietary diversity (P=.03), and have poor sleep quality (P<.001; see Table 1).

Out of 456 pregnant women, 73 had a miscarriage. The prevalence of miscarriage was 16.0% (95% CI 12.6%-19.4%). Compared with women who spent 0.5-2 hours (25/247, 10.1%) on reading COVID-19 news per day, miscarriage prevalence in women who spent <0.5 hours (5/23, 21.7%), 2-3 hours (26/132, 19.7%), and >3 hours (17/54, 31.5%) were higher (P<.001; see Figure 1). Miscarriage prevalence was also higher in pregnant women with poor sleep quality (39/174, 22.4% vs 34/282, 12.1%; P=.003) and a high education level (66/368, 17.9% vs 7/88, 8.0%; P=.02; see Table 2).

We observed a U-shape relationship between media use about COVID-19 and the risk of miscarriage (see Figure 2). In the multivariable model, poor sleep quality (adjusted RR 2.06, 95% CI 1.24-3.44; P=.006), 2-3 hours of media use daily (adjusted RR 1.74, 95% CI 1.02-2.97; P=.04), and >3 hours of media use daily (adjusted RR 2.56, 95% CI 1.43-4.59; P=.002) were associated with miscarriage (see Table 3).

In the sensitivity analysis, the association between excessive media use about COVID-19 and the risk of miscarriage was stable (see Multimedia Appendix 1). In the subgroup analysis, the risk of miscarriage was significantly higher in the <0.5 hours media use group among women living in urban areas (P=.04), who had a history of miscarriage (P=.005), and who had advanced maternal age (P<.001), and was significantly higher in the >3 hours media use group among women living in rural areas (P=.001), who had no history of miscarriage (P<.001), and who were younger (P=.004; see Multimedia Appendix 2).

To our knowledge, this is the first study exploring the status of social media use and lifestyles among pregnant women during the COVID-19 pandemic and assessing their associations with the risk of miscarriage in a prospective cohort study. Our results showed a significant association between excessive media use, unhealthy lifestyle, and the risk of miscarriage in Chinese pregnant women. No previous study has assessed the status of social media use during the COVID-19 pandemic among pregnant women. There were some studies that examined the impact of exposure to COVID-19 information on the metal health status among the nonpregnant population (eg, internet users and factory workers) [18]. Nekliudov et al [15] conducted a cross-sectional online survey in a large Russian population using multiple social media platforms and found that time spent following news on COVID-19 was strongly associated with an increased anxiety. To be specific, compared to less than 30 minutes spent reading COVID-19 news per day, the 1-2 hours group was associated with a 5.46 (95% CI 5.03-5.90) point difference, the 2-3 hours group with a 7.06 (95% CI 6.37-7.74) point difference, and the >3 hours group with an 8.65 (95% CI 7.82-9.47) point difference [15]. Pan et al [19] did a cross-sectional web-based survey of 3035 factory workers at the beginning of work resumption following the COVID-19 outbreak in Shenzhen, China. They found that higher overall information exposure to COVID-19 was associated with higher depression symptoms. Similar with the previous studies, we found that, compared with the 0.5-2 hours media use group, the risk of miscarriage was significantly higher in the 2-3 hours media use group (adjusted RR 1.74) and >3 hours media use group (adjusted RR 2.56). One possible explanation of these findings was that women who spend too much time on social media might be more likely to have unhealthy lifestyles (eg, fewer physical activities, inadequate dietary diversity, and poor sleep quality), which might be related with miscarriage. Another possible explanation of these findings was that pregnant women with exposure to the excessive media information were more likely to have psychological problems (eg, depression and anxiety), which might also increase the risk of miscarriage. Berthelot et al [20] found that women in the COVID-19 pandemic were more likely to present clinically significant levels of depression and anxiety symptoms (odds ratio 1.94) than pre–COVID-19 women in Canada. The underlying mechanism on the relationship between excessive media information and the risk of miscarriage needs to be explored in the future.

It is worth noting that a U-shape relationship between time spent on reading COVID-19 news per day and the risk of miscarriage was found in our study. The risk of miscarriage was significantly higher in the <0.5 hours media use group among women who lived in urban areas, had a history of miscarriage, and had advanced maternal age, and significantly higher in the >3 hours media use group among women who lived in rural areas, had no history of miscarriage, and were younger. Previous history of miscarriage and advanced maternal age are both risk factors for miscarriage in the literature [12]. In a multicenter European study, the risk of miscarriage was found to be higher if the woman was 35 years or older, after adjustment for various factors (eg, reproductive history and country) [12]. Inadequate information on getting essential knowledge might be the explanation for the association between inadequate media use and the risk of miscarriage. In addition, this association was more obvious in populations with other risk factors of miscarriage. Our findings highlight the importance of obtaining moderate COVID-19–related information from social media and that either inadequate (<0.5 hours) or excessive (>3 hours) exposure to COVID-19–related information was not beneficial for the individuals.

In our study, we found that the mean time spent on reading COVID-19 news was 1.8 hours per day for the pregnant women, which was slightly shorter than the general population (2.4 hours) reported in other studies [21]. The worry of radiation from phones or computers among pregnant women might be related to their relatively shorter time spent on social media. Social media has an imperative role in the world that can provide a unified platform for public health communications, comprehensive health care education guidelines, and robust social distancing strategies while still maintaining social connections [21]. Meanwhile, fake news about COVID-19 on social networks could harm public health [22]. For individuals, it is difficult to effectively identify true and false information in the mass media. Just like a “double-edged sword,” social media needs to be used properly to help provide equal access to health care and end discrimination and social stigmatization. We also found that women with excessive media use were more likely to be previously pregnant, have no physical activity, have inadequate dietary diversity, and have poor sleep quality. Moreover, too much time spent on social media might also be related with unhealthy lifestyles, such as few physical activities, inadequate dietary diversity, and poor sleep. Keeping healthy lifestyles is helpful to prevent the occurrence of infectious disease and chronic disease. Our findings provide a clue for the early identification of a potentially high-risk population and miscarriage among pregnant women during the COVID-19 pandemic.

The prospective cohort study design, controlling various risk factors related with miscarriage, and the first insight into the association of social media use with miscarriage are the strengths of this study. However, there are several limitations in this study. First, we did not collect information on the time spent on different kinds of social media (eg, official and unofficial web-based media, newspapers, and magazines) and individual behaviors on COVID-19 prevention. Second, genetic and psychological factors associated with miscarriage were not investigated in this study. The results need to be interpreted with caution. The potential intermediation role of psychological factors on the association between time spent on reading COVID-19 news and miscarriage needs to be explored in further studies. Third, this study was a single-center cohort conducted in China. A multicenter cohort study is needed to verify the findings in this study. Despite these limitations, our findings are helpful to better understand the role of social media and lifestyle on health among pregnant women during the COVID-19 pandemic. The role of media and public health communications needs to be correctly understood and explored further, as they will be an essential tool for delivering information and combating COVID-19 and health promotion, especially for vulnerable populations such as pregnant women.

Pregnant women spent about 2 hours a day reading COVID-19 news in the early stage of the COVID-19 pandemic in China. Pregnant women with excessive media use were more likely to having no physical activity, inadequate dietary diversity, and poor sleep quality. Excessive media use and poor sleep quality were associated with a higher risk of miscarriage. Our findings highlight the importance of healthy lifestyles during the COVID-19 pandemic.