Mental Illness in Japan

Currently, over 80% of the mental health hospitals in Japan are private, and most are filled primarily with involuntary patients who are physically disabled, intellectually disabled, or who suffer from a severe psychotic disorder such as schizophrenia. As in many other countries, the mental healthcare system in Japan focuses on the long-term care of chronic disorders and developmental disabilities that exceed the caring capacity of families. Consequently, there are currently few hospitals or clinics in the country that people can visit to treat more acute disorders such as depression and post-traumatic stress disorder. While there has recently been a legislative push towards a more community-based mental healthcare system in Japan that focuses on these illnesses, the country’s mental health services today are still primarily reserved for the severely mentally ill[4].

In addition to the structural problems inherent in Japan’s mental healthcare system, many people in the country also refrain from using mental healthcare services due to the negative stigma attached to mental illness. According to Dr. Hiroshi Kato of the Hyogo Institute for Traumatic Stress in Kobe, gaman, or ‘endurance,’ is a very important attitude in Japanese culture that has caused many people in the country to respond to disasters with silence, avoidance, and patience[5]. Depression, which is often referred to by its euphemistic name, “heart flu,” is a particularly taboo subject in Japanese culture. Due to intense cultural pressures, individuals suffering from depression often end up feeling guilty or ashamed about discussing their psychological pain[6].

In addition to the stigma attached to depression, there is also a strong and deep-seeded stigma in Japan attached to radiation poisoning. More specifically, following the bombings of Hiroshima and Nagasaki during World War II, survivors of the disasters–referred to as “hibakusha”–were heavily discriminated against in other parts of the country. Today, individuals displaced from Fukushima have begun reporting similar types of discrimination, with some individuals being discouraged from seeking refuge in other prefectures for that reason. Children who have had to move to areas outside of the Fukushima prefecture have also begun reporting being bullied in elementary and middle schools due to their potential exposure to radiation[7].

Pediatric Mental Illness in Japan

While the overall prevalence of pediatric mental illnesses in Japan is not known, a number of factors related to the March 11, 2011 disaster suggest that many of the children who were exposed to the earthquake, tsunami, and radiation scare are likely to suffer from moderate- to severe-levels of mental health problems such as depression and post-traumatic stress disorder.

The first such factor is the nature of the catastrophe itself. Previous studies have shown that children are often more susceptible than adults to mental health problems in the months and years following a natural disaster[8] and often manifest these problems through psychological withdrawal, temper tantrums, decreased concentration, and sleep disorders[9]. In addition to having to bear witness to the disaster itself, children are also known to worry more than adults about the possible recurrence of a disaster as well as the fate of family members, friends, and pets. Those who have already become separated from their parents often become confused about their own safety and the safety of their caregivers, while those who remain with their parents often worry that they may still become separated from them in the future. During a natural disaster, children are also known to be more vulnerable to stress than adults, particularly those who have witnessed people die or who have been trapped in damaged buildings or by debris for any length of time[10].

With regard to the radiation spill, the Japanese government’s decision to limit evacuation to within only 20 kilometers of the Fukushima Daiichi power plant has also been heavily criticized both domestically and internationally for its disproportionate impact on the region’s 300,000 children. This decision prompted six non-governmental organizations in Japan to submit a joint report to the UN Office of the High Commissioner for Human Rights (UNHCHR) describing the situation of the children in Fukushima and urging the United Nations to intervene[11]. Due to these factors, the need for more research investigating the prevalence of mental illnesses among the children of Fukushima as well as possible risk and protective factors is needed.

Risk and Protective Factors for Pediatric Mental Illnesses

Three risk factors that appear to have a significant predictive effect on the onset of pediatric mental illnesses following a natural disaster include exposure to disaster-related violence, displacement from one’s home, and impaired parental mental health. Exposure to violence has repeatedly been shown to be a strong risk factor in the development of mental disorders among children affected by natural disasters. A study of Chilean refugees in Sweden indicated that rates of sleep disturbances and anxiety were higher among children who had witnessed violent events before migration[12]. Among Hurricane Katrina survivors, youths with a high exposure to violence were also more likely to develop long-term emotional disturbances than those with low- or moderate- exposure to violence[13].

Unlike exposure to violence, displacement from one’s home appears to have a more complicated predictive effect on the onset of pediatric mental illnesses. A number of studies focusing on Bosnian and Croatian refugees, for instance, have shown that while the prevalence of most mental illnesses does not differ significantly between displaced and non-displaced children, psychosocial adaptation remains worse in displaced populations[14]. Some of these studies have also suggested that displaced children have higher levels of anxiety and fewer coping strategies to deal with stressful situations than non-displaced children[15].

The existence of emotional and mental disturbances among the parents of children affected by a disaster has also been shown to be a significant risk factor in the development of pediatric mental illnesses. While researching the mental health of children affected by the 2003 earthquake in Iran, for instance, Kalantari and Vostanis found a significant positive relationship between children’s scores on the Strengths and Difficulties Questionnaire (SDQ) and their parents’ scores on the Self-Report Questionnaire (SRQ)[16].

In addition to these three risk factors, two factors that appear to be protective against the onset of pediatric mental illnesses include male gender and a younger age. More specifically, a number of studies have shown that depression is more prevalent among female refugee children than male refugee children[17]. Pyari’s research among survivors of the 2004 Indonesia tsunami[18] and Zhang’s research among survivors of the 2008 Sichuan earthquake in China[19] have also suggested that being male can be a strong protective factor against the onset of post-traumatic stress disorder. While the exact mechanisms behind this phenomenon are unclear, some researchers believe that the distinction can be attributed to differing coping mechanisms between girls and boys as well as more limited socio-economic resources for girls in many cultures[20].

Finally, numerous studies looking at the mental health of both children and adults have suggested that older survivors of natural disasters (>40 years) are more susceptible to mental health disorders such as post-traumatic stress disorder than their younger counterparts[21]. Within the youth population, however, very little is known about the protective effects of young age against mental illness.

Accordingly, we hypothesized that pediatric mental health outcomes, measured in sampled children using standardized Strengths and Difficulties questionnaires, would be influenced by exposure to violence, displacement from one’s home, impaired parental mental health, gender, and age.

Sample Size Calculations

The sample population was divided into three groups: students displaced from the coast to the city of Koriyama (category I–displaced children), students who were local to Koriyama (category II–non-displaced children), and students who lived outside of the Fukushima prefecture entirely (category III–controls). Using the Strengths and Difficulties Questionnaire in a Japanese population, the effect size often ranges from 2 to 3 with standard deviations ranging from 5 to 6[22]. With a two-sided alpha of 0.05 and beta of 0.20, the estimated number per group would therefore need to be between 45 and 176 in order to achieve a statistically significant effect.

Exposure Variables

The students were divided into the three categories of displaced children, non-displaced children, and controls in order to distinguish the effects of the earthquake, relocation, and radiation spill (exposure variables) on the children’s mental health (outcome variable). Fig 1 shows the location of the two control cities in relation to Koriyama, the Fukushima power plant, and the earthquake’s epicenter. The epicenter of the earthquake was located approximately 80 km east of Japan’s coast, 200 km east of Koriyama, 375 km northeast of Komoro, and 400 km northeast of Tokyo.

Download:

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

Fig 1. Map of Japan.

https://doi.org/10.1371/journal.pone.0170402.g001

According to the United States Geological Survey, “perceived shaking” was recorded as “very strong” to “severe” in Koriyama and along the eastern coast of the Fukushima prefectures, with peak ground velocities ranging from 16 to 70 centimeters/second. In Tokyo and Nagano, on the other hand, “perceived shaking” was recorded as “moderate” to “strong,” with peak ground velocities ranging from 3.4 to 16 centimeters/second[23]. Those children who were located in Koriyama and along the eastern coast were therefore exposed to much stronger tremors than those children who lived in the control areas of Nagano and Tokyo.

The radiation fallout in and around the Fukushima prefecture was also higher than in the control cities as measured by surface deposits of cesium-134 and -137. Internationally, a cesium fallout density of 10,000 becquerels per spare meter is recognized as the threshold for defining an area as being affected by a nuclear accident[24]. Koriyama was exposed to a combined cumulative density of between 60,000 and 100,000 becquerels per square meter. In Tokyo and Nagano, the fallout densities were considerably lower, measuring between 0 and 10,000 becquerels per square meter. Therefore, those children who lived in Koriyama and along the eastern coast of Fukushima were also exposed to considerably higher radiation levels than those children living in the control prefectures of Tokyo and Nagano.

Participants

During the months of February and March, 2012, the researchers distributed 3,650 survey packets to nine schools across Japan. Of these, 3,000 were distributed to three elementary schools and four middle schools in Koriyama, Fukushima prefecture (categories I and II), 250 were distributed to one elementary school in Tachikawa, Tokyo prefecture (category III), and 400 were distributed to one middle school in Komoro, Nagano prefecture (category III). Because the researchers were unable to identify displaced children and their families in Koriyama using government records, surveys were distributed to all 3,000 children in the city with the hopes of receiving completed surveys from at least 45 displaced families–the threshold necessary to achieve a statistically significant effect.

By the end of April, the researchers had received a total of 1,004 completed surveys. Of these, 785 were returned from Koriyama (26%), 80 were returned from Tachikawa (32%), and 139 were returned from Komoro (35%). Of the 785 surveys returned from Koriyama, 51 were returned from displaced families and 734 were returned from native families. Table 1 shows the distribution of returned surveys by category.

Download:

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

Table 1. Sample Distribution by Category.

https://doi.org/10.1371/journal.pone.0170402.t001

Ethics

Ethics approval for this study was obtained by the Institutional Review Board of the University of California—San Francisco and participant consent was implicit in the parent’s decision to return the completed survey to the researchers. The questionnaires were also approved by local Japanese school officials.

Procedures

During the distribution of the survey packets in February and March, 2012, teachers in all nine participating schools were instructed to give the packets to their students during the last week of February and the first week of March. The students were then told by their teachers to give the packets to their parents upon returning home and to return the completed surveys via postal mail no later than the first week of April. Parents were provided with pre-addressed, stamped envelopes in order to successfully return the completed packets to the researchers. No other incentives to participate were employed.

The survey packets consisted of an original demographic survey, the Strengths and Difficulties Questionnaire, and the Impact of Event Scale-Revised, all of which were to be filled out by the children’s parents. The demographics survey was used to identify potential risk and protective factors, the Strengths and Difficulties Questionnaire (SDQ) was used to assess the mental health of the children (the main outcome variable), and the Impact of Event Scale-Revised (IES-R) was used to assess trauma in the children’s parents (one of the predictor variables). We decided upon these survey instruments over others that measure mental disorders due to the fact that they rely upon the responses of a child’s caregiver rather than the children themselves. Due to ethical and logistical constraints, we were not able to receive the necessary permission to evaluate the children directly. In addition to our research constraints, we also decided to rely upon the SDQ as our primary diagnostic tool because it is widely known for being one of the most reliable indicators of pediatric mental disturbances internationally, having been successfully used in Japan[25] as well as in other Asian countries such as Bangladesh, Pakistan, and Sri Lanka[26]. The SDQ includes 25 questions that are divided between five scales of five items each: Conduct Problems, Inattention-Hyperactivity, Emotional Symptoms, Peer Problems, and Prosocial Behaviors. These five subscores are then summed together to generate a Total Difficulties score of 0–40[27].

Measures

Aim 1: After all of the survey packets had been returned to the researchers, they were then divided into the three categories outlined above: students displaced from the coast to the city of Koriyama (category I–displaced children), students who were local to Koriyama (category II–non-displaced children), and students who lived outside of the Fukushima prefecture entirely (category III–controls). The Strengths and Difficulties Questionnaires were then scored for all 1,004 children by determining each child’s Total Difficulties Score and comparing it to clinical range cut-offs first developed by Robert Goodman in the United Kingdom[28]. More specifically, SDQ Total Difficulties scores of 0–13 were classified as “normal,” 14–16 as “borderline,” and 17–40 as “abnormal.” Mean SDQ scores for each of the three groups were then compared to one another using two-sample t-tests.

Aim 2: In order to identify potential risk and protective factors, the researchers relied upon responses to the demographic survey as well as the results of the parents’ Impact of Event Scale-Revised (IES-R) questionnaires. The IES-R was used to assess subjective distress caused by the earthquake, tsunami, and radiation spill in the children’s parents. Respondents were asked to indicate how much they were distressed within the past seven days by a number of “difficulties” that correspond to 14 of the 17 DSM-IV symptoms of post-traumatic stress disorder (PTSD). Items were rated on a 5-point scale ranging from 0 (“not at all”) to 4 (“extremely”) and total scores out of 88 were used to assess the parents’ level of distress[29]. While IES-R scores and age were both treated as continuous variables, gender, exposure to violence, and displacement from one’s home were all treated as dichotomous variables. With regard to exposure to violence, questions 13 and 14 in the demographics survey asked parents to indicate whether the child was hurt during the earthquake, whether a family member or friend was injured or died during the earthquake, or whether the child saw someone get hurt or die during the earthquake. If the parent responded positively to any of these questions, the child was marked as having been exposed to violence. After assessing the potential risk and protective factors, the researchers then used a number of regression models to identify and isolate which risk and protective factors were significantly associated with the children’s SDQ Total Difficulties scores, taking potential confounders into effect.

Exposure to Violence

Using a simple linear regression, exposure to violence appeared to be associated with children’s SDQ Total Difficulties scores (β = 2.46, p = .011). In order to account for any possible confounders, the investigators then performed a series of multivariate regressions to find out if this association remained while controlling for other potential factors.

First, the investigators looked at the relationship between exposure to violence and SDQ score while controlling for category type. Because the control children had not been exposed to disaster-related violence, the investigators looked at this correlation only within the native group and the relocated group. Among the native group, the correlation between these two variables remained significant (β = 2.41, p = .019) while, among the relocated children, there was no longer any significant correlation between SDQ score and exposure to violence (β = 2.16, p = .441).

In order to account for this discrepancy, the investigators then began looking for other possible confounders and found that, among the entire sample population, there was not a significant association between exposure to violence and SDQ score when controlling for parental trauma (β = 1.06, p = .247). When looking specifically at the native population, the correlation between exposure to violence and SDQ score also disappeared when controlling for parental trauma (β = 0.928, p = .339). Therefore, the investigators decided to discard exposure to violence as a potential risk factor for increased SDQ Total Difficulties scores.

Displacement from Home

Like exposure to violence, displacement from one’s home also appeared to have a statistically significant association with children’s SDQ scores. Because only relocated children experienced this risk factor, the investigators decided to look at the other two remaining factors–parental trauma and age–as potential confounders. In doing so, the researchers discovered that while the association remained significant when controlling for age (β = 1.85, p = .043), it again disappeared when controlling for parental trauma (β = 0.914, p = .309). Therefore, displacement from one’s home was also discarded as a potential risk factor for increased SDQ Total Difficulties scores.

Parental Trauma

Based on the simple linear regressions and the previous two multivariate regressions, parental trauma, as assessed by the Impact of Event Scale-Revised (IES-R), appeared to be a strong risk factor for impaired mental health among children. In order to confirm this, the investigators performed a series of multivariate regressions to control for age, displacement from one’s home, and exposure to violence. The association between parental trauma and children’s SDQ Total Difficulties scores remained statistically significant (p = .000) among all three analyses. Consequently, the researchers decided to keep parental trauma in their final regression model.

Age

Finally, while the effects of age on mental health have not been adequately researched within the youth population, our simple regressions and multivariate models suggest that age may, in fact, be a significant risk factor in the development of mental disturbances among children affected by a natural disaster. In order to confirm this hypothesis, the investigators again performed a series of multivariate regressions, controlling for each of the other three potential risk factors. In all three models, age continued to be significantly associated with impaired mental health among children (p = .000). What was perhaps most interesting, though, was that this association was negative, meaning that younger children were more at risk for developing mental disturbances after a natural disaster than older children.

In sum, parental trauma (β = 0.128, p = .000, 95%CI .095-.161) and a younger age (β = -0.588, p = .000, 95%CI -.799- -.377) were the only two risk factors that were significantly associated with SDQ Total Difficulties score when adjusted for other factors. More specifically, each one-point increase in a child’s SDQ Total Difficulties score was associated with a .128-point increase in parental IES-R score and a 7-month decrease in age. Together, these two variables accounted for 21% of the variation in SDQ score. Gender, displacement from one’s home, and exposure to violence were not found to have a significant association with SDQ Total Difficulties scores.