Study design

A cross-sectional survey was conducted on pregnant women from three hospitals in Ya’an City, which was the epicenter of the Lushan earthquake. Since the survey was conducted immediately after the disaster and challenges existed in finding the participants and delivering materials related to the investigation, the researchers did not have the time and resources to identify another group of women for comparison.

Participants

One hundred and thirty pregnant women who experienced the earthquake in the epicenter area were recruited by convenience sampling when they came to the three hospitals for prenatal care. When the earthquake occurred, all major roads into Ya’an were blocked except to vehicles transporting disaster relief materials, and communication between people living in the epicenter area and those outside was affected. Therefore, it was quite difficult to deliver the related information materials to pregnant and postpartum women, and to recruit participants immediately after the earthquake. This made it impossible to obtain a large sample size and to carry out random sampling. On the basis of studies on depression in pregnant women after an earthquake [7–9], it was estimated that a sample of up to 190 would be appropriate [52]. However, as the number of deliveries at the three hospitals constituted over half of the total number of deliveries in Ya’an, this survey covered the majority of pregnant women who were accessible at the time.

Participants for this study needed to meet the following inclusion criteria: 1) those who had been pregnant for over 12 weeks when they experienced the Lushan earthquake; 2) who had lived in Ya’an during and after the earthquake; 3) who had no known history of auditory, language, or cognitive problems, and were able to communicate in Chinese.

In addition, both pregnant and postpartum women were targeted as participants, in accordance with our original study design. Nevertheless, considering the small number (only 15) of postpartum women that we were able to recruit for the study, due to our limited resources immediately after the earthquake, we did not analyze the results of the postpartum women.

Procedures

From 2 to 6 weeks after the Lushan earthquake, three local volunteers who had been trained through telephone by three of the researchers (R.J, Y.J, & L.X) to work on the study instruments, interviewed the participants at the three hospitals in Ya’an. All of the participants were told the purpose of the study and gave their written informed consent to participate in the study before the questionnaires were administered to them. The interviews took place from May 4 to June 1, 2013. Data was collected only once and each instance of data collection lasted about 40 minutes. Code numbers, rather than names, appeared on the questionnaires. After all of the completed questionnaires were collected, two volunteers mailed 86 questionnaires and one volunteer carried 59 questionnaires back to Chengdu for data analysis. The data were entered and stored by one author (R.J) in a password-protected computer.

Measures

The demographic data were collected through a self-designed questionnaire, which included socio-demographic data such as age, gestation age at the time of the earthquake, education, income, and injury and/or death of relatives. The main results of this study were obtained using three questionnaires: the Edinburgh Postnatal Depression Scale (EPDS), the Social Support Questionnaire-Chinese version (SSQ), and the Coping Styles Questionnaire (CSQ).

The EPDS is a self-rating scale with 10 items and was originally designed for assessing postpartum depression [53–54]. However, it had a similar effect on pregnant women, with desirable reliability and validity [19, 55]. For instance, in their study on pregnant women, Lau et al. [19] mentioned that the Cronbach’s Alpha coefficient for the EPDS was 0.76 and that the instrument demonstrated satisfactory concurrent validity with the Beck Depression Inventory and the General Health Questionnaire. The maximum score is 30, and a score of 14 or higher suggests a depressive illness of varying severity. Considering that all of the participants were Chinese, the Chinese version of the EPDS was used so that the participants would be better able to understand the questions [56–57].

The SSQ was designed by Xiao [58], and was proven to have acceptable psychometrics among Chinese university students [59] and farmers [60], with a test-retest reliability of 0.92, and good predictive validity. This questionnaire consists of three dimensions: objective support, subjective support, and usage of support. Objective support is visible or actual support. It includes material aids, and the existence of and participation in social networks. It also includes team relations, defined as the availability of stable marriages, and unsteady social relations such as informal organizations and temporal social interactions. Subjective support is a kind of support that is experienced or felt by oneself. It is defined as the experience of, and the degree of satisfaction felt in, being supported, understood, and respected. Usage of support is the third dimension of social support, and refers to the degree to which the available support is used [58].

The CSQ was revised by Xie [39] to suit the characteristics of Chinese people based on the Ways of Coping Questionnaire. The latter was designed by Folkman and Lazarus [61] and has two dimensions (positive coping and negative coping) with 20 items. Because the survey would be conducted among the Chinese population, and because after the earthquake conditions for conducting a survey were far from ideal, we used the simplified version of the questionnaire. This scale has acceptable psychometrics in studies on earthquake victims in China [62]. The test-retest reliability of the CSQ was 0.89 and the Cronbach’sαcoefficient was 0.90. The factor analysis indicated that the items could differentiate “negative” and “positive” factors and were in line with theoretical construct [62].

Ethical considerations

The study was approved by the ethics committee of the West China Second University Hospital of Sichuan University. Written informed consent was obtained from the participants before the questionnaires were distributed to them.

Data analysis

The cut-off score of the EPDS is 14, on the basis of which the incidence rate of depression was estimated. In a series of analyses of variance (ANOVA), the differences in the EPDS scores were compared by demographic factors. Possible confounders were: 1) demographic factors that are often used as confounders in other studies [63–65] focusing on disasters, e.g., age, income, relatives wounded and dead, etc., and 2) factors that have been identified in other studies [8–9, 19] as related to depression during pregnancy or postpartum depression (e.g., education, number of pregnancies and children, etc.). The Pearson’s correlation coefficient (r) was calculated to indicate the relationship among demographic data, EPDS scores, social support scores, and coping styles scores. In order to determine whether demographical factors, social support, and coping styles are significantly associated with the prenatal depression immediately after a major earthquake, a multiple linear regression analysis was conducted against the EPDS scores. The method of the regression analysis is stepwise, which involves entering the variables in blocks one by one rather than all at once. In addition, an interaction analysis was conducted to check the interaction between the main risk or protective factors, including social support and coping styles, on depression. All of the data were inputted using the software SPSS 20.0 for statistical analysis, and the statistical significance was set at p<0.05.

Demographic features and the incidence rate of depression during pregnancy

Forty-five of the 128 participants had EPDS scores of ≥14 (the EPDS cutoff point); thus the incidence rate of prenatal depression in this study was 35.2%. The incidence rates of prenatal depression in the various studies focusing on pregnant women after earthquakes in China are listed in Table 1. The chi-square test indicated that the incidence rate in the present study is similar to that of the other two studies [7, 9] (p = 0.255).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Incidence rate of prenatal depression in different studies.

https://doi.org/10.1371/journal.pone.0135809.t001

The mean EPDS score was 11.25±5.71. The mean scores for the objective, subjective, and support usage subscale were 8.48±2.79, 24.73±4.54, and 8.05±2.0, respectively. The mean scores for negative and positive coping styles were 24.69±7.23, 34.59±8.20, respectively.

The characteristics of the participants are outlined in Table 2. The mean age of the 128 participants was 26.91±5.46 years. The mean gestation age at the time of the earthquake was 33.19±7.84 weeks.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Demographics, EPDS scores, social support, and coping styles of women who participated in the study (n = 128).

https://doi.org/10.1371/journal.pone.0135809.t002

Multivariable analysis for EPDS scores against demographic data

The multivariable analysis was performed using one-way ANOVA to compare the EPDS scores for different demographic factors by group. EPDS scores were higher in women who were farmers (p = 0.001), had more than two children (p<0.001), were at or beyond their 28 week of gestation at the time of the earthquake (p<0.01), and/or had relatives injured (p<0.001) or killed (p = 0.001) as a result of the earthquake (See Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. ANOVA of EPDS scores against demographic data.

https://doi.org/10.1371/journal.pone.0135809.t003

Correlations among the scores for EPDS, social support, coping styles, and demographic data

EPDS scores were significantly correlated with gestation age at the time of the earthquake, with three dimensions of social support, and with two dimensions of coping styles (Table 4). Higher EPDS scores were significantly related with exposure to the earthquake later in pregnancy (r = 0.181, p<0.05). Moreover, lower objective support (r = -0.324, p<0.001), subjective support (r = -0.372, p<0.001), and support use (r = -0.320, p<0.001) were related to higher EPDS scores in the women. Likewise, higher EPDS scores were related to higher negative coping scores (r = 0.420, p<0.001) but lower positive coping scores (r = -0.328, p<0.001).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Bivariate correlation analysis of scores in EPDS, social support, and coping styles, and of demographic data.

https://doi.org/10.1371/journal.pone.0135809.t004

Although EPDS scores were significantly and inversely correlated with three dimensions of social support, and with two dimensions of coping styles, the correlation coefficients between them were moderate, with the value of r ranging from -0.32 to -0.42 (Table 4).

Apart from EPDS, coping style scores were correlated with certain dimension of social support. Higher negative coping style was related to lower objective support (r = -0.201, p = 0.023), lower subjective support (r = -0.300, p = 0.001), and lower support use (r = -0.226, p = 0.010). However, higher positive coping style was only significantly associated with higher subjective support (r = 0.205, p = 0.020) and higher EPDS scores (r = -0.328, p<0.001) (Table 4).

Regression analysis against EPDS scores

In order to determine whether demographic factors, social support, and coping styles were significantly associated with prenatal depression immediately after a major earthquake, a multivariate linear regression analysis was further conducted against the EPDS scores. Since the number of dead relatives was less than 5, this variable was excluded. The results are shown in Table 5. The stepwise regression screened out five variables that could be put into the regression model. The five variables included number of children (β = 2.189), relatives wounded (β = 3.466), subjective support (β = -0.312), negative coping styles (β = 0.218), and positive coping styles (β = -0.134).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Multivariate linear regression analysis against EPDS scores.

https://doi.org/10.1371/journal.pone.0135809.t005

In the first step of the model, negative coping scores entered and accounted for 17.7% of the variance. Higher negative coping scores were associated with higher EPDS scores in the women. In the second step of the model, having a relative injured because of the earthquake entered into the model and accounted for an additional 11.1% of the EPDS scores. Women who had a relative injured had higher EPDS scores. In the third step, positive coping scores entered and explained another 6.1% of the EPDS scores. Lower positive coping scores were related to the women’s higher EPDS scores. Number of children entered into the model in the fourth step and accounted for 4.7% of the EPDS scores. The women with more than one child had higher EPDS scores than those with only one. In the final step, subjective support scores entered into the model and accounted for extra 5.3% of the EPDS scores. Altogether these five factors accounted for 44.9% of the EPDS scores.

Interaction analysis of risk factors of depression

A correlation analysis (Table 4) showed that social support, coping styles, and EPDS scores were significantly correlated with each other. Although the multiple linear regression analysis found that not all types of social support could predict the variance of EPDS scores after the earthquake (Table 5), it is reasonable to investigate possible interactions between social support and coping styles on depression. In this connection, an interaction analysis was conducted. Since objective social support and support use were excluded from the regression model (Table 5), only subjective social support, which was an associated factor for depression, was used to create interaction terms with two coping styles.

The multiple linear regression analysis (Table 6) of subjective social support scores (centered), negative coping styles (centered), and positive coping styles (centered) in relation to the EPDS scores indicated that the interaction between subjective social support and positive coping styles (centered) had statistical significance. On this basis, a scatter diagram (Fig 1) for positive coping styles with different levels of subjective social support on EPDS scores was produced to illustrate the effects of these variables. Positive coping styles were found to be strongly associated with EPDS scores when subjective social support was high.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Effect of positive coping style on EPDS score as a function of subjective social support.

* “sup.low” means low level of subjective social support. * “sup.mod” means moderate level of subjective social support.

https://doi.org/10.1371/journal.pone.0135809.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Regression analysis of social support (centered), negative coping styles (centered), and positive coping styles (centered) in relation to EPDS scores.

https://doi.org/10.1371/journal.pone.0135809.t006

The incidence rate of prenatal depression immediately after the earthquake

The rate of prenatal depression after the 2013 Lushan earthquake (35.2%) was similar to those results of the 2008 Wenchuan earthquake, which ranged from 34.5% to 40.8% [7, 9] (see Table 1). All these incidence rates were higher than those in the general population of pregnant women (7% to 14%) [10–12]. This implied that there might be somewhat a connection between higher rate of prenatal depression and being exposed to an earthquake. Although the timing of those studies differed, the similar higher rates of prenatal depression in different places after earthquakes indicated that the timing of these incidents might not necessarily affect the progress of such depression and its recovery. This suggests that psychological intervention could be conducted on pregnant women in the early stages following a major earthquake. Most researchers who have attempted to study interventions for earthquake victims have also suggested that interventions should be conducted as soon as possible after the disaster [48, 50–51].

Pregnant women being at higher risk of developing prenatal depression following an earthquake

The results of a one-way ANOVA (Table 3) showed that the EPDS scores of pregnant women differed significantly depending on their occupation, number of children, gestation age at the time of the earthquake, and whether they had any wounded or dead relatives. Our analyses (Tables 4 and 5) showed that, five factors, namely, number of children, having wounded relatives, subjective social support, and coping styles (positive and negative) were significant determinants of prenatal depression immediately after the earthquake. The results of these analyses indicated that women who used higher levels of negative coping, lower levels of positive coping and who reported lower levels of subjective social support, particularly those who farmed, had more than two children, and had wounded relatives were at higher risk of prenatal depression following the earthquake.

The women, who worked in farming, had higher EPDS scores than those in other occupations. In mainland China, especially Sichuan, the major form of farming is terrace farming, which is generally considered particularly backbreaking and unremunerative work. Fewer and fewer people are choosing to stay in the countryside to farm [66–67]. The movement from rural to urban areas disproportionately involves the male population, leaving many females, elderly people, and children in the countryside [67]. As the main laborers in a family, women may assume more farming work after their husbands leave for cities, which may lead to a further deterioration in their psychological health [68]. Another possible reason for higher EPDS scores in farming women is that the delay in communication with their husbands might pose more uncertainty and stress on them. Most husbands of these women were away from their hometown for work, and it took some time before the women could contact their husbands and discuss family events with them. Therefore, following the earthquake, these women may not have an intimate person to share responsibility and provide support. As such, they would have had to make decisions for themselves and their families alone, which may make them feel burdened and stressed [68–70]. However, given the lack of an immediate comparison group, it was difficult to determine in the present study whether those women had higher EPDS scores before or after the earthquake. Nevertheless, it is clear that women whose occupation is farming need more psychological help immediately after the occurrence of a major earthquake.

It was also found from both multivariable analyses and regression analyses that the more children pregnant women had when the earthquake occurred, the higher their EPDS scores. This result may be attributed to their worries about the safety of young children. Young children are usually considered more vulnerable than adults in an earthquake because they are less capable of coping with emergency conditions in a disaster [71]. The findings from the study by Raposa et al. [72] suggest that increased levels of child-related stress could put mothers at risk of developing depression in the future. It is very likely that the worries about the safety of their younger children following the earthquake contributed to the occurrence of their depression. In addition, there was a higher incidence of depression among women with relatives who were wounded as a result of the earthquake. This finding is similar to those of other studies [8, 73]. According to the above analyses, the higher depression scores of mothers with more than one child or with relatives who were wounded might have been derived from the worries about the wellbeing of their families. Arranging better communication between pregnant women and their families immediately after the earthquake and let them know about their families’ well-being could help to alleviate some anxiety.

The EPDS scores were negatively associated with positive coping, and positively associated with negative coping (Table 4); and in the regression analysis the β of positive and negative coping styles were -0.134 and 0.218, respectively (Table 5). It would appear from this study that adopting positive coping styles has a positive impact and adopting negative styles has a negative impact on prenatal depression immediately after an earthquake. According to Lazarus and a number of researchers [34, 74–75], coping is a mediator between stress and health. When the earthquake occurred, the coping styles of women may play a role as a mediator in alleviating or deteriorating the impact of the stressor (earthquake) on health. The results of the current study are consistent with other findings in similar studies on coping during pregnancy. A systematic review suggested that higher depressive symptoms were associated with greater use of avoidance coping styles, which was similar in characterization with negative coping strategies assessed in our study, and low levels of problem-focused coping, which was similar to positive coping used in our study [38].

In the present study, our preliminary correlational analyses (Table 4) revealed that all the three types of social support were significantly associated with prenatal depression immediately after the earthquake. A number of studies found that social support is a risk/protective factor against the development of psychological problems during stress events [76–78]. The results of the present study further substantiate the function of this factor. However, not all of the three types of social support remained as statistically significant predictors of prenatal depression in the regression model. Only subjective social support was found to be a significant predictor of prenatal depression after an earthquake (Table 5). Subjective social support is the experience and degree of satisfaction felt in being supported, understood, and respected [58]. In contrast to objective social support, which represents the actual support that is obtained, subjective social support is more about the feeling of being supported. It is believed that subjective social support is the more important of the two kinds of support because a person’s psychological perception of reality could affect that individual’s behavior and development [58]. A number of researchers have demonstrated in their studies that the women who were more satisfied with support were less prone to depression [79–80]. Although objective social support was related to EPDS scores in the correlation analysis (Table 4), this factor was eventually excluded by the findings in the regression analysis (Table 5).

According to the above analyses, it could be concluded that using higher levels of negative coping, having lower levels of subjective social support, working as a farmer, having more than two children, and having wounded relatives were risk factors for prenatal depression following the earthquake. Regulating these factors immediately after an earthquake may decrease the depression of pregnant women and lead to favorable outcomes for those suffering from prenatal depression. Therefore, the following directions should be considered in the interventions to relieve the psychological problems of pregnant women: better attention and psychological support for farming women; more efficient communication between pregnant women and their families to let them knowing that their families were unhurt; higher quality medical services for wounded relatives; facilitating the feeling of being supported; and directing women to adopt more positive coping strategies and avoid negative strategies. For example, based on operational definitions for positive and negative coping styles [39], interventions could help women adopt actions associated with positive coping styles such as actively talking and listening, seeing events from a favorable perspective, and seeking solutions to problems, and avoid actions associated with negative coping styles such as evading reality, depending on others to solve problems, and performing bodily harm activities.

While the risk factors accounted for 44.9% of the variance in prenatal depression immediately after the earthquake (Table 5), there might be some other important risk factor/s that were not identified in the present study. According to Rich-Edwards et al. [81], history of depression is one of the strongest predictors of prenatal depression in women in general. History of depression may also explain a certain percentage of the variance in prenatal depression in the immediate aftermath of the earthquake. Furthermore, a systematic review on prenatal depression of women in general identified major life events and daily hassles as significant predictors of depression [22]. The occurrence of a major earthquake may cause greater life stress by disrupting daily life which may lead to depression [82–83]. Nevertheless, given that the current study had placed the focus solely on demographic factors, social support and coping styles, additional and possible risk factors cannot be fully understood in relation to prenatal depression during the aftermath of an earthquake. These factors may become the focus for future studies on prenatal depression immediately after earthquakes.

Interaction analysis between social support and coping styles in relation to depression

The regression model (Table 6) and scatter diagram (Fig 1) for the interaction analysis indicated that the interaction term between subjective social support and positive coping styles reached statistical significance to less depression, but the term between social support and negative coping styles did not (Table 6). This result suggests that the effect of positive coping has on depression depended on the degree of subjective social support, so that there was no “unique” main-effect of positive coping. Coping capacities were different for women with varied subjective social support. It has been identified that higher subjective social support and positive coping styles, separately, are associated with lower tendency to experience prenatal depression in the current study (Table 5). However, the five factors including subjective social support and positive coping, could only account for 44.9% of the variance in depression. The emergence of interaction between subjective social support and positive coping may explain part of the variance that was not included in the 44.9%.

According to the interaction analysis, lower EPDS scores were observed when women, with lower subjective social support, adopted more positive coping strategies. This may be due to the situation that women with lower social support tend to adopt negative rather than positive coping strategies [47, 84]. In contrast to women with low social support, women who had high social support have invested more on positive coping styles. Analogous to the law of diminishing marginal effect [85], adopting more positive coping strategies may cause less increase of effect for women with high social support, and greater improvement of effect for those with low social support. This result imply that supportive relationships may impact on prenatal depression by helping women adapt to stress and changes in the aftermath of the earthquake. Clinical healthcare workers should pay more attention to other pregnant women who have negative coping styles and little social support. This also implies that a new intervention can be developed to prevent pregnant women from developing depression, involving by improving their feeling of being supported and helping them to develop positive coping styles in the aftermath of a major earthquake.

The lower correlation coefficients between EPDS scores and social support and coping styles

It is worth mentioning that the correlation coefficients between EPDS scores and social support and coping styles appear to be moderate, with the value of r ranging from 0.32 to 0.42 (Table 4). According to Burnand [86], the strength of the association is moderate when r is > 0.30 and <0.45. It is difficult to explain this phenomenon, given the paucity of the relevant literature. The interaction of subjective social support and positive coping styles was found to have an inverse association with the EPDS scores (Fig 1). There might be some moderating effects between subjective social support and positive coping styles in that they provide essential encounter-actions immediately after a major earthquake to the development of depression during pregnancy. Future studies that target this aspect of the essential variables for deeper investigation are necessary.