Psychiatric morbidity and pregnancy outcome in a disaster area of Taiwan 921 earthquake


address: Hsueh-Ling Chang, Department of Child Psychiatry, Chang-Gung Children's Hospital, Linkou Medical Center, 5 Fu-Hsin Road, Kwei-Shan, Taoyuan, Taiwan. Email:


Abstract We investigated the prevalence of minor psychiatric morbidity in a group of women who were pregnant during or immediately after a major earthquake disaster and we investigated the prognostic factors that may have influenced the perinatal outcome of the pregnancy. The study was initiated 6 months after the earthquake and enrolled 171 women in a town near the epicentre. A Post-Earthquake Questionnaire, Chinese Health Questionnaire (CHQ-12) and posttraumatic stress disorder (PTSD) symptoms checklist were completed before delivery while the perinatal data were retrieved from hospital obstetrical records. The prevalence of minor psychiatric morbidity (MPM) was 29.2%. Women with starvation experience, higher negative attitude scores about the influence of earthquake on pregnancy and more casualties among relatives were significantly correlated with high CHQ. A significant positive correlation between the MPM and PTSD scores was noticed. Among the 115 pregnancies with known perinatal outcome, there were nine (7.8%) low-birth weight neonates, defined as birth weight ≤ 2500 g. Maternal history of abdominal injury, spouse casualty and instability in living condition were significantly correlated with low birth weight. Spouse casualty was the only significant factor that predicts neonatal low birth weight.


At 01:47 am on 21 September 1999, an earthquake measuring 7.3 on the Richter scale struck central Taiwan, killing at least 2400 people and leaving 100 000 people homeless. It was followed by over 10 000 aftershocks during the first month and caused major destruction in a few cities and villages. Pu-Li is a town a few kilometers from the epicentre with a population of 87 000. The 921 earthquake, as it is known, destroyed 40% of Pu-Li, and killed 824 residents.

Studies of the short- and long-term mental health consequences that result from earthquakes have gained wide recognition in the past years. Large-scale earthquakes in Ecuador, Armenia, India, Japan and Taiwan indicate that approximately 25–50% of earthquake victims might develop acute psychological dysfunction and chronic posttraumatic stress disorder (PTSD).1–6 Impacts of earthquakes may also result in physiological changes. Earthquake-induced psychological stress was associated with increased cardiac mortality, and also compromise human Natural Killer Cell activity.7,8

The impact of the earthquake on pregnancy requires closer investigation. However, this has been a difficult area of research, where many variables needed to be controlled. Previous studies suggested an association between antenatal stress and several important birth outcomes: low birth weight, low gestational age and smaller head circumference.9–13 Fukuda et al. reported a low sex ratio (0.501) after Kobe earthquake, indicating that acute stress might have reduced sperm motility.14

The above-mentioned studies focused on outcomes of delivery after major disaster. Possibly as a result of the difficulty of disaster recovery, there has not been any prospective study on the relationship between disasters, the prevalence of minor psychiatric disorder in pregnant women, and birth outcome.

The present study investigated the prevalence of minor psychiatric morbidity, the presentation of PTSD symptoms in pregnant women that were related to the earthquake, and the effect of the earthquake on pregnancy outcome.



The present study commenced 6 months after the earthquake. The largest hospital at Pu-Li, Pu-Li Christian Hospital, was selected. It was the largest regional obstetric care centre and had about 500 births annually before earthquake. Women who attended antenatal care at the hospital from March to August 2000 (6–12 months after the earthquake) and who lived in Pu-Li during and after the 921 earthquake for at least 6 months were asked for their oral informed consent to be included in the present study.

Self-rating questionnaires were given to all subjects at antenatal visit or at prenatal nursing instruction class. Maternal and neonatal data were prospectively collected.


Post-earthquake questionnaire

This questionnaire includes two self-report checklists: Earthquake Exposure Checklist (EEC) and Post-traumatic Stress Reaction Checklist (PTSRC). The EEC consists of 30 questions designed to assess subjective aspects of the earthquake (starvation, attitude toward pregnancy after the earthquake, impact of the earthquake on mood, daily life and physical condition), and life events (casualty among family members, financial loss, dislocation of housing) and personal injury during the earthquake. The PTSRC consists of 20 statements that assess for post- traumatic stress reaction based on Diagnostic and Statistical Manual of Mental Disorders 4th edition (DSM-IV) criteria. The PTSD symptoms are further divided into three subscales according to the DSM IV criteria: re-experiencing, numbing/avoidance and arousal.

Chinese Health Questionnaire

This 12-item version of the Chinese Health Questionnaire (CHQ-12) was designed and developed by Cheng and Williams for the survey of minor psychiatric morbidity.15 It was adapted from General Health Questionnaire (GHQ) for use as a self-administered screening instrument to identify minor psychiatric disorders. The CHQ-12 emphasized more somatic items than the GHQ, as the Chinese population tend to present with depression and minor psychiatric disorders symptoms with more somatic discomfort. The CHQ-12 was validated with the Chinese population in both community and clinical settings.16 It has a weighted classification rate, sensitivity and specificity of 89%, 70% and 95%, respectively.17 A simple scoring method of 0-0-1-1 was applied to the CHQ-12. The optimal cut-off scores of 4/3, described by Chong et al., were applied to the present study.18 The grade of psychopathology was calculated by CHQ global scoring.

Sociodemographic information

Sociodemographic information including the subject’s age, marital status, parity, weight and height, education level and occupation of subject and spouse, and annual family income were obtained from a personal information questionnaire designed for the present study. The socioeconomic status was classified according to Hollingshed’s Two Factor Index of Social Position.19

Perinatal data

Perinatal information including delivery date, birth weight, body length and Apgar scores of 1 min and 5 min after delivery for each newborn were retrieved from the birth record at delivery room. Low birth weight was defined as birth weight ≤ 2500 g.

Statistical analysis

Data analyses were executed using the statistical software package system (SPSS for Windows, 8.0.1; SPSS, Chicago, IL, USA). The Student’s t-test or Pearson’s χ2 test was used to assess the association between CHQ scores and variables including psychosocial factors, PTSD subscales, life events during the earthquake, living condition, and subjective feeling toward influence of earthquake if appropriate. Variables found to be significant were put into a logistic regression model to disclose any independent prognostic factor of delivering a low birth weight neonate after adjusting maternal age, parity, and socioeconomic status. A two-sided P-value of less than 0.05 was declared as statistical significant.


A total of 192 pregnant women enrolled in the present study. Twenty-one women, most were foreign brides, were excluded at first interview because they were unable to understand the questionnaire. A total of 171 pregnant women completed the questionnaires. Fifty-six of these 171 women had their babies delivered somewhere other than at the Pu-Li Christian Hospital and their perinatal data could not be obtained.

There was no significant difference between these 56 lost to follow-up women and the 115 study women in terms of age, socioeconomic status and PTSD scores. However, the lost to follow-up group had a significantly higher CHQ-12 score.

Basic characteristics

All 171 women were married and had received at least 6 years of education with a mean education of 9.3. The mean and median age of the 115 women who were studied were 27.4 years and 27 years (range 15–43 years), respectively. A total of 4.6% of the studied women were aged over 35 years and 2.3% aged under 18 years.

Mental health

Of the 171 women, 50 (29.2%) scored higher than 3 in the CHQ-12, and were identified as suffering from a minor psychiatric disease (CHQ (+)). Analyses on the association between psychosocial factors and CHQ and between life event during earthquake and CHQ showed that a woman with starvation experience showed a significant correlation with high CHQ scores (P = 0.001). Mothers with higher negative attitude scores about the influence of earthquake on pregnancy or casualties among relatives also had a significant correlation with high CHQ scores (P = 0.004 and P = 0.005, respectively).

Table 1 shows the basic characteristics of CHQ (+) and CHQ (–) cases. CHQ (+) cases had more casualties among relatives, more starvation experience and higher negative attitude scores to the influence of the earthquake on their pregnancies. There is no difference in age, socioeconomic status, gravida, parity, maternal height and maternal weight between the CHQ (+) and CHQ (–) cases (Table 1).

Table 1.  Basic characteristics of study subjects
 CHQ (+)CHQ (−)
 n = 50n = 121
 No. (%)No. (%)P
Age (year)28.04 ± 5.527.16 ± 4.70.32
Maternal body weight (kg)54.15 ± 8.653.67 ± 8.70.23
Maternal body height (cm)158.1 ± 4.9157.6 ± 5.30.25
Gravida2.30 ± 0.952.11 ± 1.100.46
Parity0.97 ± 0.790.93 ± 0.970.53
Self injured3 (6)5 (4.1)0.001
Death among relatives4 (8)0 (0)0.02
Starvation during pregnancy13 (26)6 (4.9)0.001
Negative attitude to influence of earthquake on pregnancy30 (60)50 (41.3)0.04
Dislocation46 (92)109 (89)0.37

Impact of the earthquake

The impact of the earthquake on the studied women was divided into three categories: impact on mood, impact on physical condition, and impact on daily life. A 4-point rating scale (mild, moderate, severe and profound) was applied. A total of 9.3% of women rated ‘profound’ in ‘impact on mood’ and ‘impact on daily function’, while 0.6% of women rated ‘profound’ in ‘impact on physical condition’. There was significant difference in the distribution of all the three categories between CHQ (+) and CHQ (–) individuals (Table 2).

Table 2.  Impact of earthquake on mood, physical condition and daily life between Chinese Health Questionnaire (CHQ) groups
 CHQ (+)CHQ (–)Comparison
Impact of
earthquake on
n = 50n = 121P
Physical condition
Daily life

CHQ (+) cases scored significantly higher than CHQ (–) cases in all PTSD subscales, and the total PTSD score (Table 3).

Table 3.  PTSD subscales in Chinese Health Questionnaire CHQ (+) and CHQ (–) groups
 CHQ (+)CHQ (–)
 n = 50n = 121P
Re-experiencing9.24 ± 5.416.44 ± 3.880.001
Numbing/9.68 ± 8.096.88 ± 3.820.003
Arousal3.06 ± 1.712.10 ± 1.270.000

Birth outcomes

Among the 115 pregnancies with known perinatal outcome, there were nine (7.8%) low birth weight neonates.

Univariate analysis showed significant associations between low neonatal birth weight and maternal abdominal injury or spouse casualty during the earthquake or instability in living condition (Table 3). Logistic regression after adjusting maternal age, parity, and socioeconomic status revealed that spouse casualty was the only significantly factor that predicts neonatal low birth weight (Table 4).

Table 4.  Intercorrelation between maternal body height, abdominal injury experience, spouse casualty and relatives and dislocation in housing towards neonatal low birth weight. (n = 40)
  • *

    P< 0.05;

  • **

    P< 0.01;

  • ***

    P< 0.001.

1. Low birth weight 
2. Body height− 0.26** 
3. Abdominal injury0.32**− 0.35 
4. Spouse casualty0.32**− 0.190.32***
5. Injury of relatives0.31*− 0.130.32***0.49*** 
6. Dislocation0.25**


The prevalence of minor psychiatric morbidity found in the present prospective study was significant, and supports previous studies which stated that increased stresses may be associated with higher rates of psychiatric morbidity. This is the first study, however, that focused on post-disaster psychiatric morbidity in pregnant women.

It was observed in the present study that CHQ (+) cases differ significantly with CHQ (–) cases in terms of PTSD scores, starvation experience, negative attitude towards the earthquake’s influence on pregnancy and casualties among relatives. This may represent a subpopulation in need of special care in the study population.

The present finding is in line with other studies conducted after the 921 earthquake in Taiwan. Several surveys in different disaster areas revealed high exhibition of PTSD symptoms among earthquake victims. In general, female sex, preschool children and elderly over 65 years, higher exposure to various types of stress, and more difficulties in maintaining resources were related to higher psychiatric morbidity or poorer mental health.20–22

In the present study, we noted the persistence of post-traumatic stress symptoms 6 months after the earthquake. Both CHQ (+) and CHQ (–) subjects scored highest on ‘numbing/avoidance’ subscales, and lowest on ‘arousal’. In another study on earthquake victims, Goenjian et al. postulated that elderly victims scored highest on ‘arousal’ because of increase vulnerability to dysregulation due to age, and younger victims scored highest on ‘reexperiencing’ because of more participation in rescue efforts.2 The present finding also showed different evidence from another study conducted 3 weeks after 921 earthquake, which noted more ‘arousal’-related symptoms than ‘numbing/avoidance’ symptoms.6 A longer time lag between the disaster and time of investigation may result in the difference between PTSD symptom presentations. Further investigation is needed to understand the role of gender, pregnancy and culture in PTSD symptoms presentation.

The previous prevalence of MPM in Taiwan ranges from 11.5% to 21% in the general population.23 A study in Nigeria using the General Health Questionnaire (GHQ-30) in pregnant women revealed a 12.5% prevalence of psychiatric disorder.24 As for post-disaster studies, Lima et al. found that 40% earthquake victims were still distressed 3 months after the event.25 Araki et al. found that 67% of volcanic eruption evacuees suffered from psychological trauma;26 and Sharan et al. noted 59% earthquake victims were receiving a psychiatric diagnosis at interview.3 The present study used different measures to evaluate the victim’s mental health, making it difficult to compare the present results with the above-mentioned studies.

The psychiatric morbidity in the present study is higher than in other Taiwanese studies using similar measures, but the difference was not overwhelming. The fact that the present study was conducted from an antenatal clinic, which provided good early support for the pregnant victims in a post-disaster period, and the exclusion of lost to follow-up women with higher CHQ scores, could explain the relatively lower prevalence of MPM.

The prevalence of low birth weight was 7.8%. Previous surveys in Taiwan reported prevalences of 2.8–6.2%.27,28 In the present study, low birth weight is associated with maternal history of abdominal injury, spouse casualty and instability in living conditions. However, after adjusting for maternal age, parity, and socioeconomic status, spouse casualty was the only significant factor that predicted neonatal low birth weight. The stresses and hazards resulting from spouse casualty and the significance of spouse support in the Chinese culture needs deeper investigation.

There are a number of limitations to the present study. The optimal threshold score to determine the prevalence of minor psychiatric morbidity form of the CHQ-12 has not been established for pregnant women. Although the CHQ-12 was well accepted and easy to administer, the interpretation of the results for prevalence estimates is not straightforward unless and an optimal cutoff score is established for the specific population. Two previous studies using a cutoff point of 3/4 for the same questionnaire in Chinese population, revealed a prevalence of 11.5% and 19%.16,29 These two studies focused on the general population, but they did mention a female predominance in prevalence. No cutoff scores have been published for disaster victims. It is very likely that the prevalence of minor psychiatric morbidity is much higher than in the general population.

The present study can no doubt be improved with matching control samples of low-exposure pregnant women from similarly developed areas and measures of pre-existing psychiatric disorder, family history and other pre-existing vulnerability to stresses.

The contribution of the present study may be, among the sparse literature on the impact of natural disaster, to provide an important clue for the relationship between pregnancy in disaster and birth outcome. The key psychosocial variables identified are likely to increase CHQ scores, and low birth weight may prove useful in screening and helping future pregnant victims in natural disaster.


The authors would like to express their sincerest respect to 921 survivors for their efforts and courage to recover from the disaster. We would also like to thank the many people who offered their help to make this study possible. Special thanks go to Lu-Yi Chou and the staff of Pu-Li Christian and Chang-Gung Hospitals.