Health impact of disaster-related stress on pregnant women living in the affected area of the Noto Peninsula earthquake in Japan

Authors


*Yuri Hibino, PhD, Department of Environmental and Preventive Medicine, Graduate School of Medical Science, Kanazawa University 13-1 Takaramachi, Kanazawa, Japan. Email: hibino@pub.m.kanazawa-u.ac.jp

Abstract

Aims:  The present study assessed the health impact of stress on women who were pregnant during, or immediately after, a major earthquake and were living in the disaster area. Inherent resistance against the stress induced by the earthquake was also assessed.

Methods:  The panel study consisted of 99 women who provided responses before and after delivery (response rate, 77.9%). Psychological impact was assessed on the Edinburgh Postnatal Depression Scale (EPDS), and stress resistance was assessed on the Sense of Coherence Scale (SOC).

Results:  In adjusted multivariate models, the significant earthquake factor that predicted postnatal depression (EPDS) was ‘existing anxiety about an earthquake’ (β = 0.27, P = 0.01) and ‘parity’ (β = −0.26, P = 0.02). The SOC during pregnancy significantly moderated between ‘existing anxiety about an earthquake’ and ‘EPDS’ (β = −0.21, P = 0.02). During pregnancy the EDPS was a significant predictor of a physical abnormality during pregnancy or childbirth (odds ratio, 1.21; 95% confidence interval: 1.04–1.41). The SOC during pregnancy did not moderate between a physical abnormality and earthquake-related stress.

Conclusions:  Provision of an adequate support system and improvement of the SOC of young women affected by a disaster may be two ways of reducing the deleterious effects of disaster-related stress on maternal well-being.

AN EARTHQUAKE MEASURING 6.9 on the Richter scale struck the Noto Peninsula of Japan at 09.42 hours on 25 March 2007. The earthquake was followed by the sustained occurrence of >500 aftershocks. The Noto Peninsula earthquake severely damaged the region and completely destroyed 684 houses. More than 2500 people escaped to 47 temporary shelters. Many residents still live in temporary housing.

The disaster area is located in northern Japan, a relatively depopulated area with a large elderly population. Therefore, most of the victims of the Noto Peninsula earthquake were elderly; pregnant women are a minority group in this area. Nevertheless, 1000 neonates on average are born annually in the area most devastated by the earthquake. Several studies have documented maternal vulnerability in earthquakes. Antenatal stress is associated with a wide range of adverse birth outcomes in humans and other animals.1–3 Therefore, the effect of disaster-induced stress on pregnant women cannot be ignored.

Although several publications have addressed health problems, including psychological distress, in non-pregnant survivors of devastating earthquakes,4–6 few studies have examined the effects of disaster-related stress on pregnant women.7–10 Several prior studies suggest that prenatal stress caused by natural and man-made disasters may result in minor psychiatric distress, low birthweight,8 preterm delivery,7,11 smaller head circumference,11,12 skewed sex ratios,10 retardation of fetal brain development,13 and reduction of breast milk.9 Moreover, pregnancy is more susceptible to stress effects in the first trimester than in the second and third trimesters.7,11–13

Given that psychiatric stress has known detrimental effects on maternal health, the present study focused on the health impact of stress on pregnant women living in an area affected by a major earthquake. We explored whether existing stress resistance among these women moderated the earthquake-related stress. Understanding the relationship between stress and maternal health is critical for the development of a complete support system in the setting of a natural disaster.

METHODS

Data collection

Pregnant women who were receiving routine prenatal care at four hospitals in the Noto area were recruited. All of the women were pregnant during or immediately after a major earthquake and were living in the disaster area. They provided informed consent to participate in the study. Three months after the earthquake, a self-rating questionnaire was distributed to each study participant during a visit to her physician or midwife. Completed questionnaires were sealed in a white envelope and returned to the hospital or mailed directly to the researchers by each respondent. Some participants in this panel study completed questionnaires both before and after their deliveries. From 25 June 2007 to 31 January 2008, 311 of 399 questionnaires were returned (response rate, 77.9%). Among a total of 199 participants, 93 women completed two questionnaires and 19 completed three. The questionnaires of 99 women who provided repeat responses before and after their deliveries were analyzed; if there were two questionnaires in one category, the earlier questionnaire was adopted for analysis. To avoid selection bias, we first compared the mean Sense of Coherence (SOC) and Edinburgh Postnatal Depression Scale (EPDS) scores between analyzed and non-analyzed participants. No statistical difference was observed (SOC, P = 0.34; EPDS, P = 0.43). Second, we compared the means of SOC and EPDS scores of the 199 and 93 and 19 questionnaires and found that P did not reach significance (SOC, P = 0.91; EPDS, P = 0.15). This study was approved by the Ethics of Committee of the Kanazawa Graduate of School of Medical Sciences (No. 525).

Assessments

Stress outcome variables

One outcome variable in the present study was postnatal psychological depression. Psychological impact was assessed before and after delivery on the EDPS, which was developed by Cox et al. in 1987.14 Its validity and reliability have been confirmed in the Japanese population.15 The EPDS consists of 10 items, each of which is a 4-point Likert scale. The total score ranges from 0 to 30. The cut-off for the EPDS is a score of 9. Respondents with scores 9 are considered to be at significantly increased risk of maternal depression. The EPDS was originally designed to assess postnatal maternal depression but the scale has been used in pregnant women16–19 and even fathers of neonates.20–22 The present study adopted the EPDS to assess depression mood or status in women before and after delivery, and thus cannot determine or accurately diagnose ‘postnatal depression’ or ‘maternity blues’ among participants.

Another outcome variable in the present study was the presence of a physical abnormality during pregnancy or childbirth (yes, no). Information was obtained from women following delivery. Forty-seven postnatal women who responded ‘yes’ (47.5%) were further queried regarding which abnormalities they had experienced (Table 1).

Table 1.  Women participating in the study (n = 99)
  • Seismic Intensity is measured with seismic intensity meters.

  • Physical abnormalities during pregnancy or childbirth were vacuum extraction (n = 8), anemia (n = 11), bleeding (n = 4), GBS (n = 2), threatened premature delivery or threatened abortion (n = 21), breast disorders (n = 3), use of oxytocic agents (n = 1), cesarean section (n = 8), velamentous cord insertion (n = 2), pregnancy hypertension (n = 4), and other (n = 1). Total number exceeds 47 because of multiple answers.

  • EPDS, Edinburgh Postnatal Depression Scale; JMA, Japan Meteorological Agency; SOC, Sense of Coherence.

Demographic factors
 Age (years) Mean ± SD (range)29.6 ± 4.6 (21–43)
 Marital status n (%)
  Married99 (100.0)
 Parity n (%)
  (−)51 (53.1)
  (+)45 (46.9)
 Years living in the community n (%)
  <347 (56.0)
  ≥3 − <511 (13.1)
  ≥5 − <1017 (20.2)
  ≥109 (10.7)
Family background
 Living with spouse n (%)
  (−)7 (7.1)
  (+)92 (92.9)
 Living with parents n (%)
  (−)57 (57.6)
  (+)42 (42.4)
 Living with grandmother/father n (%)
  (−)79 (79.8)
  (+)20 (20.2)
Earthquake-related factor
 Gestational age at earthquake exposure (mean ± SD, range)13.1 ± 6.7 (0–25)
  First trimester (0–12 weeks) n (%)36 (36.7)
  Second trimester (13–27 weeks) n (%)59 (60.2)
  Third trimester (28–40 weeks) n (%)3 (31.0)
 House damage n (%)
  (−)55 (64.0)
  (+)31 (36.0)
 Evacuation n (%)
  (−)67 (72.8)
  (+)25 (27.2)
 Intensity of an earthquake (JMA: Seismic Intensity Scale) n (%)
  5 lower and upper11 (12.1)
  6 lower21 (23.1)
  6 upper59 (64.8)
 Subjective feelings regarding the earthquake n (%)
  Not so fearful32 (37.2)
  Very fearful54 (62.8)
 Existing anxiety about an earthquake n (%)
  (−)75 (75.8)
  (+)24 (24.2)
Participation before childbirth (mean ± SD, range)
 Weeks after earthquake exposure19.6 ± 5.8 (0–41)
 Gestational age (week)32.1 ± 5.1 (19–41)
 SOC score59.2 ± 10.3 (28–82)
 EPDS score3.9 ± 4.1 (0–19)
Participation after childbirth (mean ± SD, range)
 Weeks after earthquake exposure30.4 ± 8.1 (18–48)
 Days after delivery26.7 ± 12.5 (2–68)
 SOC score60.5 ± 9.5 (39–83)
 EPDS score3.8 ± 4.1 (0–22)

Sociodemographic and perinatal data

Sociodemographic data, including participant age, marital status, parity, family members living together, years living in the community, and participant address, were obtained from the questionnaire before delivery. Perinatal data included gestational age at the time of the earthquake exposure and the number of days after delivery.

Exposure variables

Information on earthquake exposure was obtained. Earthquake intensity was calculated from the participant's address where the earthquake occurred (5 lower and upper/6 lower/6 upper). The Japanese scale for seismic intensity is measured with seismic intensity meters. Information about injuries was solicited but no injuries existed among participants or their family members. Additional variables were whether or not an escape shelter or temporary housing were used (yes, no), subjective feelings regarding the earthquake (not so fearful, very fearful), and existing anxiety about an earthquake (yes, no). The level of damage to a participant's home was originally assessed with the terms ‘completely destroyed’, ‘half destroyed’, ‘partly destroyed’, or ‘none’, which resulted in a skewed distribution. Therefore, answers to this variable were integrated into the outcomes of ‘yes’ or ‘no’ before analysis.

Stress moderator

The SOC scale was used to assess the effect of existing stress resistance among pregnant women in moderating the relationship between stress outcome and earthquake exposure. The SOC scale, originally developed by Antonovsky in 1987, measures how well a person controls his or her environment in order to assign meaning to daily events.23 A person with a strong SOC is less likely to perceive a situation as stressful, and is more likely to adapt positively to stressful situations such as a natural disaster. Individuals with a weak SOC are more likely to develop feelings of anxiety and maladaptive responses to stressful situations. The short version of the SOC scale adopted in the present study consisted of 13 items with a 7-point Likert scale and a total score ranging from 7 to 91.

Statistical analysis

Pearson's correlation coefficient (r) or correlation ratio (η) was calculated to explore the relationship between dependent and independent variables. Multiple linear regression analysis was conducted against postnatal depression (EPDS) and stepwise multiple logistic regression was conducted against physical abnormality during pregnancy or childbirth. Variables with P > 0.1 were excluded, and those with P < 0.05 were included in the final model. Evaluation of the SOC during pregnancy as a moderator was conducted with interaction terms on hierarchical multiple regression analysis. In the first step, earthquake-related factors and demographic variables were included in the model. In the second step the interaction term was evaluated. Assessment was performed with each increment of F and its P. In all multivariate analyses, statistical adjustment was conducted with age, parity and SOC score during pregnancy. Statistical analysis was performed using SPSS (statistical package for the social sciences) version 14.0 for Windows.

RESULTS

Characteristics of the study subjects including earthquake-related factors are outlined in Table 1. The mean age (±SD) of the 99 participants was 29.6 ± 4.6 years. The mean number of weeks (±SD) after earthquake exposure, before and after the delivery was 19.6 ± 5.8 and 30.4 ± 8.1, respectively. The mean gestational age (±SD) before delivery was 32.1 ± 5.1 (week), and the mean number of days after delivery was 26.7 ± 12.5. The mean SOC scores before and after the delivery were 59.2 ± 10.3 and 60.5 ± 9.5, respectively. These scores were weakly correlated (r = 0.18, P = 0.07). The mean EPDS scores before and after delivery were 3.9 ± 4.1 and 3.8 ± 4.1, respectively, and were highly correlated (r = 0.66, P < 0.001).

Correlations between outcome and potential predictor variables are presented in Table 2. The EPDS scores after delivery significantly correlated with ‘existing anxiety about an earthquake’ (η = 0.33, P < 0.01) ‘parity’ (η = 0.36, P < 0.01), and EPDS during pregnancy (r = 0.66, P < 0.01). ‘Physical abnormality during pregnancy or childbirth’ was significantly correlated with ‘parity’ (r = −0.29, P < 0.01) and EPDS during pregnancy (η = 0.35, P < 0.01).

Table 2.  Intercorrelations among dependent and potential predictor variables (n varies from 82 to 99)
 123456789101112
  • *

    P < 0.05;

  • **

    P < 0.01.

  • 1, evacuation (−/+); 2, existing anxiety about an earthquake (−/+); 3, house damage (−/+); 4, intensity of an earthquake (less than 6 upper/6 upper); 5, gestational age at earthquake (second and third trimester/first); 6, subjective feelings regarding the earthquake (−/+); 7, age; 8, parity (−/+); 9, SOC before delivery; 10, EPDS before delivery; 11, EPDS after delivery; 12, physical abnormality during pregnancy or childbirth (−/+).

  • Intercorrelations are presented using correlation ratio (η) in cases between continuous variables and dichotomous variables. Others were presented using Pearson's correlation coefficient (r).

 11.00           
 20.141.00          
 3−0.110.141.00         
 40.22*0.110.021.00        
 50.120.07−0.030.031.00       
 6−0.060.050.23*−0.18−0.021.00      
 7−0.21*−0.13−0.09−0.32**0.010.21*1.00     
 8−0.18−0.30**−0.010.050.20*−0.010.19*1.00    
 90.210.25*0.070.120.040.040.120.021.00   
100.080.50**0.040.16**0.21*0.03−0.120.40**0.211.00  
110.000.33**0.050.180.060.15−0.150.36**−0.090.66**1.00 
120.110.120.09−0.24*−0.070.020.04−0.29**0.010.35**0.181.00

Multiple linear regression analysis against postnatal depression (EPDS) indicated that significant earthquake-related factors were ‘existing anxiety about an earthquake’ (β = 0.27, P = 0.01,) and ‘parity’ (β = −0.26, P = 0.02). In this model, women who were anxious about an earthquake during pregnancy and were nulliparous had an increased risk of depression after delivery. Other variables (age, evacuation, house damage, gestational age at earthquake, intensity of an earthquake, subjective feelings regarding the earthquake, and SOC during pregnancy) were excluded from the model (Table 3).

Table 3.  Predictor variables against postnatal depression (EPDS)
 Partial regression coefficient (B)Standard error (SE)Standardized partial regression coefficient (β)tP
  1. Age, evacuation, house damage, gestational age at earthquake, intensity of an earthquake, subjective feelings regarding the earthquake, and Sense of Coherence were excluded from the stepwise model.

  2. EPDS, Edinburgh Postnatal Depression Scale.

Anxiety about an earthquake (−/+)2.250.880.272.570.01
Parity (−/+)−1.880.78−0.26−2.420.02
R2    0.18
Adjusted R2 (d.f. = 1)    0.16

Multiple logistic regression analysis against a physical abnormality during pregnancy or childbirth indicated that EPDS during pregnancy (odds ratio, 1.21; 95% confidence interval: 1.04–1.41) was significant. Women with a high EPDS score during pregnancy had an increased risk of a physical abnormality during pregnancy or childbirth. Other variables (age, parity, evacuation, house damage, gestational age at earthquake, intensity of an earthquake, subjective feelings regarding the earthquake, and SOC during pregnancy) were excluded from the model (Table 4).

Table 4.  Predictor variables against physical abnormality during pregnancy or childbirth
 Odds ratios95%CI P
  1. Age, Parity, evacuation, existing anxiety about an earthquake, house damage, gestational age at earthquake, intensity of an earthquake, subjective feelings regarding the earthquake, and Sense of Coherence were excluded from the stepwise model.

  2. CI, confidence interval; EPDS, Edinburgh Postnatal Depression Scale.

EPDS during pregnancy1.211.041.410.01
Nagelkerke R2   0.13
Hosmer and Lemeshow (χ2 = 2.92, d.f. = 6)   0.82

Evaluation of the SOC as a moderator between the outcome variables and earthquake-related stress are presented in Table 5. The SOC during pregnancy moderated between EPDS after delivery and ‘existing anxiety about an earthquake’ (P = 0.02) (Figure 1), and tended to moderate between ‘physical abnormality during pregnancy or childbirth’ and ‘evacuation’ (P = 0.07). The remaining earthquake-related factors were not moderated by the SOC during pregnancy.

Table 5.  Evaluation of SOC during pregnancy as a moderator against postnatal depression (EPDS) and physical abnormalities caused by earthquake-related stress
 StepPostnatal depressionPhysical abnormality
βR2R2P for ▵FβR2R2P for ▵F
  • Step 1 was adjusted for age and parity, SOC;

  • final model.

  • SOC, Sense of Coherence.

1Evacuation (−/+)−0.080.23  0.150.043  
2Evacuation × SOC−0.04 0.0010.71−0.19 0.0340.07
1House damage (−/+)0.040.232  0.140.031  
2House damage × SOC0.01 00.950.02 0.0010.84
1Existing anxiety about an earthquake (−/+)0.160.269  0.070.051  
2Existing anxiety about an earthquake × SOC−0.21 0.040.020.05 0.0020.64
1Gestational age at earthquake (second and third trimester/first)0.080.251  −0.020.062  
2Gestational age at earthquake × SOC0.06 0.0030.50−0.07 0.0050.47
1Intensity of an earthquake (less than 6 upper/6 upper)−0.150.241  −0.220.1  
2Intensity of an earthquake × SOC−0.12 0.0120.23−0.17 0.0230.13
1Subjective feelings regarding the earthquake (−/+)−0.030.175  0.000.026  
2Subjective feelings regarding the earthquake × SOC0.00 0.0000.990.14 0.0180.21
Figure 1.

Moderating effect of SOC during pregnancy. Predictor variable presented as mean-centered; lines represent regression slopes at 1 SD above and below the mean. Heavy line: women with earthquake anxiety; thin line: women with no earthquake anxiety.

DISCUSSION

The mean level of EPDS for the study participants after delivery was 3.8 ± 4.1, and the percentage of high-risk individuals (those with a total score <9 points) was 13.1% before delivery and 8.1% after. In one study the reported prevalence of postnatal depression in a large Japanese population was 13.9%.24 The prevalence of postnatal depression in the present study might be in the lower range for the general Japanese population. One possible interpretation of this result is that the earthquake had little psychological impact as indicated by the EPDS, on pregnant women living in the Noto area. Another possibility is that the EPDS of inhabitants of rural areas such as Noto is naturally lower than that of the average Japanese population, especially those living in urban areas.

Studies on the psychological impact of earthquakes indicate that 25–50% of earthquake survivors suffer from psychological dysfunction or post-traumatic stress disorder (PTSD).8,25–27 Again, the present rate of these more serious psychological problems was lower than in prior studies. A possible explanation for this is that women with severe psychiatric symptoms may have been less likely to enroll in the present hospital-based study, thereby resulting in underestimation of the prevalence of psychological dysfunction.

Earthquake exposure is a classic example of an event that might cause PTSD. Previous studies have used this scale on pregnant women.8,12,28 We did not use the PTSD scale on the pregnant women in the present study, however, because they were regarded as a susceptible population. Therefore, the prevalence of PTSD among the present study participants was unknown, and the prevalence of depressive symptoms was within the normal range.

Another reason why the present study may have found a relatively low incidence of psychological dysfunction when compared to other studies is that the participants may have experienced only minimal physical and material damage. Previous studies demonstrate a positive correlation between physical and material damage caused by an earthquake, and increased psychological morbidity.8 The fact that no injuries were reported in the present study might contribute to the low prevalence of depression. Existing selection bias in the present study, however, might also be one reason why study participants had minimal physical and material damage.

Nevertheless, results from the multivariate analysis against EPDS after delivery indicated that the significant predictor variable with regard to the earthquake was ‘existing anxiety about an earthquake’. These results imply that the earthquake might arouse various negative psychological symptoms, such as anxiety and fear, among pregnant women, and that these psychological symptoms might lead to postnatal depression. We conclude, based on these results, that mental health care should be provided for pregnant women, in particular those with anxiety.

The sociodemographic factor ‘parity’ was also a significant factor for EPDS after delivery. It is understandable that nulliparous women may be at higher risk for depressive symptoms simply due to a lack of experience with pregnancy, childbirth, and child rearing. This lack of experience may also render the woman more vulnerable to earthquake-induced stress. Previous studies also found that other sociodemographic and personal factors such as single motherhood,29 lack of support from a spouse,30 presence of extended family,31 breast feeding cessation,15,32 and social isolation33 were significant risk factors for the development of postnatal depression. These risk factors may also be applicable for women living in a disaster area. Thus, support for women with these factors, especially nulliparous women, is important following an event such as an earthquake.

A physical abnormality during pregnancy or childbirth was significantly predicted by prenatal EPDS level on multivariate analysis. Although the literature is somewhat mixed, the overall consensus is that antenatal stress increases risk during pregnancy and childbirth. Stress may increase the production of cortisol and corticotrophin-releasing hormone, which promote the biological cascade leading to delivery.34 Not all types of stress, however, have the same effect on cortisol production. For instance, major depression tends to be associated with increased cortisol production.28 Antenatal depression was found to be a risk factor for preterm delivery.35 The present finding thus suggests that psychological responses, such as anxiety about an earthquake, may be mediated by EPDS, and that EPDS may be severe enough to lead to physical abnormalities in pregnancy or childbirth. Therefore, pregnant women with high EPDS scores and additional risk factors should be identified and carefully followed, especially after a natural disaster, to minimize physical dysfunction.

Glynn et al. reported that mothers exposed to an earthquake in their first trimester were more biologically vulnerable than mothers who were exposed in the third trimester.7 We did not detect a similar relationship, however, in the present study. In order to understand the causal relationship between trimester and earthquake exposure, an additional, more rigorous study that completely controls for time-related factors is warranted.

Individual psychological characteristics can modify individual response to stress. The SOC is a measure of this intrinsic stress resistance. SOC level in the present study was higher than that of non-pregnant women from previous studies. Yamazaki and Inoue estimated an SOC level in Japanese women of 44.1–47.6.36 The SOC level among pregnant women, however, is higher than in non-pregnant women. Matsushita et al. found an SOC level of 62.5 ± 12.9 in 52 pregnant Japanese women,37 and Engelhard et al. found an SOC of 65.1 ± 9.6 in 1372 early pregnant women.38 These SOC levels are consistent with those in the present study. Therefore, we infer that the SOC of the participants in the present study was within the normal range.

The results of studies evaluating SOC as a moderator are mixed. Retrospective studies report a relationship between a high SOC and less post-traumatic stress after various traumatic events.39,40 The SOC may also protect against depressive symptoms after a stressful life event.41 In a longitudinal study of pregnant women, the SOC was a resilience factor for psychological distress after pregnancy loss.38 Virtually most previous studies were retrospective and assessed the SOC after the traumatic event took place. This does not establish causation; the SOC may affect psychological distress, or vice versa. Although the SOC is stable over time, it may be influenced by stressful events, particularly when efforts at mastery fail.42 The present study has the same limitations as any retrospective study, but this is the first study to explore the relationship among earthquake exposure, SOC during pregnancy, and postnatal psychological and physical status.

The most notable results of the present study were that ‘existing anxiety about an earthquake’ during pregnancy predicted postnatal depression, and that SOC during pregnancy moderated between the two states. In addition, SOC tended to moderate the relationship between ‘physical abnormality during pregnancy or childbirth’ and ‘evacuation’, although the main effect of evacuation was not significant. Because of the small sample size, this study may have been underpowered to demonstrate a significant moderator effect in this relationship. Therefore, interventions that improve the SOC among young women based on a population approach may equip them to deal well with stressful life events such as natural disasters. Targeting SOC alone, however, is not sufficient. For example, a more intensive approach is likely needed for women with extremely high EPDS scores who might have been excluded from the present study. Moreover, an adequate and comprehensive support system for pregnant women should be in place for those exposed to natural disasters.

There are a number of limitations to the present study. First, the sampling method may have created selection bias that resulted in an underestimation of negative health impacts. Second, the small sample size may have led to inadequate statistical power and warrants caution in interpretation. Third, this is a retrospective design, and a causal relationship cannot be determined. Last, findings from the present study may not generalize to other populations or to other types of disasters. In particular, this study had no reference group, so the present findings are specific for pregnant women who experienced the Noto Peninsula earthquake. But literature on this subject is sparse. Thus, the present study makes an important contribution to the understanding of the impact of a natural disaster upon pregnant women.

ACKNOWLEDGMENTS

The authors would like to thank the participants of the present study. We would also like to thank those who made this study possible: Dr Aoyama Kouya at Wajima Hospital, Dr Nakahama Yukio at Anamizu General Hospital, Dr Fujita Tomoko at Noto General Hospital and Dr Kohama Takafumi at Keiju General Hospital.

Ancillary