Examination of factors affecting insomnia in older victims of the Kumamoto earthquake

Introduction

Eight years have passed since the April 2016 Kumamoto earthquake (Kumamoto City, 2016). Due to natural disasters worldwide, health crisis management is becoming increasingly important, and post-disaster victims are at an increased risk of mental disorders such as depression (Matsubara et al., 2014) and post-traumatic stress disorder (hereinafter referred to as PTSD). They also need long-term support (Ide-Okochi et al., 2022). In addition, mental disorders such as depression increase the risk of developing further mental disorders (Li et al., 2016) as well as the appearance of insomnia from early onset.

According to a country-by-country comparison of average sleep duration released in 2021 by the Organization for Economic Cooperation and Development (OECD), Japan had the shortest sleep (7 h 22 min per day) duration among 33 countries around the world, mainly developed countries (OECD, 2021). The United States, which has the longest sleep duration among OECD member countries, had 8 h 51 min, an hour and a half longer than Japan. Therefore, Japan’s (Ministry of Health Labor Welfare, 2023) proposed the Sleep Guide for Health Promotion 2023 and indicated that it is important to ensure that both quality (sleep restfulness) and quantity (sleep duration) of sleep are adequate. In addition to daytime sleepiness, sleep problems are likely to cause increased psychosomatic complaints such as headaches, emotional instability (Vandekerckhove & Wang, 2017), and impaired attention and judgment (Groeger et al., 2014). It is also associated with an increased risk of developing or worsening symptoms of obesity (Häusler et al., 2019), hypertension (Wang et al., 2017), type 2 diabetes (Reutrakul & Van Cauter, 2018), heart disease (Korostovtseva, Bochkarev & Sviryaev, 2021), cerebrovascular disease (Chaudhry et al., 2020) and mortality (Ensrud et al., 2012).

On the other hand, economic hardship, loss of employment, and a pessimistic outlook reportedly have a significant impact on mental health transitions, including insomnia and suicidal ideation (Orui, Kuroda & Yusumura, 2019; Xu et al., 2018). Furthermore, there is evidence that the standardized mortality rate for suicide increases after survivors leave temporary housing (Orui et al., 2020). Since many depressed individuals commit suicide, it is important to focus on insomnia, which tends to develop early in the course of depression. It is also essential that insomnia be detected early and that local health care professionals provide appropriate support in post-earthquake areas, where the risk of mental disorders is likely to increase.

However, in recent years, the elderly tend to sleep longer because of reduced sleep restfulness caused by difficulty falling asleep, and reduced sleep efficiency (Yuan et al., 2022) due to mid-sleep awakenings. Prolonged sleep increases the risk of future mortality by a factor of 1.33 (Da Silva et al., 2016). Therefore, it is necessary to take measures to reduce the difficulty of falling asleep and awakening in the middle of the night, which cause long hours of sleep. Specifically, being active during the day and having a good day/night pattern contributed to good sleep (Ministry of Health Labor Welfare, 2023). Conversely, a lack of daytime activity and reduced day-night crispness can facilitate the emergence of daytime sleepiness and fatigue (Buysse et al., 2011) and lead to prolonged naps (Da Silva et al., 2016), preventing good sleep at night. In contrast, exercise helps shorten the sleep onset latency (Matsumoto et al., 2017) and improves subjective sleep quality (Hasan et al., 2022); eating breakfast prevents the body clock from going backward, contributing to sleepiness and sleep deprivation (Otsuka et al., 2023). Social capital such as community activities, interpersonal relationships, and trust (Ministry of Health Labor Welfare, 2023) also promotes good sleep and physical activity. Therefore, it is important to actively promote daytime activities such as exercise, diet, and social capital to achieve good night sleep.

Many people were forced to relocate after the disaster because of housing damage. Sugawara, Tohmata & Tsuji (2019) found that the odds ratio of insomnia was significantly higher in the group that relocated than in the group that did not and that the association with insomnia was stronger when the relocation site was further away from the district where the person had lived before the disaster. In addition, in the areas affected by the Great East Japan Earthquake, the rate of insomnia increased after people moved from temporary housing to permanent housing, which they moved into immediately after the disaster (Tsuji, 2019). Therefore, it is speculated that relocation may also affect insomnia.

Based on the above, it is necessary to understand the actual situation and factors affecting sleep-related problems among older adults affected by the disaster and to consider appropriate support after the earthquake. During the Kumamoto earthquake, over 180,000 people were affected and up to 110,000 were forced to move to temporary housing (Kumamoto City, 2016), of which approximately 24,000 (21%) were older adults. Therefore, support for insomnia among older adults affected by the Kumamoto earthquake is considered a medium- to long-term issue. However, at present, there are no findings that clarify the actual situation of insomnia and its factors. Therefore, we examined whether the extent of relocation and daytime activities such as exercise, diet, and social capital explain insomnia among older adults affected by the Kumamoto earthquake. Identifying the factors that influence insomnia will help local health and welfare professionals develop and implement measures to prevent mental disorders in older adults affected by disasters. Therefore, this study examined the factors affecting insomnia among older adults affected by the Kumamoto earthquake and to obtain suggestions for future health activities.

Materials & Methods

Variables

Basic attributes

Regarding basic attributes, the respondents were asked about their gender, age, cohabitants, temporary housing category, and current residence. Age was measured using a descriptive form in which the respondents were asked to state their current age. Gender was linked to the Kumamoto City ledger. Question: Do you have anyone living with you? The respondents were asked to answer by circling in either (1) yes or (2) no. The classification of temporary housing is linked to the Kumamoto City ledger. The question was: “Please check all that apply to your current residence.” The options were (1) owned house, (2) houses for rent, (3) public housing, (4) public housing for disaster, (5) Hospitals and institutions, and (6) other.

Scope of relocation

For relocation (within or outside the community), participants were asked the following question: Is your current residence in the elementary community where you lived before the Kumamoto earthquake? Please circle the answer that applies. The answer choices were (1) in the same elementary community, (2) changed elementary community, and (3) did not know. This study was set up on an elementary school district basis, referring to the fact that previous studies have investigated the range of movement in residential areas or outside the city (Otsuka et al., 2023). Preventive care activities in Kumamoto City are conducted on a community basis (elementary school district basis) because information about community activities is mainly communicated through posts, circulars, verbal communication at meetings, and invitations among residents.

Daytime activities

Exercise Habits (1/w) included: “Do you walk or perform other exercises or household chores that replace exercise (cleaning, yard care, etc.) at least once a week? Please check all that apply”. The question was: “Do you perform household chores (cleaning, yard maintenance, etc.) to replace exercise and exercise such as walking at least once a week?” The answer choices were (1) Yes or (2) No. Questions on exercise habits were set up with reference to the Active Guide (Ministry of Health Labor Welfare of Japan, 2023) based on the Physical Activity Standards for Health Promotion 2013.

Regarding meal habits, please answer the question: Do you eat three meals per day? Please circle the answer that applies. The answer choices were (1) I eat and (2) I do not eat. The question about appetite was: “How is your appetite these days? Please circle the answer that applies.” The answer choices were as follows: (1) increased, (2) same as usual, and (3) decreased. The question about solitary eating was: “Who do you eat meals with most often? Please circle the answer that applies.” The answer choices were as follows: (1) eating alone, (2) eating with family, and (3) eating with friends, where “Eating alone” means solitary eating.

Social capital was ascertained through participation in community activities and the availability of social support (advisors). Participation in community activities was measured by asking: “Do you participate in events and social gatherings held in your community?” Respondents were asked to answer by circling in one of the following options: (1) I participate, (2) I do not participate, or (3) I do not know of any such information. The question was set because the same question was asked in an earlier recovery survey conducted in Kumamoto Prefecture; 62.1% of the respondents did not participate, and 4.4% did not have such information (Kumamoto Prefectural Government, 2021). Also, in Hikichi et al. (2017), which revealed social capital and physical and mental health, the question “Do you participate in community events?” and a yes/no response method was used.

The social support (advisor) question asked: “Who is the person you talk to about your problems?” The respondents were asked to circle in all the following options. The choices were: (1) family; (2) friend; (3) neighbors; (4) coworker; (5) district welfare commissioners; (6) hospital; (7) nursing care office; (8) city office; and (9) no one. For example, “(1) family” means to consult with one’s family. (2) to (9) are the same. Regarding social support, the roles of friends (Wu & Sheng, 2019), neighbors (Yoshiyuki & Kono, 2020), and welfare commissioners (Sugihara, 2018) are particularly important. In addition, medical institutions and nursing care offices are involved depending on the health conditions. These options were established based on the aforementioned criteria.

Insomnia

Insomnia was assessed using the Japanese version of the Athens Insomnia Scale (AIS), an 8-item insomnia rating scale developed by the WHO World Project on Sleep and Health (Okajima et al., 2013). Scores range from 0 to 24, with a score of six or higher being rated as “suspected insomnia” and is used for screening purposes (Kawasaki, Nishitani & Sakakibara, 2015).

Analysis method

After calculating the basic statistics, a test of independence (X² test) is performed for insomnia and basic attribute of relocation, and daytime activities. Next, binomial logistic regression analysis was conducted with insomnia as the dependent variable to examine the factors that influence insomnia. Three models were created, with the independent variables being basic attribute of relocation, and daytime activity (Fig. 1). Model I attempted to explain how insomnia is affected by basic attributes. Model II attempted to explain how insomnia was affected by adding the scope of relocation to Model I. Model III attempted to explain the impact of daytime activity on insomnia by adding daytime activity to Model II.

The independent variables are gender, age (whether the person falls into the category of 75 or older), presence of a roommate, temporary housing category, current residence, extent of relocation, exercise habits, dietary habits, and social capital. Dummy variables were used as independent variables, and the Variance Inflation Factor (VIF) was calculated using linear regression analysis to check for multicollinearity before conducting a dichotomous logistic regression analysis. Variables were selected using the forced entry method, and the two-sided statistical significance level was set at 0.05. The goodness of fit of the model will be performed with the X² test and the Hosmer–Lemeshow test of the model. For the goodness of fit of the model, Negelkerke R² values are calculated. Missing values for each variable were excluded. SPSS Statistics 29.0 for Windows was used as the statistical software.

Results

Participants included 2,010 men (42.2%) and 2,748 women (57.8%). The average age of the participants was 75.36 ± 7.33 years (65–105), with 2,485 (52.2%) in the 65–74 age group and 2,273 (47.8%) in the 75+ age group. The basic attributes of the participants are listed in Table 1. Among the respondents, 3,473 (74.0%) live with a cohabitant. Most respondents (4,064 (85.4%)) lived in temporary housing. Regarding their current residence, 2,675 (56.7%) lived at home (owner-occupied), 1,024 (21.7%) lived in rented housing, and 768 (16.3%) lived in public housing. In addition, 1,494 (32.1%) participants moved out of the community.

In addition, 3,460 (75.9%) participants were in the habit of exercising; 3,955 (84.8%) were in the habit of eating three meals per day; 1,035 (22.2%) reported a decrease in appetite; and 1,467 (32.3%) ate alone. Community activities were participated in by 1,331 (28.9%) participants. In contrast, 2,863 (62.1%) did not participate in community activities and 415 (9.0%) did not have such information. Regarding social support, 238 (5.2%) respondents had no one to talk to about social support. On the other hand, 3,601 (78.1%) answered that they consulted family members, 1,657 (36.0%) friend, 404 (8.8%) neighbors, 190 (4.1%) coworker, 88 (1.9%) disrict welfare commissioners, 502 (10.9%) hospital, 293 (6.4%) nursing care office, 113 (2.5%) city office, and 2.5%), and 242 (5.1%).

Table 2 shows the actual situation regarding insomnia; the mean AIS score was 5.01 ± 4.14 points. A total of 2,663 (62.4%) participants had a total score between 0 and 5, and 1,606 (37.6%) had a total score of 6 or more. Regarding sleep induction, 1,498 (32.7%) were “slightly delayed”, 595 (13.0%) were “markedly delayed ”, and 157 (3.4%) were “very delayed or did not sleep at all”. Regarding awakenings during the night, 1,236 (27.2%) were “minor problem”, 410 (9.0%) “were considerable problem”, and 50 (1.1%) were “serious problem or did not sleep all”. Regarding final awakening earlier than desired, 1,797 (39.6%) “a little earlier”, 521 (11.5%) “markedly earlier”, and 121 (2.7%) were “much earlier or did not sleep at all”. Regarding total sleep duration, 1,741 (37.6%) were “slightly insufficient”, 366 (7.9%) were “markedly insufficient”, and 59 (1.3%) were “very insufficient or did not sleep at all”. Regarding very insufficient or did not sleep at all, 1,972 (42.6%) were “slightly unsatisfactory”, 451 (9.7%) were “markedly unsatisfactory”, and 58 (1.3%) were “very unsatisfactory or did not sleep at all”. Regarding sense of well-being during the day, 1,074 (23.4%) were “Slightly decreased”, 199 (4.3%) were “markedly decreased”, and 43 (0.9%) were “very decreased”. As functioning during the day, 1,448 (31.4%) were ”slightly decreased”, 590 (12.8%) were “markedly decreased”, and 149 (3.2%) were “very decreased”. Sleepiness during the day was “mild” in 3,195 (69.5%), “considerable” in 492 (10.7%), and “intense” in 23 (0.5%).

The results of the cross-tabulation of independent variables with and without insomnia are shown in Table 3. The variables significantly associated with the presence or absence of sleep were gender, age, cohabitant, type of temporary housing, current residence, scope of relocation, exercise habits, eat three meals a day, appetite, solitary meal, community participation, and social support. Regarding social support, significant differences were found in the presence or absence of anyone and family, coworker, hospital, nursing care office, city office, or other, while no significant differences were found for friend, neighbors, and district welfare commissioners.

To confirm multicollinearity, the VIF were calculated as 34.4, 23.6, 18, and 6 for the current type of residence (home, rental, and public housing, respectively). Therefore, current type of residence was removed from the independent variables in the binomial logistic regression analysis. The results of this analysis are shown in Table 4, with the presence or absence of insomnia as the dependent variable and basic attribute of change in residence, and daytime activities (exercise habits, meal habits, and social capital) as independent variables. Model I showed p < 0.001, Negelkerke R² was 0.024, and the Hosmer–Lemeshow test showed p = 0.946. Model II showed p < 0.001, Negelkerke R² showed 0.027 (p = 0.545 by the Hosmer–Lemeshow test). Model III also showed p < 0.001, Negelkerke R² was 0.189, and p = 0.397 by the Hosmer–Lemeshow test. Therefore, both models fit. Furthermore, a comparison of Negelkerke R² determined that the best model fit was Model III (0.189). Therefore, Model III was adopted for the binomial logistic regression analysis.

Insomnia was associated with age 75+ (1.17, 1.004–1.35), no exercise habit (1.23, 1.04–1.47), not eating three times a day (1.40, 1.13–1.72), no appetite (3.79, 3.18–4.52), solitary meal (1.60, 1.27–2.02), no community participation (1.34, 1.14–1.59), no community information (1.94, 1.47–2.55), no one (1.49, 1.04–2.13), no family to consult (1.49, 1.21–1.84), and no colleagues to consult (1.66, 1.14–2.43).

Discussion

Limitations

This study has a number of limitations.

The first limitation is due to the study design. Although temporal order is important in post-earthquake studies, this study was a cross-sectional study owing to the lack of longitudinal data. In addition, the limited explanatory power of the final model with an R² of 0.189, the failure to distinguish between pre-existing and new cases of insomnia, and limited control for psychiatric comorbidities such as depression and PTSD limit the interpretation of causal relationships. In addition, the use of binomial logistic regression assumes linearity and independence of predictors but does not examine interactions between variables. This may increase the risk of Type I errors.

Second, this study omits trauma-related variables that are predictors of post-disaster mental health problems. This may confound the relationship between relocation and insomnia and may weaken the validity of the findings.

Third, there is the issue of self-reported data and measurement. Considering that AIS alone is insufficient for assessing the severity and functional impairment of insomnia, it is desirable to consider complementary assessment methods in accordance with study objectives in the future. In addition, social capital and exercise habits are operationalized in a simplistic manner, and so nuances such as frequency, intensity, and quality of interactions may be overlooked. Additionally, it may have been difficult to accurately answer some of the questions. For example, the question, “Do you perform household chores in place of exercise?” could not have been answered by those who do perform household chores but not in place of exercise. Additionally, the question, “How is your appetite these days?” does not specify a comparison timepoint, and so the responses may have varied.

Fourth, there is a selection bias. The sample was limited to those who had left temporary housing, excluding those who were still in temporary housing and those who chose not to participate. This may have resulted in a small number of responses from very vulnerable groups and may have biased the results toward healthier respondents.

Fifth, this study set the scope of relocation at the elementary school district level, in line with the activity base of Kumamoto City, the affected area. Therefore, there may be limitations in generalization.

Conclusions

This study showed that daytime activities such as exercise habits, meal habits, and social capital, as well as the extent of relocation, were significantly associated with insomnia among the elderly affected by the Kumamoto earthquake.

Therefore, support for insomnia requires both informal supports to prevent isolation based on mutual community support and formal support for those suspected of insomnia. We hope that study will continue on effective ways of providing support in the future. It is assumed that individual and group support for subjects will be realized through collaboration between the two, mainly by health and medical welfare professionals in the community.

Supplemental Information