The data for this study were derived from the Mental Health and Lifestyle Survey, which is included in the Fukushima Health Management Survey, a detailed description of which can be found elsewhere . Briefly, a self-administered questionnaire on mental health and various lifestyle habits according to age category (0–6 years, 7–15 years, and ≥16 years) was delivered in January 2012 to all residents who lived in the evacuation zones caused by the radiation accident on March 11, 2011, in Fukushima Prefecture. The evacuation zone was a government-designated area (20 km radius) around the nuclear power plant. Of the target population of 180,604 participants aged ≥16 years, 73,433 (40.7%) responded.
We excluded the following participants from this study: those who did not provide information about subjective health (n = 1879) or current living arrangements (n = 14,859), those who reported “other” for current living arrangements (n = 2761), and those aged <20 or ≥65 years (n = 20,584). We excluded the latter set of participants because this group included many participants who were not working due to their status as students or retirees. Thus, we analyzed a total of 33,350 participants (14,913 men and 18,437 women).
The study protocol was approved by the Ethics Committee of Fukushima Medical University. Participants who returned the self-administered questionnaires were considered to have consented to participate.
The self-administered questionnaire included questions about disaster-induced changes in SES that assessed participants’ living and working situations. Change in living arrangements was assessed through each participant’s response to the question: “How have your living arrangements changes?” The participants indicated their current residence by circling one of the following: “evacuation shelter”, “temporary housing”, “rental housing/apartment”, “relative’s home”, or “own home”. We defined participants as having a “change in living arrangements” if their response was “evacuation shelter”, “temporary housing”, or “rental housing/apartment”. Change in working conditions was assessed with the question: “Has your work situation changed as a result of the natural disaster or nuclear accident?”. If participants responded “yes”, they checked the following items as appropriate: “became unemployed” or “income has decreased”. We defined participants as having a disaster-induced change in SES if they provided any of the above responses (i.e., “evacuation shelter”, “temporary housing”, “rental housing/apartment”, “became unemployed”, or “income has decreased”).
The self-administered questionnaire included a question on subjective health. Subjective health was assessed through each participant’s response to the question: “Describe your current state of health”. The participants chose one of the following answers: very good (men 5.6%; women 3.9%), good (men 18.3%; women 13.8%), normal (men 61.3%; women 66.5%), poor (men 13.4%; women 14.4%), or very poor (men 1.4%; women 1.4%). “Poor subjective health” was defined by the answers “poor” or “very poor”.
We regarded smoking, alcohol consumption, sleep, participation in recreation and community activities, and exercise as lifestyle-related factors. These factors were assessed using the following questions and associated choices: “Do you smoke cigarettes (excluding cigars and pipes)?” (“never smoke”, “quit”, or “current smoker”); “Do you drink alcohol?” (“don’t drink or only rarely (less than once/month)”, “quit”, or “drink (at least once a month)”); “Are you satisfied with the quality of your sleep over the past month (regardless of sleep duration)?” (“satisfied”, “slightly dissatisfied”, “quite dissatisfied”, or “very dissatisfied or haven’t slept at all”); “Do you participate in recreation (karaoke or gateball, etc.) and community activities (festivals, etc.)?” (“never or rarely”, “sometimes”, or “often”); and “Do you exercise regularly?” (“almost every day”, “2–4 times/week”, “once/week”, or “almost never”).
We used Poisson regression with robust error variance to derive prevalence ratios (PRs) and 95% confidence intervals (CIs) of poor subjective health according to disaster-induced changes in SES and to adjust for potential confounding factors because the prevalence of poor subjective health was not rare (≥10%) . We calculated using the SAS software package, version 9.3 (Cary, NC, USA). Participants who did not undergo disaster-induced changes in SES (i.e., “living in relative’s home” or “own home” and did not “become unemployed” and “decrease income”) were selected as the reference group. All P values were two-tailed, and differences at P < 0.05 were accepted as statistically significant. In model 1, we considered the following variables to be potential confounding factors: age (5-year categories: 20–24, 25–29…up to and including 60–64 years), history of disease (yes or no: hypertension, diabetes, hyperlipidemia, stroke, heart disease, cancer, chronic hepatitis, pneumonia, bone fracture, or thyroid disease), history of mental illness (yes or no), activities of daily living (go shopping for daily necessities: can do by myself or can’t do by myself), education (elementary school・junior high school, high school, or vocational college/ junior college or university (4 years)・graduate school), and evacuation place (Fukushima or other prefectures). In model 2, we further adjusted for the following lifestyle-related factors: smoking (never smoked, quit, or current smoker), alcohol consumption (less than once/month, quit, or at least once a month), satisfaction of sleep (satisfied, slightly dissatisfied, or complaint (quite dissatisfied, or very dissatisfied or haven’t slept at all), participation in recreation and community activity (never or rarely, sometimes, or often), and regular exercise (almost every day, 2–4 times/week, or ≤1 time/week). Additionally, we repeated the analyses for each disaster-induced change in SES (i.e., change in living arrangements, became unemployed, and decrease income).
We calculated the percentage of excess risks explained by lifestyle-related factors as follows: (PRmodel 1 − PRmodel 2)/(PRmodel 1–1)) × 100 .