Participants were recruited from two maternity and childcare centers in the cities of Changsha and Yiyang in Hunan province, China. We recruited women in the obstetrics clinics of these two centers from September 2016 to February 2017. The following inclusion criteria were applied: women who were pregnant, aged ≥18 years, and had a gestation period of ≤13 weeks (pregnancy weeks were estimated based on the first day of the last menstrual period).

The study was approved by the institutional review board of the Institute of Clinical Pharmacology of Central South University (ChiCTR-ROC-16009255). A total of 1126 women were recruited. In total, 15,647 WeChat Moments images and 6609 posts from 419 mothers, 123 of whom had postpartum depression, were used in this study. The participant recruitment and data collection process are depicted in Figure 1. The use of all the data from WeChat Moments in this study was authorized by the users.

A number of different tools were used to collect data. We used questionnaires to measure the participants’ perinatal depression levels and other psychological factors from the first trimester to 6weeks postpartum and collected the mothers’ WeChat Moments data from childbirth to 6 weeks postpartum. The questionnaire was designed using simple items to investigate the participants’ general demographic characteristics including age, first pregnancy, monthly income, education, and history of depression. A token gift of 10 RMB (US $1.5) was given to each participant who completed a questionnaire.

We used the Pittsburgh Sleep Quality Index [22] to measure the sleep quality of the participants during the postpartum period. In this index, sleep quality is divided into different grades based on the total score, where 0-5 indicates very good sleep quality, 6-10 indicates good sleep quality, 11-15 indicates average sleep quality, and 16-21 indicates poor sleep quality.

We used the 7-item Generalized Anxiety Disorder Scale [23] to measure the anxiety of the participants during the postpartum period. A total score of 10 or higher on this scale is indicative of anxiety.

We collected the images that mothers had posted on WeChat Moments from childbirth to 6 weeks postpartum using the WebCrawler document assembler. The privacy settings allowing image visibility to online friends in WeChat Moments include “all,” “last six months,” “month,” and “three days” (friends will not be able to see the WeChat Moments earlier than the selected timeframe), so images were collected every 2 days.

The Edinburgh Postnatal Depression Scale (EPDS) [24,25] was used to screen for postpartum depression. The EPDS is a 10-item self-rated questionnaire. Each item is scored from 0 to 3, with a total score ranging from 0 to 30. The Chinese language EPDS used in this study was translated by Wang Yuqiong [26]. The critical value was 9.5.

For image analysis, the following 3 errors were preprocessed and removed: (1) images that returned link error messages or manifested incompletely, (2) images with a screen resolution less than 800x600, and (3) black images. The remaining eligible images were converted to readable formats such as JPG and PNG.

Several different types of information were automatically extracted from the collected WeChat Moments data. Information categories included the following: post coloring, post volume and frequency, and the time posts were made.

Different features of mothers with or without postpartum depression were analyzed using the chi-square test and t test. While controlling for the factors associated with postpartum depression, a binary logistic regression was conducted to explore the association between the image features and postpartum depression using the image features as the independent variables and with or without postpartum depression as the dependent variables. The statistical analyses were performed using SPSS (version 25; IBM). The classification of images in this study is based on previous studies [14-16,27-29].

We collected a total of 15,647 images and 6609 posts from 419 mothers on WeChat Moments. The demographic characteristics of the sample are shown in Figure 2. The average participant age was 27.6 years (SD 3.8 years). Participants predominantly had a monthly income of 2000-5000 yuan (US $282-705) and had completed undergraduate or college education. The average hue, saturation, and value of the images was 33.9 (SD 10.6), 0.3 (SD 0.1), and 0.6 (SD 0.1), respectively. During the postpartum period, mothers posted an average of 0.89 (SD 5.07) images per day. The incidence of postpartum depression was 29.4% (123/419). Sample comparison analyses were conducted to compare the demographic features of mothers with postpartum depression and the mothers without postpartum depression (Figure 2).

We used logistic regression to explore the relationship between the image features and the risk of 6-week postpartum depression (Table 1). In this model (r² = 0.20), selfies showed statistical significance after controlling for potential confounders. The factors associated with postpartum depression were included in the model as potential confounders. Women who posted selfies in the postpartum period were more likely to have postpartum depression (P=.003; odds ratio 2.27, 95% CI 1.33-3.87).

We put the variables of advanced age, first pregnancy, monthly income, education, depression history, anxiety, sleep quality, frequency of posts, late-night posts, colors, facial expressions, selfies, pictures of their children, others, and memes into the logistic regression model and presented the meaningful results in Table 1 (α=.05).

We found that, during the postpartum period, the mothers in this study tended to post pictures of their children and that the majority of the “people” in the collected pictures were categorized as experiencing positive emotions. However, the mothers who posted selfies during the postpartum period were more likely to have postpartum depression. Therefore, images posted by new mothers on WeChat Moments could be predictive of the mothers’ mental health during the postpartum period.

Pictures of their children accounted for the largest proportion of images posted by the mothers (3909/6887, 56.8%), and 77.1% (323/419) of the mothers posted pictures of their children to WeChat Moments. This result is similar to that of the Bartholomew et al [30] study in which the proportion was 78.6%. Maternal identity confirmation may be the main reason that mothers share pictures of their children during the postpartum period. Johnson [31] suggested that social media was the main social platform mothers used to announce their pregnancies and the births of their babies. Another important reason mothers post pictures of their children may be that they need to strengthen ties with family and friends so that they can obtain more support after childbirth. The Gameiro et al [32] study found that mothers believed they received increased support from their nuclear family and friends in the pre- to postpartum period. Given that adults are more likely to give attention to babies’ faces, many mothers who uploaded photos of their children to social media thought it was “likely” that the photos would be acknowledged (ie, commented on or “liked”) by their social network friends [30]. It is possible that the majority of mothers consider posting pictures of their children as a particularly compelling way to engage with their online friends [30,33]. Accordingly, WeChat Moments may provide mothers with a platform through which to maintain strong ties with, for example, family and close friends. Meanwhile, sharing pictures of their children on WeChat Moments could be a particularly important means for mothers to identify themselves, build and maintain social capital, and find social support.

There were more positive emotions in the facial expressions of the people in the pictures and memes in this study than negative ones. In Chinese tradition, giving birth is a happy life event that should be celebrated by the whole family. Although we assumed that posting on WeChat Moments would reflect the feelings of the mothers in this study more accurately than posts made on other social media due to the WeChat Moment privacy settings, an element of social desirability response bias may have affected the expression of positive emotions. Pressures to portray one’s best self may lead people to display deceptive versions of themselves on social media [28,34]. Social media users tend to present the happiest and most ideal sides of themselves even if these versions do not align with their actual emotions or experiences [35,36]. The findings of this study suggest that even if a new mother shows positive emotions on WeChat Moments, the risk of postpartum depression cannot be ruled out.

The posting of selfies on WeChat Moments seems to indicate the need to pay more attention to possible depressive symptoms in the users, as the women in our study who posted selfies during the postpartum period were more likely to have postpartum depression. This supports the findings of previous studies, which have shown a positive association between selfies and depression [16,28]. One potential explanation for the trend is that mothers posting selfies intend to show an idealized version of themselves, which may cause them to focus on perceived self-deficiencies, thus leading to depression [16,28]. In addition, a study found that women who posted selfies felt more anxious, less confident, and less physically attractive after posting, indicating that selfies may have harmful effects on mental health [16]. Posting selfies may be indicative of mothers’ concerns regarding others’ opinions of their images as well as their desire for more social attention [27,28]. Posting selfies may be one of the things that mothers can do during the “doing the month” period when their social contacts may be weaker. However, excessive concern for others’ opinions, especially those of close friends and family, could negatively impact women’s happiness and increase their risk of postpartum depression [37].

This study had several limitations. First, the convenience sampling method used in this study limited the extrapolation of the results. Second, WeChat Moments allows users to upload pre-processed images, and some women were accustomed to using filters to embellish their images. Our inability to remove the effects of filters on the original images may have led to some bias. However, we believe that filter tones may have reflected some of the users’ emotional states [14]. Third, WeChat has a feature to hide certain posts from certain people, so we cannot be confident that we collected all the mothers’ images on WeChat Moments, even though we asked for all posts to be open to us when we requested informed consent. Fourth, we collected images posted before the EPDS measurement. However, the EPDS measurement reflects symptoms over the 7-day period before the EPDS questionnaire is administered. This means that we could not be certain about the order or cause-and-effect relationship between the images posted and the presence of postpartum depressive symptoms. Longitudinal research is needed to detect whether posting selfies during the postpartum period has a detrimental effect on mothers’ mental health. Lastly, we analyzed the relationship between the mothers’ posted images and postpartum depression but ignored the text accompanying the posts. The text could have helped us understand the content of the images better. We believe that a combined analysis of both text and images could provide more information, and this could be a direction for future research.

WeChat Moments is an important social media platform for Chinese mothers to share pictures of their newborns. Although mothers tend to post images featuring positive emotions, these pictures may actually mask depression. Particular attention should be given to new mothers who have posted selfies, as they seemed to be at higher risk for postpartum depression.