Introduction
On March 11, 2011, the Great East Japan Earthquake (GEJE) and subsequent tsunami devastated the northeastern coast of Japan. As of March 1, 2023, 22,318 people were recorded as dead or missing.1 Most deaths were due to drowning (92.4%).2 After the GEJE, some of the coastal municipalities were hit by a tsunami just 30 minutes after the earthquake occurred.3
In the case of a large earthquake accompanied by a tsunami, tsunami evacuation, especially between the onset of the earthquake and tsunami arrival, is a crucial factor in determining the fate of seaside community residents. In the Sanriku region of northern Japan, a rule called Tsunami-tendenko, which dictates people to “run for your life to the top of the hill and never mind others or even your family when the tsunami comes,” has been passed down through the generations. Disaster drills have long been considered an effective disaster preparedness activity. Our previous study reported that tsunami evacuation after the GEJE was significantly higher among those who had experienced tsunami evacuation drills before the GEJE than among those who had not.Reference Nakaya, Nemoto and Yi4 To promote evacuation and reduce human casualties in the event of a large-scale disaster in the future, it is important not only to conduct tsunami evacuation drills, but also to explore other factors related to evacuation and promote evacuation.
In 2017, Thompson et al. conducted a systematic review of 83 studies that examined evacuations in response to natural hazards.Reference Thompson, Garfin and Silver5 Regarding demographic factors, factors that promote evacuation include being female, ethnic group (Caucasian/white), and living with children. Factors that inhibit evacuation include the elderly and living with pets. On the other hand, educational background, income, and home ownership did not yield consistent results. Additionally, factors associated with evacuation include expectation for evacuation, having an evacuation plan, evacuation order issued by the government, length of residence in a disaster-prone area, previous disaster experiences, information of evacuation warning, high self-efficacy, and degree of the risk (scale or threat of disasters). This systematic review included five papers regarding evacuation in tsunami disaster.Reference Charnkol and Tanaboriboon6–Reference Rød, Botan and Holen10 Factors related to tsunami evacuation include gender,Reference Rød, Botan and Holen10 education,Reference Charnkol and Tanaboriboon6, Reference Charnkol and Tanaboriboon9 distance to nearest body of water,Reference Charnkol and Tanaboriboon6, Reference Charnkol and Tanaboriboon9 disaster knowledge,Reference Charnkol and Tanaboriboon6, Reference Charnkol and Tanaboriboon9 family members,Reference Charnkol and Tanaboriboon6, Reference Charnkol and Tanaboriboon9 occupation,Reference Charnkol and Tanaboriboon9 alert signals,Reference Lachman, Tatsuoka and Bonk7 community tenure,Reference Lindell, Prater and Gregg8 dialogue with experts,Reference Rød, Botan and Holen10 more concern about other issues,Reference Rød, Botan and Holen10 distrust in the experts,Reference Rød, Botan and Holen10 and compliance with the evacuation procedure.Reference Rød, Botan and Holen10 Furthermore, our previous study found that tsunami evacuation drill was associated with tsunami evacuation.Reference Nakaya, Nemoto and Yi4 Regarding tsunami disasters, further study is required on factors related to tsunami evacuation.
Natural hazards include not only tsunamis following earthquakes, but also hurricanes, floods, and volcanic eruptions. Adequate evacuation is crucial, not only to reduce fatal outcomes but also to prevent poor health consequences following a disaster. Here, we conducted a large-scale population study to comprehensively explore the factors that promoted tsunami evacuation after the GEJE. The Miyagi Prefecture in northern Japan, where this study was conducted, faces the Pacific Ocean and was one of the prefectures that was most damaged by the GEJE.
Material and Methods
Study Setting
For this cross-sectional study, data were obtained from the Tohoku Medical Megabank (TMM) Community-based Cohort Study (hereafter referred to as the TMM CommCohort Study) conducted in Miyagi Prefecture, northern Japan (these data were previously published elsewhere).Reference Kuriyama, Yaegashi and Nagami11–Reference Nakaya, Xie and Scheerder13 For the TMM CommCohort Study, participants were recruited for the baseline survey using two approaches between May 2013 and March 2016. Participants were recruited from sites of annual community health examinations conducted by local governments in the Miyagi Prefecture for insured persons aged 40-74 years (Type 1 survey). Additionally, seven community support Centre facilities were established in Miyagi Prefecture for voluntary admission-type recruitment and participant health assessment (Type 2 survey). In the baseline survey, blood and urine samples were collected, and self-administered questionnaires that included information on lifestyle habits, medical histories, and family relationships common to the Type 1 and Type 2 surveys were administered. This study only used data from participants in the Type 2 survey because the tsunami evacuation data was collected using a touchscreen computer, which was only used in the Type 2 survey.
Data Collection and Variables
Tsunami evacuation
Tsunami evacuation after the GEJE was assessed using a self-response questionnaire on a touchscreen PC. Participants were asked to choose one of the following three options: (1) I evacuated without preparation, (2) I evacuated after preparation, or (3) I did not evacuate. Participants were divided based on whether they evacuated (1 and 2) or not (3).Reference Nakaya, Nemoto and Yi4 Factors potentially related to tsunami evacuation were selected as comprehensively as possible based on previous studies,Reference Nakaya, Nemoto and Yi4–Reference Rød, Botan and Holen10 and are described as follows.
Basic characteristics
This study conducted a cross-sectional study on factors related to tsunami evacuation and explored whether various factors that existed before the GEJE were related to tsunami evacuation.
The basic characteristics included sex (male and female); age at baseline in years old (20-29, 30-39, 40-49, 50-59, 60-69, 70-); educational level: (low [elementary school, junior high school, high school], medium [vocational school, college, technical college], high [university, graduate school], and unknown; working status: (not working [looking for a job, not working], working [full-time employee, self-employed, temporary employee, part-time], unknown); and extent of damage to the home (no damage, small scale damage, large scale damage, not living in the disaster area, unknown). This study treated “extent of damage to the home” as a risk of health damage.
Socio-behavioral factors
The socio-behavioral factors included marital status (unmarried, married, unknown); presence of cohabitants (none, living with one other person, living with two or more other people, unknown); having pets (dogs [none, yes, unknown], cats [none, yes, unknown]); participating in tsunami drills before the GEJE (not experienced, experienced)Reference Nakaya, Nemoto and Yi4; participating in earthquake drills before the GEJE (not experienced, experienced)Reference Nakaya, Nemoto and Yi4; hearing about past earthquakes or tsunamis from a family member or an acquaintance before the GEJE (not experienced, experienced).Reference Nakaya, Nemoto and Yi4
Physical and lifestyle factors
We assessed the following factors: Physical activity (metabolic equivalents: METs) (Q1 [the lowest quartile] for METs, Q2, Q3, Q4 [the highest quartile], unknown)Reference Nakaya, Xie and Scheerder13; smoking (never smokers, current smokers, past smokers, unknown); alcohol drinking (never drinkers, current drinkers, past drinkers, unknown); 24-hour urinary sodium excretion in gram per day (<9.09 gram/day, ≥9.09 gram/day, unknown); body mass index in kg/m2 (<18.5, 18.5-24.9, ≥25.0, unknown); history of physical diseases using a self-administered questionnaire (vascular disease [any one of cerebral hemorrhage, cerebral infarction, subarachnoid hemorrhage, myocardial infarction, angina pectoris, aneurysm/aortic dissection, heart failure, Marfan syndrome, atrial fibrillation, pacemaker, and ventricular fibrillation]; respiratory disease [any one of bronchial asthma, chronic bronchitis, chronic sinusitis, and chronic obstructive pulmonary disease]; musculoskeletal disorders [any one of scoliosis, knee osteoarthritis, collagen disease, autoimmune disease, systemic lupus erythematosus, and rheumatism]; and cancer [absence or presence]). The estimated 24-hour urinary sodium excretion was calculated using the Tanaka formula,Reference Tanaka, Okamura and Miura14 as follows:
$$ \mathrm{Estimated}\;24\mathrm{HUNaV}\hskip0.24em \left[\mathrm{mEq}/\mathrm{day}\right]=21.98\times {\mathrm{XNa}}^{0.392} $$
$$ \begin{array}{c}\mathrm{PRCr}\;\left[\mathrm{mg}/\mathrm{day}\right]=-2.04\times \mathrm{age}+14.89\times \mathrm{weight}\;\left[\mathrm{kg}\right]\\ {}+16.14\times \mathrm{height}\;\left[\mathrm{cm}\right]-2244.45\end{array} $$
where XNa [mEq/day] = (SUNa [mEq/L] /SUCr [mg/dL] /10) × PRCr [mg/day]
24HUNaV= 24-h urinary sodium excretion
PRCr = predicted value of 24-h urinary creatinine excretion
SUNa = Na concentration in the spot urine
SUCr = creatinine concentration in the spot urine
The 24-hour urinary sodium excretion was classified into two categories—above the median (<9.09) and below the median (>9.09) (Supplement).
Mental factors
In this study, for the first time, we analyzed mental factors related to tsunami evacuation during large-scale disasters. We assessed the following: self-reported history of depression or post-traumatic stress disorder (PTSD) before GEJE.
Ethical Considerations
The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki, and the Ethics Committee of ToMMo, Tohoku University (Sendai, Japan) reviewed and approved the study protocol (first edition: 2012-4-617, latest edition: 2021-4-113).
Statistical Analyses
The objective variable was the presence or absence of evacuation after the GEJE, and the explanatory variables were comprehensively analyzed with reference to previous studies, including the socio-behavioral, physical, and mental factors mentioned above. Multivariate logistic regression analysis was performed, and odds ratios and 95% confidence intervals were calculated. Basic characteristics, such as sex, age, educational level, working status, and extent of damage to homes, were selected as covariates. The explanatory variables of interest were built into the models individually, and the covariates were built into all models. Address information on the municipality level was available. We classified the municipalities into two categories depending on whether they were coastal or not. We conducted a sensitivity analysis of only those surveyed in this study who lived in municipalities bordering the coast (n = 7917). Missing values were treated as an unknown category and used in the analysis. All statistical analyses were performed using SAS software (version 9.4; SAS Institute Inc., Cary, NC, USA).
Results
Participants in the TMM CommCohort Study
Overall, 17,575 individuals agreed to participate in a detailed survey (baseline survey 2013-2016) at the Community Support Center in the TMM Community Cohort Study, who continued to consent to this study until October 15, 2021. Of the 17,575 participants, we excluded 1640 who did not respond to the question about whether they had evacuated after the GEJE. Consequently, 15,935 participants were included in the final analysis.
Participants were divided into evacuees (n = 7365) and non-evacuees (n = 8570) after the GEJE. Females (70%) were the most common sex in this study, and 38% were in their 60s. Additionally, 48% of the participants were employed. Of the analyzed subjects, 69% reported that their residence was damaged on a small or large scale during the GEJE (Table 1).
Tsunami evacuation after the GEJE and the related basic characteristic factors (n = 15,935)
* Multivariate logistic regression analysis was performed, and multivariate odds ratios (MORs) and 95% confidence intervals (95% CIs) were calculated. Covariates were selected as possible confounders including sex, age, educational level, working status, and extent of damage to the homes.
† SD, standard deviation.
‡ Educational level was summarized as follows: low (elementary school, junior high school, or high school), medium (vocational school or college or technical college), and high (university or graduate school). Other was treated as unknown.
Multivariate logistic regression analysis showed that the evacuation odds ratio was significantly higher for those in their 30s than for those in their 20s, and the evacuation odds were significantly lower for those in their 60s and 70s. Regarding education level, the odds of evacuation were significantly lower in the moderate and high education groups than in the low education group. In terms of working status, the odds of evacuation were significantly higher among employed participants than among non-working participants. Furthermore, tsunami evacuation was significantly higher among participants with large- and small-scale damage than among those who did not report damage.
Tsunami Evacuation After the GEJE and the Related Socio-Behavioral Factors
Most of the participants were married (78%) and 8% lived alone. In addition, 18% of the participants had dogs and 11% had cats. Most participants had not experienced a tsunami (93%) or participated in earthquake evacuation drills (67%) prior to the GEJE. Twenty-two percent of participants were socially isolated (Table 2).
Tsunami evacuation after the GEJE and the related socio-behavioral factors (n = 15,935)
* Multivariate logistic analysis was performed, and multivariate odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated. In addition to each item, covariates were selected as possible confounders including sex, age, educational level, working status, and extent of damage to the homes.
Multivariate logistic regression analysis revealed that those who did not live in areas affected by the GEJE exhibited significantly lower tsunami evacuation rates. Married participants were significantly less likely to evacuate than unmarried participants. In addition, those who were living with one other person, or were living with two or more other people, were significantly less likely to evacuate than those who were living alone. Additionally, those who had dogs or cats were significantly less likely to evacuate than those who did not. Tsunami evacuation was significantly higher among those who had experienced tsunami or earthquake evacuation drills before the GEJE. Tsunami evacuation was also significantly higher among those who had heard of past earthquakes or tsunamis from family members or acquaintances before the GEJE.
Tsunami Evacuation After the GEJE and the Related Physical and Lifestyle Factors
Of the analyzed participants, 55% were drinkers at the time. And 24% of the participants were obese (BMI > 25.0). Less than 10% had a history of physical diseases: vascular disease (6%), respiratory disease (7%), musculoskeletal disorder (7%), and cancer (7%). (Table 3).
Tsunami evacuation after the GEJE and the related physical and lifestyle factors (n = 15,935)
* Multivariate logistic regression analysis was performed, and multivariate odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated. In addition to each item, covariates were selected as possible confounders including sex, age, educational level, working status, and extent of damage to the homes.
Vascular diseases: cerebral hemorrhage, cerebral infarction, subarachnoid hemorrhage, myocardial infarction, angina pectoris, aneurysm/aortic dissection, heart failure, Marfan syndrome, atrial fibrillation, pacemaker, and ventricular fibrillation.
Respiratory diseases: bronchial asthma, chronic bronchitis, chronic sinusitis, and COPD.
Musculoskeletal diseases: scoliosis, knee osteoarthritis, collagen disease, autoimmune disease, systemic lupus erythematosus, and rheumatism.
Physical activity (metabolic equivalents [METs]) (hours per week)
Q1: ~8.99, Q2: 9.00~74.99, Q3: 75.00~209.99, Q4: 210.00~
Compared with that in Q1 of the METs, tsunami evacuation was significantly lower in Q2-Q4. Other than physical activity, smoking, 24-hour urinary sodium excretion, and alcohol consumption did not show any relationship with tsunami evacuation. History of physical diseases was also not associated with tsunami evacuation.
Tsunami Evacuation After the GEJE and the Related Mental Factors
Of the analyzed participants, 3% had a self-reported history of depression before GEJE. Self-reported history of post-traumatic stress disorder (PTSD) before GEJE was observed in 0.23% of the participants (Table 4). The history of depression or PTSD before GEJE was not associated with tsunami evacuation. We conducted a sensitivity analysis of only those surveyed in this study who lived in municipalities bordering the coast (n = 7917). No significant changes were observed in point estimates of the main results (data not shown).
Tsunami evacuation after the GEJE and the related mental diseases (n = 15,935)
* Multivariate logistic regression analysis was performed, and multivariate odds ratios (ORs) and 95% confidence intervals (95% CIs) were calculated. In addition to each item, covariates were selected as possible confounders including sex, age, educational level, working status, and extent of damage to the homes.
Discussion
This study in Japan comprehensively analyzed the factors that promoted tsunami evacuation after the GEJE.
Tsunami Evacuation After the GEJE and the Related Basic Characteristic Factors
The promotion of evacuation among young people is consistent with the results of previous studies.Reference Van Willigen, Edwards and Lormand15–Reference Meyer, Broad and Orlove17 As people age, their experience of not having to evacuate may be reinforced. Women tend to evacuate more often than menReference Rød, Botan and Holen10, Reference Meyer, Broad and Orlove17, Reference Morss, Demuth and Lazo18 and our result showed significant sex differences. Previous studies have not found consistent results regarding education level.Reference Charnkol and Tanaboriboon6, Reference Charnkol and Tanaboriboon9, Reference Reininger, Raja Alam and Sanchez Carrasco16, Reference Morss, Demuth and Lazo18 This study did not support our hypothesis that a higher educational background would lead to a higher level of health and/or disaster knowledge, which would increase tsunami evacuation. Those with higher education tend to work in white-collar jobs, who were located far from the tsunami inundation zone, and might have discouraged them from evacuating. As for those with lower education, they tend to work in agriculture and construction, which are dependent on weather and natural conditions, which might lead to higher awareness of disaster risks.Reference Tatsuki, Hayashi and Zoleta-Nantes19 Evacuation was promoted among workers in this study. Currently, in Japan, the Fire Services Act obliges schools and workplaces to perform disaster drills (Article 8 of the Fire Service Act); therefore, workers promote tsunami evacuation. Tsunami evacuation was significantly higher among subjects who reported damage than among those who did not. Previous studies have shown that large-scale natural hazardsReference Thompson, Garfin and Silver5 or disasters close to disaster areasReference Thompson, Garfin and Silver5, Reference Charnkol and Tanaboriboon6, Reference Charnkol and Tanaboriboon9 promote evacuation, which is consistent with the results of this study.
Tsunami Evacuation After the GEJE and Related Socio-Behavioral Factors
Married participants and/or those living with many cohabitants were less likely to evacuate. Number of household members was revealed to be an important predictor of evacuation because of both evacuation decision and time to evacuate; larger households often took longer to do so than smaller households.Reference Charnkol and Tanaboriboon6, Reference Dixit, Pande and Radwan20, Reference Solís, Thomas and Letson21 It is possible that they did not evacuate because they prioritized caring for their cohabitants before evacuation.Reference Van Willigen, Edwards and Edwards22 We found that people with pets (dogs and cats) were less likely to evacuate. Previous studies have shown that people who own pets have more limited tsunami evacuation than people who do not own pets.Reference Mozumder, Raheem and Talberth23, Reference Petrolia and Bhattacharjee24 Considering pet allergies, it is possible that fewer people evacuated with their pets because there were fewer evacuation shelter environments where pets could be evacuated.Reference Schoos, Nwaru and Borres25 Securing evacuation sites specifically for pets or those where people can spend time together with their pets may increase tsunami evacuation rates.
Having experience with tsunami and earthquake evacuation drills before the GEJE outbreak and hearing about past earthquakes or tsunamis from family members or acquaintances led to the promotion of evacuation. Previous studies have reported that experience with tsunami evacuation drills promotes tsunami evacuation.Reference Nakaya, Nemoto and Yi4 In terms of its novelty, this study demonstrated that having experience with earthquake evacuation drills and hearing about such disasters in the past can lead to the promotion of tsunami evacuation.
Tsunami Evacuation After the GEJE and Related Physical and Lifestyle Factors
In this study, contrary to our hypotheses, tsunami evacuation was found to be inhibited in those who were physically active. The reason for these results is unknown, but it is possible that they did not evacuate because they thought they could escape at any time or that they did not evacuate because they were caring for others. Previous studies have reported that tsunami evacuation has increased significantly among people with a history of stroke.Reference Kulkarni, Gu and Tsai26 The reason for this was thought to be that it was necessary to evacuate early because it was difficult to move. In this study, tsunami evacuation was not associated with cardiovascular diseases, respiratory diseases, musculoskeletal diseases, or cancer history.
Tsunami Evacuation After the GEJE and Related Mental Factors
A few studies have examined mental factors as they relate to tsunami evacuation. Previous research has reported that high self-efficacy promotes tsunami evacuation.Reference Samaddar, Misra and Chatterjee27 It was unclear whether participants had depression or PTSD during the GEJE or how these conditions influenced tsunami evacuation rates. Future studies should analyze the impact of depression or PTSD at the GEJE on tsunami evacuation rates.
Strengths and Limitations
This study has a few strengths. First, this is the largest study on this topic to date (15,935 subjects). This study had the largest number of participants compared to the 83 papers included in the systematic review.Reference Thompson, Garfin and Silver5 Second, we comprehensively analyzed basic characteristics, socio-behavioral, physical, lifestyle, and mental factors. The value of this study, which comprehensively examined factors related to tsunami evacuation with one of the largest sample sizes, is extremely high.
However, there are also some limitations. First, because of the cross-sectional design, the factors before the GEJE could not be accurately assessed, and the factors investigated in this study may have changed after the evacuation. There is a high possibility that some of social, physical, and lifestyle factors will change because of the damage caused by the GEJE. Therefore, it is possible that these factors were misleading in this study. Second, since the participants of this study were survivors of the GEJE, it was not possible to assess population who lost their lives at the GEJE. Therefore, the results of the present study may have been underestimated. However, it is difficult to examine factors related to tsunami evacuation among residents of actual disaster-stricken areas.Reference Thompson, Garfin and Silver5 Third, we might not explore important items such as the factors associated with tsunami evacuation. The Protective Action Decision ModelReference Lindell and Perry28 and the Disaster SchemeReference Tatsuki, Hayashi and Zoleta-Nantes19 are related to evacuation. In the Protective Action Decision Model, evacuation is determined by threat, protective action, and stakeholder perceptions. In the Disaster Scheme, evacuation is related to perceived risk and normative beliefs. In future research, factors related to tsunami evacuation should be selected based on these theories. Fourth, this study included a relatively large proportion of women and older adults, and participants may have been more health conscious or in good health status, which might have led to selection bias. Finally, the survey was conducted using a self-reported questionnaire including the tsunami evacuation and the extent of damage to the home. Regarding the tsunami evacuation, it was not clear when people evacuated, whether it was before the tsunami arrived or a few days later. Even if the structure of their home was safe, some people might have evacuated because their lifelines were cut off. Also, regarding the extent of damage to the home, this answer contained a subjective aspect. Even if the participants live in a coastal municipality that was severely affected by the tsunami, the study included participants who answered, “I did not live in the disaster area” because the tsunami did not reach my location. We then conducted a sensitivity analysis of only those who lived in municipalities bordering the coast (n = 7917). No significant changes were observed in point estimates of the main results.
Conclusion
This cross-sectional study explored the factors that were associated with tsunami evacuation after the GEJE. It showed necessary to enhance the evacuation processes of elderly people, cohabitants, and households with pets. It was important to conduct evacuation drills and hold discussions about disasters within families and households. To prevent damage from tsunamis after earthquakes, it is important to address the modifiable factors related to this study. Notably, the number of people who are unable to evacuate due to physical, mental, or family circumstances associated with an aging society is increasing. In addition to encouraging people to evacuate during tsunamis, the most effective way to avoid the threat of tsunamis may be to live in a safe place that is less susceptible to tsunami damage.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/dmp.2025.10242.
Authors contribution
NN and KN: Methodology; NN, RN, and KN: Formal analysis; NN, NR, KN, RH, MK, CI, ST, TM, SK, and AH: Writing—original draft; AH: Supervision.
Acknowledgments
This study was supported in part by the Japan Agency for Medical Research and Development (AMED) [grant number JP21tm0124005]. The supercomputer system used in this study was provided by the Tohoku Medical Megabank Project and funded by AMED [grant number JP21tm0424601]. This study was also supported by a grant from JSPS KAKENHI [grant number: 23K09734].
Open access
