Bulletin of the World Health Organization

Effect of having a subsequent child on the mental health of women who lost a child in the 2008 Sichuan earthquake: a cross-sectional study

Yao Xu a, Helen Herrman b, Rebecca Bentley a, Atsuro Tsutsumi c & Jane Fisher d

a. Centre for Women’s Health, Gender and Society, Melbourne School of Population and Global Health, University of Melbourne, Level 3, 207 Bouverie Street, Carlton, Victoria, 3010, Australia.
b. Orygen Youth Health Research Centre, University of Melbourne, Melbourne, Australia.
c. International Institute for Global Health, United Nations University, Kuala Lumpur, Malaysia.
d. The Jean Hailes Research Unit, Monash University, Melbourne, Australia.

Correspondence to Yao Xu (e-mail: xuy@student.unimelb.edu.au).

(Submitted: 16 May 2013 – Revised version received: 13 December 2013 – Accepted: 02 January 2014 – Published online: 17 March 2014.)

Bulletin of the World Health Organization 2014;92:348-355. doi: http://dx.doi.org/10.2471/BLT.13.124677


In most natural disasters, one third to one half of the casualties are children.1 Direct exposure to natural disasters and the loss of a child are extremely stressful life events with long-term health consequences. Among these, increased vulnerability to psychological problems such as depression, anxiety, post-traumatic stress disorder (PTSD) and complicated grief (CG) are specially common.2 Parents who have lost young or adolescent children are at higher risk for mental disorders than those who have lost adult children.3,4 Furthermore, bereaved mothers appear to be more vulnerable to psychological illness than bereaved fathers.5

Several factors influencing emotional recovery from the loss of a child after a disaster have been identified. First, professional support, including a structured psychological intervention, appears to be effective in reducing psychological morbidity2 and can improve physical and psychological health and social adjustment after traumatic experiences.68 Second, social support from spouses or one’s parents, friends and colleagues is also linked to better post-disaster mental health status among bereaved parents.2,6 Third, whether or not parents viewed their child’s body appears to influence recovery. According to the limited amount of available research, parents who viewed their children’s bodies recovered better than those who did not on account of the loss or disfigurement of the body.5,7,9 Finally, parents who lost a child but who have at least one surviving child have lower mortality rates10 and a lower prevalence of psychiatric illness than those who have lost an only child.1113

Some experts have postulated that having a new baby might favourably influence parents’ psychological well-being after losing a child. In some cases, having another child has been positively associated with decreased rates of depression in bereaved parents;14 in others this has not been the case. In an 18-year study of parents who had lost one of their children, no significant association was found between having a subsequent child and recovery from grief or depression.15 There are currently no systematic population-based studies investigating the psychological effect of having a subsequent baby after losing a child in a natural disaster.

The aim of this study was to compare the psychological state of women who had and had not given birth to a baby after losing a child in the 2008 Sichuan earthquake in China. Most of the children who died were only children, since China has a one-child policy.16 Three months after the earthquake, the National Health and Family Planning Commission announced the establishment of reproductive health services to help bereaved families to have a subsequent child. The services included free psychological counselling, fertility assessment, cessation of long-term contraception, recanalization surgery, treatment of reproductive tract infections and treatment of infertility with assisted reproductive technologies.17


A community-based cross-sectional survey was conducted in Dujiangyan, China, 29 to 34 months after the 2008 Sichuan earthquake. Data were collected from October 2010 to March 2011. Dujiangyan is a city in the province of Sichuan. It is situated 50 kilometres south-west of Wenchuan, the epicentre of the earthquake, which killed nearly 1000 children in Dujiangyan. The project was conducted in collaboration with the Dujiangyan Family Planning and Human Development Bureau (hereafter the Bureau), which provided a list of women who had lost a child in the earthquake in Dujiangyan.

Conceptual model and design

Since the determinants of psychological health are known to be multifactorial, we tested a conceptual model that took into account the potential confounding effects of age, educational level, employment status, psychological treatment, social support and whether or not the body of the deceased child had been viewed. (Fig. 1).

Fig. 1. Model used to assess the relationship between the mental health of bereaved mothers and having had a subsequent child
Fig. 1. Model used to assess the relationship between the mental health of bereaved mothers and having had a subsequent child
CES-D, Centre for Epidemiological Studies Depression Scale; ICG, Inventory of Complicated Grief; MSPSS, Multidimensional Scale of Perceived Social Support; PCL-S, PTSD Checklist-Specific; PTSD, post-traumatic stress disorder; SAS, Zung’s Self-Rating Anxiety Scale.

Participants and sample

Adult women of reproductive age living in Dujiangyan who had been exposed directly to the earthquake and had lost a child 18 years of age or younger in the 2008 Sichuan earthquake were eligible for participation. Women were excluded if they were unable to read or speak Chinese and if their new child was under 6 months of age. To detect a difference of 20% between groups at a significance level of 0.05 and a power of 0.80 (two-sided statistical significance test), at least 93 participants were required in each group (a total sample of 186 women).


The Bureau sent invitation letters with response forms to each woman meeting the inclusion criteria. On the response form, women were invited to endorse one of three choices: (i) interest in participation; (ii) no interest in participation; (iii) a request for further information about the study. Women who asked for more information about the study were contacted by the Bureau. Women who had not returned the response form three weeks after it was mailed were contacted by telephone once to ask whether they had received the letter or whether they wanted more information.

Data were collected during individual structured face-to-face interviews by a native Chinese-speaking female medical graduate. Interviews lasted up to 90 minutes, either at the research office or the participant’s home. No formal psychiatric diagnoses were made, since the interviewer was not a trained mental health clinician.

Data sources

The structured interview schedule included study-specific questions and standardized measures of psychological well-being in three sections: (i) sociodemographic information (age, education, employment and marital status); (ii) post-disaster experiences (whether or not the child’s body had been viewed, whether or not a psychological intervention or social support had been received); and (iii) mental health (anxiety, depression, PTSD and CG).

Five widely used standardized scales were incorporated, including four for the mental health outcomes and one appraising social support (Table 1).18,20,22,24,27 All scales, translated from English into Chinese, have been used in China and have shown high internal consistency (Table 1).19,21,23,25,26,28

Data management and analysis

All data analyses were conducted in Stata 11.0 for Mac (StataCorp. LP, College Station, United States of America). Descriptive statistics comparing sociodemographic characteristics, post-disaster experiences and social and professional supports for the groups with and without a subsequent child were calculated and statistically significant differences between the two groups were estimated by using the χ2 test. Logistic regression was used to investigate the association between the presence or absence of a subsequent child (with the presence of a subsequent child as the reference category) and mental health. Confounder selection was based on our evidence-informed, theoretical conceptual model (Fig. 1).


Permission to conduct the study was obtained from the University of Melbourne’s Human Research Ethics Committee (HREC No.1033862) and the Human Research Ethics Committee of the Chengdu Fourth People’s Hospital (Sichuan, China).


Overall, 509 (81.4%) of the 625 eligible women returned the response forms. Of these women, 415 agreed to participate and 94 refused. In total, 18 women (12 with a subsequent child) were not able to complete the interviews. Four chose to complete the interview in two or three stages but were too busy to return to finish it; five found it too difficult to answer the questions in the mental health section; nine could not answer questions about the quality of support because their husband or another relative remained in the room during the interview. In total, 110 bereaved women with a subsequent child and 116 without a subsequent child provided complete data (Fig. 2).

Fig. 2. Recruitment of bereaved mothers after the 2008 Sichuan earthquake and response rates, China, October 2010 to March 2011
Fig. 2. Recruitment of bereaved mothers after the 2008 Sichuan earthquake and response rates, China, October 2010 to March 2011

Sociodemographic characteristics

Most participants were married, performed unpaid household or voluntary work and had lived in Dujiangyan for at least 10 years. All except one woman had lost an only child. Except for age, the sociodemographic characteristics of the two groups did not differ significantly (Table 2). Although the mean difference was only one year, women without a subsequent child were significantly older than those with a subsequent child.

Post-disaster experiences

No significant difference was found between the two groups of women in terms of having viewed or not viewed the bodies of their deceased children (Table 2). However, the two groups did differ significantly in both the professional and the social support they had received since the earthquake. A substantially higher proportion of women without a subsequent child had received a psychological intervention (Table 2), but a significantly lower proportion had received social support from family members and friends (Table 3).

Mental health status

Overall, more than 80% of the participants had clinically significant symptoms of at least one psychological condition. Depression, PTSD and CG were more common than anxiety. Women who had not given birth to a subsequent child were consistently found to have higher symptom scores and higher odds of being symptomatic than those who had given birth to another child. A substantially higher proportion of women without a subsequent child had clinically significant symptoms of all four of the mental conditions examined in this paper than those with another child. In particular, over 90% of women without a subsequent child had clinically significant symptoms of depression or CG (Table 4).

Table 5 shows the association between mental health and the presence or absence of a subsequent child in unadjusted and adjusted models. Mothers who had not given birth to a new child had significantly higher odds of having psychological symptoms than those who had. The largest difference between the two groups was found with respect to CG symptoms. Women who did not have a subsequent child had more than 10 times the odds of having such symptoms than women who had given birth to another child (P = 0.008).


To our knowledge, this is the first population-based systematic investigation of the effect of having or not having a subsequent child on the long-term psychological health of women who have lost a child in a natural disaster. By exploring the effects of confounding factors identified in previous research2 as potential determinants of the long-term mental health of women in this situation, we were able to estimate the association between having a subsequent child after losing one in a natural disaster and recovery after bereavement. These data suggest that, for women who have lost their only child, the birth of a subsequent child is associated with milder psychological morbidity, especially depression and prolonged grief disorder.

We acknowledge that this study has limitations. First, this was a cross-sectional survey and cause and effect associations cannot be inferred from the findings. Therefore, we cannot conclude that having another child after losing one in a disaster will help women to recover psychologically. Second, we were unable to control for other types of traumatic events caused by the earthquake, such as losing one’s parents, house or livelihood, so separating their effects on mental health from those of the loss of a child was not possible. However, the modest resources we had for this project led us to focus on comparing mental health status in women who had and had not given birth to a new child after the earthquake. Finally, all the psychometric instruments we used are based on self-reporting and yield scores that are indicative of clinically significant symptoms but are not diagnostic. Although these instruments have been widely used in China29, none has been formally validated against a gold standard and no local sensitivity and specificity data have been generated. All these three factors could have resulted in an over or underestimation of the association between having a subsequent child and women’s mental health. However, since ours was a systematically recruited population-based sample, we have no reason to believe that the prevalence of these conditions was higher in the sample than in the general population. In fact, a previous study has shown that seven to eight months after the 2008 Sichuan earthquake, 80% of bereaved parents had clinically significant symptoms of PTSD and 81.8% had symptoms of depression.3 These data are consistent with our findings.

Curiously, women without a subsequent child had more severe psychological problems, on average, than women who went on to have another child, even though more women in the former group had received psychological treatment. This finding is in contrast to the results of a study by Li et al., whereby bereaved parents who had been exposed to professional psychological intervention were found to have better mental health status than those who had not.12 One possible explanation lies in the type of intervention received, which we were unable to ascertain. It may have been ineffective because of its content or because it was not long enough. However, it is also possible that women who longed to conceive another child were too distressed over this to find relief in mental health interventions. The Chinese central government changed the family planning policy immediately after the earthquake, particularly for women who had lost a child, and provided free comprehensive reproductive health services to couples wanting to conceive. This departure from the one-child policy appears to have benefited many couples but may have failed to meet the reproductive needs of women whose age made it difficult to conceive, even with assisted reproductive technologies. Furthermore, in China bloodlines are centrally important when it comes to maintaining cultural traditions. Families without a child are viewed as incomplete and unfortunate. In China’s male-dominated culture, women are often blamed for the couple’s infertility. Because all but one of the deceased children in this study were only children, their mothers may have felt pressured by their families to give birth to another child. Women who did not have another child may have experienced greater pressure than other women.

This study suggests that women who have decreased fertility by virtue of their age and who have lost their only child in a natural disaster are especially vulnerable to long-term psychological problems. It is clear that much more evidence is required in this relatively under-researched field. Future research should focus on the development and systematic evaluation of interventions designed to assist families in the circumstances described. Social and family support is clearly important to women’s post-disaster recovery.2 The data presented here indicate that communities need guidance on how to strengthen social support activities, in addition to mental health care.

Although women who had another child had lower odds of experiencing psychological symptoms, at least two out of three such women had clinically important symptoms of depression and CG. Thus, most of the children born to women in this category were being cared for by mothers with poor mental health. Research on the development of children born in these circumstances is needed. The data suggest that in bereaved women who are pregnant or have recently given birth to another child, perinatal mental health care is essential to reduce psychological morbidity, strengthen the mother–infant relationship and facilitate optimal infant development.


We thank the Dujiangyan Family Planning and Human Development Bureau and the Dujiangyan Mental Health Centre in Sichuan Province, China, for their valuable support. We also thank the participants who shared their experiences.


YX received funding from the Population Health Investing in Research Students’ Training (PHIRST) and the Harold Mitchell Foundation Travel Fellowship/Scholarship.

Competing interests:

None declared.