Health survey on victims
We conducted a health survey on residents living in the Ogatsu and Oshika districts of Ishinomaki city, Miyagi Prefecture, which were severely affected by the GEJE. The details of this health survey have been reported elsewhere [25,26,27]. The health survey on victims consisted of a physical examination and self-reported questionnaires, which were carried out every year since the GEJE. The questionnaires are composed of various items: residence status, medical history, physical health status, smoking habits, drinking habits, dietary habits, sleep status (based on the 8-item Athens Insomnia Scale [AIS]) [28], psychological distress (based on the 6-item Kessler Psychological Distress Scale [K6]) [29], working status, social networks (based on the 6-item Lubben Social Network Scale [LSNS-6]) [30], social trust, norms, and networks.
Ethical issues
The study protocol was approved by the Institutional Review Board of the Tohoku University Graduate School of Medicine (approval No.: 2011–92, 2017–1-069). Consent to participate in the study was obtained from the participants either face-to-face or via the signed self-administered questionnaires.
Study participants
Of the 5065 residents in this area, 1398 provided valid responses to the initial health survey, which was carried out from June to August 2011. For the present analysis, we excluded residents who did not provide consent to participate (N = 104) and who had missing data regarding social trust (N = 1). Finally, 1293 adults were included in the study cohort. Follow-up data were then collected every year after the GEJE (Fig. 1).
Exposure data
We obtained data on social trust by asking whether “People in my community can be trusted” at the first health survey. The responses were given on a five-point scale (“Strongly agree”, “Somewhat agree”, “Neither agree nor disagree”, “Somewhat disagree”, and “Strongly disagree”). The data were then analyzed with regard to the relation between social trust and sleep disorders, and based on the results, the participants were categorized into two groups: 1) High; those who answered “Strongly agree” or “Somewhat agree”, and 2) Low; those who answered “Neither agree nor disagree”, “Somewhat disagree”, or “Strongly disagree. To assess the reliability of the questionnaire, we calculated the Spearman correlation coefficient (95% confidence interval) between the two surveys (first survey and second survey). The Spearman correlation coefficient for social trust was 0.47 (p < 0.01).
Outcome variables
The outcome variables for sleep disorders were measured using the AIS [28], which is a self-assessment instrument to report any sleep disorders experienced by the participant, provided that they occurred at least three times per week during the last month. The AIS is composed of eight items rated from 0 to 3, with a total score ranging from 0 to 24 [28].
Other measurements
Residence status was assessed in the questionnaire by selecting one of the following responses: “same as that before the GEJE (no relocation)”, “shelter”, “temporary house”, “rental house”, “family’s house or friend’s house” and “others”. Resident status after the GEJE was categorized into two groups: same housing as that before the GEJE, or relocation after the GEJE.
Economic status was assessed by responses to the following question: “How do you feel about your current house-hold economic status?” The respondents were asked to choose one of the following responses: “very hard”, “hard”, “a little hard” or “normal”. Economic status was categorized into two groups: “very hard” and “hard”, or “a little hard” and “normal”.
Psychological distress was measured using the K6, which is composed of six items rated from 0 to 4, with a total score ranging from 0 to 24 [29].
Social networks were measured using the LSNS-6 [30, 31], which is composed of six items rated from 0 to 5, with a total score ranging from 0 to 30. .
Statistical analysis
First, we analyzed mean the AIS scores during the 6 years since the GEJE across the social trust groups. The mean AIS scores at every health survey were compared using t tests. We then investigated the association between social trust and sleep disorders using a linear mixed model (random effects model). To estimate the effects of sleep status according to social trust at the first health survey, the time from the GEJE (in months) and social trust × time from the GEJE (in months) were entered into the model.
We considered the following variables as possible confounders: sex, age (continuous variable), resident status after the GEJE (same as that before the GEJE, or relocating after the GEJE), and economic status (very hard, hard, a little hard, or normal) (Model 1). These items were chosen as covariates because living environment and economic status may contribute to sleep status. We also added the LSNS-6 score as a social factor because social isolation may affect the association between social trust and sleep status (Model 2) [32, 33]. We further added the K6 score because psychological distress may strongly affect the association between social trust and sleep status (Model 3).
We conducted further stratified analyses according to age group (< 65 years and ≥ 65 years), because age-related differences have been observed in terms of one’s connection with the local community [27].
All P values were two-sided, and P < 0.05 was considered statistically significant. All statistical analyses were performed using the SAS statistical software package (version 9.4; SAS Institute Inc., Cary, NC, USA).