Abstract
We prospectively examined whether community-level social capital plays a significant role in developing posttraumatic growth (PTG) among older survivors of the 2011 Japan Earthquake and Tsunami. The baseline survey was conducted 7 months before the disaster among residents of a city located 80 km west of the earthquake epicenter. The survey inquired about participants’ health status and social capital (informal socializing and social participation, as well as social cohesion). Approximately 2.5 years after the disaster, we surveyed older survivors to assess their disaster experiences. A follow-up survey in 2022 inquired about PTG in the 11 years following experiences of the disaster (n = 1819). Multilevel linear regression analysis showed that predisaster community-level informal socializing and social participation was associated with higher PTG scores (coefficient = 0.25; 95% CI, 0.02-0.47). In cross-classified multilevel regression, maintenance of higher community-level informal socializing and social participation during the postdisaster period was associated with higher scores of PTG (coefficient = 0.22; 95% CI, 0.07-0.37). Predisaster community-level informal socializing and social participation were associated with higher PTG scores among older survivors. Interventions encouraging social interactions among neighbors may be effective in promoting PTG of survivors after natural disasters.
Keywords: social capital, posttraumatic growth, natural disaster, older survivors
Introduction
Large-scale natural disasters have been shown to produce persistent adverse impacts on the health and well-being of survivors. Older people are especially vulnerable in the aftermath of natural disasters, because of preexisting functional limitations, comorbidities, and/or social isolation.1,2
Nevertheless, despite the experience of considerable psychological strain, a subset of survivors has been shown to achieve personal growth through struggling with adversities. Posttraumatic growth (PTG) is a positive psychological change that is sometimes witnessed after traumatic events and comprises 5 major dimensions: improved relationships with others, increased personal strength, identification of new possibilities, positive spiritual changes, and increased appreciation of life.3 We recently suggested cross-sectional associations between PTG and both better functional capacity and cognitive function among older survivors of the 2011 Japan Earthquake and Tsunami.4 In addition, a meta-analysis demonstrated that approximately 38% of trauma-exposed adults over the age of 60 had moderate to high levels of PTG.5
Research suggests that the preservation of social connections in the community (ie, Social Capital: SC) plays a significant role in achieving PTG.6 Communities endowed with higher stocks of social capital (ie, stronger bonds of trust between community members, as well as norms of mutual assistance) are better equipped to cope with the devastating consequences of disaster.7 A cross-sectional study suggested that the availability of assistance from family, friends/neighbors, and acquaintances/coworkers during disasters was positively associated with higher PTG among 216 adults who experienced Hurricane Katrina (2005) and the Deepwater Horizon oil spill (2010).8 However, few studies have examined prospectively the association between social capital and PTG.
Social capital is often separated into cognitive and structural components. The former refers to how people perceive social relations in their community (eg, trust in others, norms of mutual help, and community attachment), while the latter component captures what people actually do (eg, informal socializing with their neighbors and participation in social activities).9 We previously reported that individual and community-level informal socializing and social participation helped to prevent cognitive decline among older residents who experienced the 2011 Great East Japan Earthquake and Tsunami.10,11
There remains an additional question of whether community-level social capital can also account for individual variations in posttraumatic growth. Even if an individual did not personally socialize with friends or participate in social activities, they could still benefit from living in a community where their neighbors participate and socialize with each other. That is, we hypothesized a contextual effect of the community social capital on posttraumatic growth among older survivors. To examine this hypothesis, we analyzed data from a 11.5-year postdisaster follow-up of a cohort of community-dwelling older adults who survived the 2011 Great East Japan Earthquake and Tsunami.
Methods
Study participants
The Japan Gerontological Evaluation Study (JAGES) was established in 2010 as a nationwide sample of community-dwelling residents aged 65 years or older. One of the field sites of the JAGES cohort is based in the city of Iwanuma (total population 44 187 in 2010). Iwanuma City was administratively divided into 99 subdistricts—called gyousei-ku—which formed the basis of social organization. These administrative units were judged to be suitable for the purposes of defining a community in our study because they form the basis on which residents were organized for conducting periodic disaster preparedness drills together and local volunteer fire brigades before the disaster. We mailed questionnaires to every resident aged 65 years or older in August 2010 (n = 8576), using the official residential register of Iwanuma City. The survey inquired about personal characteristics, social capital, and health status. The response rate was 59.0% (n = 5058), which is comparable to other surveys of community-dwelling residents.12
The earthquake and tsunami occurred on March 11, 2011, 7 months after the baseline survey was completed. Iwanuma City is a coastal municipality located approximately 80 km west of the earthquake epicenter, making it directly in the line of the tsunami. That disaster killed 180 residents, damaged 5542 housing units, and inundated 48% of the land area (Figure 1).
Figure 1.
Map of a tsunami-inundated area in Iwanuma City, Japan.
We conducted follow-up surveys of survivors every 3 years until 2022 and used data obtained from the 2010, 2013, and 2022 surveys (ie, a 3-wave panel) (Figure 2), because all variables to examine our hypothesis were included in these waves.
Figure 2.
Participant flow in this study.
In October 2013, we conducted a follow-up survey of survivors approximately 2.5 years after the disaster. The survey gathered information about personal experiences during the disaster and updated information about health status.
The detailed flowchart of the analytical sample is presented in Figure 2. Of the 4380 eligible participants from the baseline survey, we managed to recontact 3594 individuals (participation rate: 82.1%). Of the participants, 27 individuals were excluded due to incompletely signed informed consent forms.
Approximately 9 years after the second wave (in November 2022), we administered the fourth survey wave. We updated their health status and inquired about posttraumatic growth. As shown in Figure 2, we collected data from 1819 individuals (participation rate: 69.9%; 3-wave follow-up rate of baseline respondents: 36.0%) and dropped 21 respondents due to invalid identification. Respondents with physical and/or cognitive impairments (n = 33) were excluded from this study, as cognitive and physical functioning are associated with reduced social participation at baseline13 as well as severe trauma-related psychopathologies,14 which may hinder the achievement of PTG.4 In addition, we excluded 2 respondents who were observed once in 2 small communities to reduce estimation bias in the multilevel model. Finally, the analytic sample was 1763 respondents from 97 communities at the baseline, ranging from 2 to 67 respondents per community (median: 24 people).
The survey protocol was reviewed and approved by the Ethics Committees on Research of Human Subjects at the Harvard T.H. Chan School of Public Health (23143), Tohoku University (21-40, 24-29), Nihon Fukushi University (10-05, 13-14), and Chiba University (2493). Each respondent provided written informed consent to participate in the study.
Outcome variable
Our primary outcome is PTG in the 2022 survey. PTG was measured using the Short Form of the Expanded Version of the Posttraumatic Growth Inventory.15 This scale assessed 5 dimensions of PTG: improved relationship with others, increased personal strength, identification of new possibilities, positive spiritual changes, and increased appreciation of life.
We instructed participants to “Please tell us to what extent your way of living has changed as a result of experiencing the 2011 Japan Earthquake and Tsunami” and asked them the following questions. Improved relationship with others was measured using the following 2 items: “I have a greater sense of closeness with others” and “I more clearly see that I can count on people in times of trouble.” Increased personal strength was measured using the following 2 questions: “I know better how to handle difficulties” and “I discovered that I’m stronger than I thought I was.” The following 2 items were used to assess the identification of new possibilities: “New opportunities are available which wouldn’t have been otherwise” and “I established a new path for my life.” Positive spiritual changes were measured using the following 2 questions: “I feel more connected with all of existence” and “I have a greater sense of harmony with the world.” We used the following 2 questions to measure the increased appreciation of life: “I have a greater appreciation for the value of my own life” and “I can better appreciate each day.” Responses were ordered along a 6-point Likert scale (1, not at all; 6, very strong).
We implemented a confirmatory factor analysis to check the construct validity of PTG measured by 10 questions. As shown in Figure S1, we found a reasonable 5-factor solution with acceptable goodness-of-fit indices (comparative fit index 0.981, root mean square error of approximation [RMSEA] 0.080, standardized root mean square residual 0.024, and Tucker-Lewis Index 0.966). Although the RMSEA score of 0.080 indicated only a fair degree of fit,16 other fit indices and factor loadings have reasonably high scores, indicating an acceptable fit of our model.
We calculated the arithmetical mean of the responses to the 10 items to create a total PTG score, with higher scores indicating a well-developed PTG. This scale exhibited good internal consistency (Cronbach’s α = 0.94).
Explanatory variables
Our primary explanatory variable is community-level social capital, divided into 2 subdimensions: cognitive and structural.9 The cognitive component refers to perceptions of social relations (eg, trust, mutual help, and community attachment), while the structural component reflects actual behaviors (eg, informal socializing and participation in social activities). The cognitive dimension was assessed based on answers to questionnaire items about residents’ perceptions of trust in the community, norms of mutual help, and community attachment. These were evaluated using the following questions: “Do you think that people living in your community can be trusted in general?” (Generalized trust); “Do you think people living in your community try to help others in most situations?” (Mutual help); and “How attached are you to the community in which you live?” (Community attachment). Responses were ordered along a 5-point Likert scale, with 1 indicating not at all and 5 indicating strong agreement.
The structural dimension of social capital was measured by the frequency of meeting with friends, the number of friends with whom the respondent met during the past month, and the frequency of participating in sports and hobby clubs per week. Respondents were asked, “How often do you see your friends?” (frequency meeting with friends: 1, rarely, to 6, almost every day); “How many friends/acquaintances have you seen over the past month? Count the same person as one, no matter how many times you have seen him/her” (number of friends met: 1, none, to 5, 10 or more); “How often do you attend sports club activities?” (frequency participating in sports clubs: 1, none, to 6, almost every day); and “How often do you attend hobby groups?” (frequency participating in hobby groups: 1, none, to 6, almost every day).
Although we derived these subdimensions of social capital from theory, our measurements have not been psychometrically validated. Therefore, we conducted an exploratory factor analysis to check whether our original items aligned with the factor structures. The results supported our approach to create 2 subscales for social capital (see Table S1). We calculated the arithmetic means for each subdimension, showing good internal consistency with Cronbach’s α values of 0.76 and 0.71, respectively.
Community-level social capital scores were then derived by aggregating the individual responses clustered by the 97 gyousei-ku units. We assumed that this aggregated individual perception of social capital is an appropriate scale for community-level social capital,17 since “gyousei-ku” was a unit for conducting social activities such as periodic disaster preparedness drills.
Another primary exposure variable of interest is personal experiences of trauma in the disaster. In the 2013 survey, respondents retrospectively reported housing damage and loss of relatives or friends. Housing damage in Iwanuma City was independently assessed by property inspectors and classified into 5 levels according to objective criteria: (1) no damage, (2) partial damage, (3) minor, (4) major, and (5) complete destruction. Respondents also reported loss of loved ones (close friends and/or relatives).
Covariates
According to previous studies,10,11 we selected several baseline demographic variables as potential confounding variables for the association of social capital with PTG and depressive symptoms: age, sex, equivalized household income, educational attainment, divorce or bereavement, working status, and depressive symptoms.
Household income was equivalized by dividing the gross income by the square root of the number of household members.
Depressive symptoms were measured by the Japanese version of the Geriatric Depression Scale–15.18 We created a binary variable representing lower risk (4 points and under) and higher risk (5 points and over).19
Statistical analysis
A multilevel linear regression analysis was implemented to examine associations between the baseline community-level cognitive and structural social capital and PTG. The 2-level model specified individuals at level 1 nested within small communities at level 2 as follows:
 |
where 
is the PTG score of individual i who resided in community j, 
represents the average score of PTG across individual i living in community j, 
represents the coefficients for individual-level predictors 
, 
represents the coefficients for community-level predictors 
, 
is the random effect of the community j, and 
is individual-level residual.
In a sensitivity analysis, we examined whether changes in community-level social capital over time were associated with the level of PTG. In our sample, 267 respondents relocated to another community until the 2022 survey. That is, these respondents were cross-nested within 2 different communities through the survey times (ie, 2010 and 2022). Therefore, we employed a cross-classified multilevel model to use an adequate model, considering the cross-nested structure as follows20:
 |
where 
is the PTG score of individual i who resided in community j and k, 
represents the grand mean of PTG score, 
represents the coefficients for individual-level predictors 
, 
represents the coefficients for community-level predictors 
(community j), 
represents the coefficients for community-level predictors 
(community k), 
and 
are the random effects of each community, and 
is the individual-level residual.
We also implemented multiple imputation for missing values using the Markov chain Monte Carlo method that assumes missingness at random. Five data sets were created and combined each result of analyses using the Stata command “mi estimate.” All analyses were performed using STATA version 17.0 (STATA Corp LP).
Results
Table 1 presents descriptive statistics for the outcomes, explanatory variables, and potential covariates. The average PTG score was 3.15 points with a 1.13 standard deviation (SD) in the 2022 survey, ranging from 1 to 6 points. In the first postdisaster survey (2013), respondents reported the level of housing damage they experienced on a 5-point scale, ranging from 1 (no damage) to 5 (housing loss). The average level of housing damage was 1.87 with a 0.98 SD. Additionally, 41% reported loss of loved ones during the disaster.
Table 1.
Characteristics of the analytic samples (n = 1763).
| Characteristic | Value |
|---|---|
| Outcome | |
| Posttraumatic growth, mean (SD) | 3.15 (1.13) |
| Missing, n (%) | 357 (20.2) |
| Explanatory variable | |
| Individual-level social cohesion, mean (SD) | 3.78 (0.63) |
| Missing, n (%) | 61 (3.46) |
| Individual-level social participation and informal socializing, mean (SD) | 2.98 (1.01) |
| Missing, n (%) | 293 (16.6) |
| Housing damage, mean (SD) | 1.87 (0.98) |
| Missing, n (%) | 40 (2.3) |
| Loss of loved ones, n (%) | 716 (40.6) |
| Baseline characteristics | |
| Age, mean (SD), y | 71.0 (4.75) |
| Female, n (%) | 1001 (56.8) |
| Educational attainment (1: <6 y to 4: ≧13 y), mean (SD) | 2.93 (0.73) |
| Missing, n (%) | 49 (2.8) |
| Equivalized household income (10 000 JPY), mean (SD) | 239.67 (137.54) |
| Missing, n (%) | 238 (13.5) |
| Divorce/bereavement, n (%) | 362 (20.5) |
| Missing, n (%) | 49 (2.8) |
| Working, n (%) | 363 (20.6) |
| Missing, n (%) | 152 (8.6) |
| Depressive symptoms, ≧5p of GDS-15 score, n (%) | 407 (23.1) |
| Missing, n (%) | 189 (10.7) |
Abbreviations: GDS, Geriatric Depression Scale; JPY, Japanese yen.
As shown in Table 2, the multilevel model showed that the experience of housing damage and loss of loved ones was associated with higher levels of PTG (coefficient = 0.18; 95% CI, 0.10-0.26; coefficient = 0.21; 95% CI, 0.09-0.32, respectively, in model 2). The baseline individual perception of social cohesion was positively associated with PTG (coefficient = 0.16; 95% CI, 0.07-0.26, in model 2), whereas community-level (aggregated) social cohesion was not significantly associated (coefficient = 0.28; 95% CI, –0.03 to 0.59, in model 3), although the point estimate was in the expected direction. Baseline informal socializing and social participation were significantly associated with higher PTG scores at both the individual and community levels (coefficient = 0.08; 95% CI, 0.02-0.14, in model 4; coefficient = 0.28; 95% CI, 0.06-0.49, in model 5, respectively).
Table 2.
Associations of baseline social capital and posttraumatic growth in the 2022 survey.
| Cognitive SC | Structural SC | Both SC | ||||
|---|---|---|---|---|---|---|
| Characteristic | Model1: null | Model 2: individual level | Model 3: community level | Model 4: individual level | Model 5: community level | Model 6: both levels of SC |
| Coef. (95% CI) | Coef. (95% CI) | Coef. (95% CI) | Coef. (95% CI) | Coef. (95% CI) | Coef. (95% CI) | |
| Disaster experiences | ||||||
| Housing damage | 0.18 (0.10 to 0.26) | 0.18 (0.10 to 0.26) | 0.18 (0.10 to 0.26) | 0.18 (0.10 to 0.26) | 0.18 (0.10 to 0.26) | |
| Loss of loved ones | 0.21 (0.09 to 0.32) | 0.21 (0.09 to 0.32) | 0.21 (0.09 to 0.32) | 0.21 (0.10 to 0.33) | 0.20 (0.09 to 0.32) | |
| Baseline variables | ||||||
| Age | –0.01 (–0.03 to 0.01) | –0.01 (–0.03 to 0.01) | –0.01 (–0.02 to 0.01) | –0.01 (–0.03 to 0.01) | –0.01 (–0.03 to 0.01) | |
| Female | 0.07 (–0.06 to 0.19) | 0.06 (–0.06 to 0.19) | 0.03 (–0.10 to 0.16) | 0.03 (–0.09 to 0.16) | 0.05 (–0.08 to 0.17) | |
| Education | 0.06 (–0.03 to 0.16) | 0.06 (–0.03 to 0.16) | 0.05 (–0.04 to 0.15) | 0.05 (–0.04 to 0.15) | 0.05 (–0.04 to 0.15) | |
| Equivalized household income (10 000 JPY) | –0.01 (–0.02 to –0.01) | –0.01 (–0.02 to –0.01) | –0.01 (–0.02 to –0.01) | –0.01 (–0.02 to –0.01) | –0.01 (–0.02 to –0.01) | |
| Divorce/bereavement | 0.05 (–0.11 to 0.20) | 0.05 (–0.10 to 0.20) | 0.04 (–0.11 to 0.19) | 0.03 (–0.12 to 0.18) | 0.04 (–0.11 to 0.19) | |
| Working | 0.06 (–0.14 to 0.27) | 0.06 (–0.14 to 0.27) | 0.08 (–0.13 to 0.29) | 0.09 (–0.13 to 0.30) | 0.08 (–0.13 to 0.29) | |
| Severe depressive symptomsa | –0.19 (–0.34 to –0.04) | –0.19 (–0.34 to –0.04) | –0.20 (–0.35 to –0.05) | –0.20 (–0.34 to –0.05) | –0.16 (–0.32 to –0.01) | |
| Individual-level SC | ||||||
| Social cohesion | 0.16 (0.07 to 0.26) | 0.16 (0.06 to 0.25) | 0.15 (0.05 to 0.24) | |||
| Social participation and informal socializing | 0.08 (0.02 to 0.14) | 0.07 (0.01 to 0.13) | 0.06 (–0.01 to 0.13) | |||
| Community-level SC | ||||||
| Social cohesion | 0.28 (–0.03 to 0.59) | 0.17 (–0.15 to 0.49) | ||||
| Social participation and informal socializing | 0.28 (0.06 to 0.49) | 0.25 (0.02 to 0.47) | ||||
| Constant | 3.17 (3.10 to 3.25) | 3.01 (1.82 to 4.21) | 2.60 (0.94 to 4.27) | 3.33 (2.15 to 4.51) | 2.79 (1.58 to 3.99) | 2.35 (0.72 to 3.98) |
| Community-level variance (SE) | 0.200 (0.049) | 0.078 (0.069) | 0.072 (0.077) | 0.077 (0.068) | 0.054 (0.164) | 0.015 (1.298) |
| Individual-level residual error variance (SE) | 1.117 (0.025) | 1.094 (0.025) | 1.094 (0.025) | 1.095 (0.026) | 1.095 (0.261) | 1.091 (0.026) |
Abbreviations: JPY, Japanese yen; SC, Social Capital.
≧5 points of Geriatric Depression Scale–15.
When we included both dimensions of social capital into the same model (model 6), individual-level perceptions of social cohesion and community-level informal socializing/social participation remained significantly associated with higher PTG scores (coefficient = 0.15; 95% CI, 0.05-0.24; coefficient = 0.25; 95% CI, 0.02-0.47, respectively). From model 1 (the null model) through model 6, the variance partition coefficients (VPCs), calculated as the proportion of variance in PTG attributed to community-level factors over the total variance (community and individual levels combined), decreased from 15% to 1%. This implies that most of the variance in PTG is now attributed to individual-level factors rather than community-level differences.
In a sensitivity analysis, we examined whether changes in individual and community-level social capital over time were associated with PTG scores. In our sample, 267 respondents relocated to another community by the time of the last (2022) survey. We used a cross-classified multilevel model to capture the fact that these respondents were cross-nested in 2 different communities throughout the survey period. As shown in Table 3, decreasing individual perceptions of social cohesion (2010 high–2022 low), as well as stably high perceptions of social cohesion (2010 high–2022 high), were associated with higher PTG scores (coefficient = 0.37; 95% CI, 0.16-0.58; coefficient = 0.39; 95% CI, 0.24-0.53, respectively, in model 2), while none of the trajectories of community-level social cohesion were significantly associated with PTG (model 3). For the structural dimension of SC, decreasing (2010 high–2022 low) and stably high (2010 high–2022 high) levels of individual-level informal socializing/social participation were associated with higher levels of PTG (coefficient = 0.30; 95% CI, 0.12-0.48; coefficient = 0.30; 95% CI, 0.14-0.46, respectively, in model 4), while stably high (2010 high–2022 high) community-level informal socializing/social participation was associated with higher levels of PTG (coefficient = 0.22; 95% CI, 0.08–0.36, in model 5).
Table 3.
Associations of change in social capital over time and posttraumatic growth in the 2022 survey.
| Cognitive SC | Structural SC | Both SC | ||||
|---|---|---|---|---|---|---|
| Model1: null | Model 2: individual level | Model 3: community level | Model 4: individual level | Model 5: community level | Model 6: both levels of SC | |
| Coef. (95% CI) | Coef. (95% CI) | Coef. (95% CI) | Coef. (95% CI) | Coef. (95% CI) | Coef. (95% CI) | |
| Disaster experiences | ||||||
| Housing damage | 0.19 (0.11 to 0.27) | 0.19 (0.11 to 0.26) | 0.18 (0.10 to 0.26) | 0.18 (0.10 to 0.26) | 0.18 (0.11 to 0.26) | |
| Loss of loved ones | 0.20 (0.09 to 0.32) | 0.20 (0.09 to 0.32) | 0.20 (0.09 to 0.32) | 0.21 (0.09 to 0.32) | 0.19 (0.08 to 0.31) | |
| Baseline variables | ||||||
| Age | –0.01 (–0.03 to 0.01) | –0.01 (–0.03 to 0.01) | –0.01 (–0.02 to 0.01) | –0.01 (–0.02 to 0.01) | –0.01 (–0.02 to 0.01) | |
| Female | 0.07 (–0.05 to 0.20) | 0.07 (–0.05 to 0.20) | 0.02 (–0.10 to 0.14) | 0.02 (–0.10 to 0.14) | 0.05 (–0.07 to 0.17) | |
| Education | 0.06 (–0.03 to 0.16) | 0.06 (–0.03 to 0.16) | 0.06 (–0.03 to 0.15) | 0.06 (–0.04 to 0.15) | 0.05 (–0.04 to 0.15) | |
| Equivalized household income (10 000 JPY) | –0.01 (–0.02 to –0.01) | –0.01 (–0.02 to –0.01) | –0.01 (–0.02 to –0.01) | –0.01 (–0.02 to –0.01) | –0.01 (–0.02 to –0.01) | |
| Divorce/bereavement | 0.05 (–0.10 to 0.20) | 0.05 (–0.10 to 0.20) | 0.04 (–0.11 to 0.19) | 0.03 (–0.12 to 0.18) | 0.04 (–0.11 to 0.19) | |
| Working | 0.06 (–0.14 to 0.26) | 0.06 (–0.14 to 0.26) | 0.06 (–0.15 to 0.27) | 0.07 (–0.14 to 0.28) | 0.06 (–0.14 to 0.26) | |
| Severe depressive symptomsa | –0.17 (–0.32 to –0.02) | –0.17 (–0.32 to –0.02) | –0.20 (–0.35 to –0.06) | –0.20 (–0.35 to –0.05) | –0.15 (–0.30 to 0.01) | |
| Individual-level SCb | ||||||
| Social cohesion | ||||||
| 2010 low–2022 high | 0.09 (–0.09 to 0.27) | 0.09 (–0.10 to 0.27) | 0.08 (–0.11 to 0.26) | |||
| 2010 high–2022 low | 0.37 (0.16 to 0.58) | 0.37 (0.15 to 0.59) | 0.34 (0.12 to 0.56) | |||
| 2010 high–2022 high | 0.39 (0.24 to 0.53) | 0.38 (0.23 to 0.53) | 0.36 (0.20 to 0.51) | |||
| Social participation and informal socializing | ||||||
| 2010 low–2022 high | 0.01 (–0.20 to 0.20) | –0.01 (–0.21 to 0.19) | –0.03 (–0.24 to 0.17) | |||
| 2010 high–2022 low | 0.30 (0.12 to 0.48) | 0.29 (0.12 to 0.47) | 0.25 (0.08 to 0.43) | |||
| 2010 high–2022 high | 0.30 (0.14 to 0.46) | 0.27 (0.11 to 0.43) | 0.24 (0.08 to 0.39) | |||
| Community-level SCb | ||||||
| Social cohesion | ||||||
| 2010 low–2022 high | 0.01 (–0.20 to 0.20) | 0.01 (–0.19 to 0.20) | ||||
| 2010 high–2022 low | 0.01 (–0.19 to 0.20) | –0.01 (–0.21 to 0.18) | ||||
| 2010 high–2022 high | 0.03 (–0.12 to 0.18) | –0.04 (–0.20 to 0.11) | ||||
| Social participation and informal socializing | ||||||
| 2010 low–2022 high | 0.06 (–0.10 to 0.22) | 0.07 (–0.09 to 0.22) | ||||
| 2010 high–2022 low | 0.16 (–0.02 to 0.33) | 0.15 (–0.03 to 0.32) | ||||
| 2010 high–2022 high | 0.22 (0.08 to 0.36) | 0.22 (0.07 to 0.37) | ||||
| Constant | 3.17 (3.10 to 3.25) | 3.38 (2.21 to 4.55) | 3.37 (2.20 to 4.54) | 2.98 (1.86 to 4.09) | 2.93 (1.80 to 4.06) | 2.87 (1.73 to 4.01) |
| 2022 community-level variance (SE) | 0.001 (0.004) | 0.001 (0.038) | 0.001 (0.031) | 0.001 (0.001) | 0.001 (0.001) | 0.001 (0.001) |
| 2010 community-level variance (SE) | 0.199 (0.053) | 0.074 (0.092) | 0.066 (0.116) | 0.078 (0.071) | 0.001 (0.131) | 0.001 (0.002) |
| Individual-level variance (SE) | 1.117 (0.025) | 1.085 (0.026) | 1.085 (0.026) | 1.089 (0.026) | 1.087 (0.026) | 1.075 (0.025) |
Abbreviations: JPY, Japanese yen; SC, Social Capital.
≧5 points of Geriatric Depression Scale–15.
The reference group is 2010 low–2022 low.
When we included both dimensions of social capital in the same model (model 6), decreasing and stably high individual perceptions of social cohesion and informal socializing/social participation were positively associated with PTG (decreased social cohesion: coefficient = 0.34; 95% CI, 0.12-0.56; stably high social cohesion: coefficient = 0.36; 95% CI, 0.20-0.51; decreased informal socializing/social participation: coefficient = 0.25; 95% CI, 0.08-0.43; stably high informal socializing/social participation = 0.24; 95% CI, 0.08-0.39). Moreover, stably high community-level informal socializing/social participation was associated with higher scores of PTG (coefficient = 0.22; 95% CI, 0.07-0.37). From model 1 (the null model) through model 6, the VPC for the community-level variance in PTG (based on the respondents’ geo-location prior to the disaster) decreased from 15.1% to 0.1%. By contrast, there was no significant community variance in PTG based on the respondents’ location in 2022.
Discussion
The experiences of disaster-related trauma (housing damage and loss of loved ones) were associated with higher PTG scores 11 years after the disaster. Baseline individual perceptions of social cohesion and community-aggregated informal socializing/social participation were significantly associated with higher PTG scores. In addition, both decreasing and stably high individual perceptions of social cohesion and informal socializing/social participation over time were positively associated with PTG. Importantly, stably high community-level informal socializing/social participation was associated with higher scores of PTG. Through these analyses, our explanatory variables explained over 90% of the community-level variance in PTG.
Older people gradually lose their social connections, primarily due to the loss of peripheral relationships. Individuals tend to prefer emotionally meaningful interactions with close relationships over less emotional communication with casual acquaintances.21 Our respondents were able to continue exchanging emotional and instrumental social support through the 2022 survey, provided they had established close relationships in their neighborhood before the disaster. Consequently, the group with decreased social participation and informal socializing may have shown a positive association with PTG.
Additional descriptive analysis showed that “appreciation of life” had the highest PTG score (3.76), while “new possibilities” had the lowest (2.76) (Figure S2). A systematic review6 found that appreciation of life was consistently increased among patients with cancer, veterans, and Holocaust survivors. Our respondents likely recognized life’s fragility due to the disaster, leading to a greater appreciation of small moments. Furthermore, volunteer groups and nonprofit organizations supported victims, and the Iwanuma government built permanent housing for survivors in 2016. These social support and recovery efforts may have further enhanced their sense of appreciation.
According to the theoretical model of PTG,22 a certain level of struggle is crucial for achieving PTG. We demonstrated that disaster experiences were positively associated with the level of PTG, which supports the theory. However, the question remains as to why some survivors achieved PTG, whereas others failed to overcome their traumatic experiences.
Our results suggest that community-level social capital is crucial for achieving PTG among older survivors in the aftermath of natural disasters. This contextual effect suggests that even survivors who did not personally socialize with neighbors or participate in social activities could benefit from living in a community of socially active residents.
A plausible mechanism by which community-level social capital enhances PTG is through the exchange of social support among community members. Social support not only alleviates mental distress caused by trauma but also aids in reconstructing belief systems, fostering growth.23 The growing field of social epigenomics seeks to understand the interaction between social environments and DNA methylation.24 For instance, neighborhood environments characterized by low social cohesion, reduced aesthetic quality, and decreased safety have been linked to higher methylation levels in stress- and inflammation-related genes in a study of 1264 adults across 6 US districts.25 In a cohort of 3338 postmenopausal US women, lower social support amplified the effect of stressful events on accelerated epigenetic aging.26 Considering these findings, social capital before the disaster may have mitigated epigenetic changes related to mental distress and inflammation, contributing to posttraumatic growth.
One potential approach to preserving social connections after a disaster is exemplified by the method by which survivors were relocated from emergency shelters to temporary housing. In the aftermath of the 2011 disaster, Iwanuma City officials offered 2 choices for displaced residents: either they could individually relocate to temporary housing via a lottery system, or they could choose to relocate together with their neighbors. The former method allowed individuals to leave the emergency shelters sooner, but it had the effect of separating people from their neighbors. By contrast, the “group relocation” method helped to preserve social connections that existed before the disaster.27 The municipal authority closed the temporary housing complex and opened a permanent housing community in April 2016. At that time, residents were again offered a choice of moving together with their neighbors. These measures may have helped to preserve community-level social capital that could lead to promoting the exchange of social support among survivors, thereby enhancing the potential for PTG.
A major strength of the present study is the availability of predisaster information about social capital and confounders. This allowed us to avoid recall bias, an issue unaddressed in most previous studies.
Our study also has some limitations. First, the one-time assessment of PTG could have resulted in underestimating the associations between social capital and PTG. A systematic review of PTG studies suggested that victims of natural disasters can achieve PTG within a few years after the disaster.28 However, we did not measure PTG until the 2022 survey. Therefore, the association between social capital and PTG might have been underestimated compared to a hypothetical scenario where PTG had been measured since the 2013 survey. Second, measurement bias may arise due to unvalidated social capital measures. We have demonstrated that community-level social capital is protectively associated with cognitive function using the same items,29 supporting the scale’s predictive validity. Additionally, the Cronbach’s α values were acceptable,30 although slightly below the ideal 0.80. Notably, the 2022 Cronbach’s α values (0.72 cognitive, 0.74 structural) were similar to those from 2010. Test-retest correlations (0.47 and 0.53, both P < .01) also indicate good reliability and stability, considering the 12-year interval. However, further validation with different cohorts is needed. Third, attrition bias may have influenced the results. As shown in Table S2, we compared the baseline characteristics of the valid respondents in the 2010 survey (n = 4957) with those of attrition cases from the 2013 (n = 1390) and 2022 (n = 3194) surveys. There were no significant differences in the prevalence of characteristics, including the risk of depressive symptoms, among these groups. However, by the 2022 survey, 507 of 520 people (97.5%) aged 85 and older had been lost, while 1294 of 2648 individuals (48.9%) aged 65 to 74 had dropped out over the 12-year period. While attrition due to aging is inevitable in longitudinal studies involving older adults, our results may be biased toward younger and healthier respondents, who may have been more socially active and better able to overcome their traumatic experiences. Fourth, unmeasured confounding related to COVID-19 may have occurred due to stay-at-home measures. As shown in Figure S3, individual social participation and informal socializing decreased by 0.5 points by 2022, while social cohesion remained stable. In addition to the 12 years of aging during the follow-up period, COVID-19 may have impacted the frequency of socializing. Since we did not measure COVID-19–related experiences in the 2022 survey, this may have resulted in confounding bias when analyzing the relationship between changes in social capital and PTG.
In conclusion, our study demonstrated that community-level informal socializing and social participation were protective factors associated with the level of PTG among older survivors. Interventions encouraging social interactions among neighbors may be effective in developing PTG of survivors after natural disasters. For example, group relocation may have helped to preserve community-level social capital that could lead to promoting the exchange of social support among survivors, thereby enhancing the potential for PTG.
Supplementary Material
Acknowledgments
We thank Professor Jun Aida at the Department of Oral Health Promotion, Graduate School of Medical and Dental Sciences, Tokyo Medical and Dental University, for invaluable advice.
Contributor Information
Hiroyuki Hikichi, Department of Public Health, Kitasato University School of Medicine, 1-15-1 Kitazato, Minami, Sagamihara, Kanagawa 252-0374, Japan.
Katsunari Kondo, Center for Preventive Medical Sciences, Chiba University, 1-8-1 Inohana, Chuo, Chiba, Chiba 260-8670, Japan; Center for Gerontology and Social Science, National Center for Geriatrics and Gerontology, 7-430 Morioka, Obu, Aichi 474-8511, Japan.
Ichiro Kawachi, Department of Social and Behavioral Sciences, Harvard T.H. Chan School of Public Health, 677 Huntington Avenue, Boston, MA 02115, United States.
Supplementary material
Supplementary material is available at American Journal of Epidemiology online.
Funding
This study was supported by the National Institutes of Health (R01 AG042463); Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (KAKENHI 23243070, KAKENHI 22390400, KAKENHI 20H00557, KAKENHI 24390469, and KAKENHI 22 K10564); Health Labour Sciences Research Grant from the Japanese Ministry of Health, Labour, and Welfare (H24-Choju-Wakate-009); the Strategic Research Foundation Grant-Aided Project for Private Universities from the Japanese Ministry of Education, Culture, Sports, Science, and Technology (S0991035); and Japan Agency for Medical Research and Development (JP17dk0110017, JP18dk0110027, JP18ls0110002, JP18le0110009, JP19dk0110034, JP19dk0110037, and JP20dk0110034).
Conflict of interest
The authors declare no conflicts of interest.
Disclaimer
The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Data availability
All data needed to evaluate this study’s conclusions are present in the article and/or the Supplementary Materials. The JAGES data used in this study will be made available upon request, as per NIH data access policies. The authors require the applicant to submit an analysis proposal to be reviewed by an internal JAGES committee to avoid duplication. Confidentiality concerns prevent us from depositing our data in a public repository. Authors requesting access to the Iwanuma data need to contact the principal investigator of the parent cohort (K.K.) and the Iwanuma substudy principal investigator (I.K.) in writing. Proposals submitted by outside investigators will be discussed during the monthly investigators’ meeting to ensure that there is no overlap with ongoing analyses. If approval to access the data is granted, the JAGES researchers will request the outside investigator to help financially support our data manager’s time to prepare the data for outside use.
References
- 1. Shiba K, Hikichi H, Okuzono SS, et al. Long-term associations between disaster-related home loss and health and well-being of older survivors: nine years after the 2011 Great East Japan Earthquake and Tsunami. Environ Health Perspect. 2022;130(7):77001. 10.1289/EHP10903 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Hikichi H, Shiba K, Aida J, et al. Association between sense of coherence and health and well-being among older survivors of a natural disaster: a prospective outcome-wide study. Sci Rep. 2023;13(1):16385. 10.1038/s41598-023-43672-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Tedeschi RG, Calhoun LG. Posttraumatic growth: conceptual foundations and empirical evidence. Psychol Inq. 2004;15(1):1-18. 10.1207/s15327965pli1501_01 [DOI] [Google Scholar]
- 4. Hikichi H, Taku K, Aida J, et al. Longitudinal associations between posttraumatic stress and posttraumatic growth among older adults 11 years after a disaster. Epidemiol Psychiatr Sci. 2024;33:e33. 10.1017/S2045796024000362 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Wu X, Kaminga AC, Dai W, et al. The prevalence of moderate-to-high posttraumatic growth: a systematic review and meta-analysis. J Affect Disord. 2019;243:408-415. 10.1016/j.jad.2018.09.023 [DOI] [PubMed] [Google Scholar]
- 6. Kadri A, Gracey F, Leddy A. What factors are associated with posttraumatic growth in older adults? A systematic review. Clin Gerontol. 2022;48(1):1-18. 10.1080/07317115.2022.2034200 [DOI] [PubMed] [Google Scholar]
- 7. Aldrich DP. Building Resilience : Social Capital in Post-Disaster Recovery. University of Chicago Press; 2012. 10.7208/chicago/9780226012896.001.0001 [DOI] [Google Scholar]
- 8. Lee J, Blackmon BJ, Lee JY, et al. An exploration of posttraumatic growth, loneliness, depression, resilience, and social capital among survivors of Hurricane Katrina and the Deepwater Horizon oil spill. J Community Psychol. 2019;47(2):356-370. 10.1002/jcop.22125 [DOI] [PubMed] [Google Scholar]
- 9. Harpham T. The measurement of community social capital through surveys. In: Kawachi I, Subramanian SV, Kim D, eds. Social Capital and Health. Springer; 2008:51-62. 10.1007/978-0-387-71311-3_3 [DOI] [Google Scholar]
- 10. Hikichi H, Tsuboya T, Aida J, et al. Social capital and cognitive decline in the aftermath of a natural disaster: a natural experiment from the 2011 Great East Japan Earthquake and Tsunami. Lancet Planet Health. 2017;1(3):e105-e113. 10.1016/S2542-5196(17)30041-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Hikichi H, Aida J, Matsuyama Y, et al. Community-level social capital and cognitive decline after a natural disaster: a natural experiment from the 2011 Great East Japan Earthquake and Tsunami. Soc Sci Med. 2020;257:111981. 10.1016/j.socscimed.2018.09.057 [DOI] [PubMed] [Google Scholar]
- 12. Santos-Eggimann B, Cuénoud P, Spagnoli J, et al. Prevalence of frailty in middle-aged and older community-dwelling Europeans living in 10 countries. J Gerontol Ser A Biol Sci Med Sci. 2009;64(6):675-681. 10.1093/gerona/glp012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Hashidate H, Shimada H, Fujisawa Y, et al. An overview of social participation in older adults: concepts and assessments. Phys Ther Res. 2021;24(2):85-97. 10.1298/ptr.R0013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. van Zelst WH, de Beurs E, Beekman ATF, et al. Prevalence and risk factors of posttraumatic stress disorder in older adults. Psychother Psychosom. 2003;72(6):333-342. 10.1159/000073030 [DOI] [PubMed] [Google Scholar]
- 15. Oshiro R, Soejima T, Kita S, et al. Reliability and validity of the Japanese version of the short form of the Expanded Version of the Posttraumatic Growth Inventory (PTGI-X-SF-J): a cross-sectional study. Int J Environ Res Public Health. 2023;20(11):5965. 10.3390/ijerph20115965 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Hu L, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Struct Equ Model. 1999;6(1):1-55. 10.1080/10705519909540118 [DOI] [Google Scholar]
- 17. Diez Roux AV. Neighborhoods and health: where are we and where do we go from here? Rev Epidemiol Sante Publique. 2007;55(1):13-21. 10.1016/j.respe.2006.12.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Sugishita K, Sugishita M, Hemmi I, et al. A validity and reliability study of the Japanese version of the Geriatric Depression Scale 15 (GDS-15-J). Clin Gerontol. 2017;40(4):233-240. 10.1080/07317115.2016.1199452 [DOI] [PubMed] [Google Scholar]
- 19. Weintraub D, Oehlberg KA, Katz IR, et al. Test characteristics of the 15-item geriatric depression scale and Hamilton depression rating scale in Parkinson disease. Am J Geriatr Psychiatry. 2006;14(2):169-175. 10.1097/01.JGP.0000192488.66049.4b [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Naess O, Leyland AH. Analysing the effect of area of residence over the life course in multilevel epidemiology. Scand J Public Health. 2010;38(5 Suppl):119-126. 10.1177/1403494810384646 [DOI] [PubMed] [Google Scholar]
- 21. Rook KS, Charles ST. Close social ties and health in later life: strengths and vulnerabilities. Am Psychol. 2017;72(6):567-577. 10.1037/amp0000104 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Long LJ, Phillips CA, Glover N, et al. A meta-analytic review of the relationship between posttraumatic growth, anxiety, and depression. J Happiness Stud. 2021;22(8):3703-3728. 10.1007/s10902-021-00370-9 [DOI] [Google Scholar]
- 23. Ning J, Tang X, Shi H, et al. Social support and posttraumatic growth: a meta-analysis. J Affect Disord. 2023;320:117-132. 10.1016/j.jad.2022.09.114 [DOI] [PubMed] [Google Scholar]
- 24. Martin CL, Ghastine L, Lodge EK, et al. Understanding health inequalities through the lens of social epigenetics. Annu Rev Public Health. 2022;43(1):235-254. 10.1146/annurev-publhealth-052020-105613 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Smith JA, Zhao W, Wang X, et al. Neighborhood characteristics influence DNA methylation of genes involved in stress response and inflammation: the multi-ethnic study of atherosclerosis. Epigenetics. 2017;12(8):662-673. 10.1080/15592294.2017.1341026 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Skinner HG, Palma-Gudiel H, Stewart JD, et al. Stressful life events, social support, and epigenetic aging in the Women’s Health Initiative. J Am Geriatr Soc. 2024;72(2):349-360. 10.1111/jgs.18726 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Hikichi H, Sawada Y, Tsuboya T, et al. Residential relocation and change in social capital: a natural experiment from the 2011 Great East Japan Earthquake and Tsunami. Sci Adv. 2017;3(7):e1700426. 10.1126/sciadv.1700426 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Amiri H, Nakhaee N, Nagyova I, et al. Posttraumatic growth after earthquake: a systematic review and meta-analysis. Int J Soc Psychiatry. 2021;67(7):867-877. 10.1177/0020764021995856 [DOI] [PubMed] [Google Scholar]
- 29. Hikichi H, Aida J, Matsuyama Y, et al. Community-level social capital and cognitive decline after a natural disaster: a natural experiment from the 2011 Great East Japan Earthquake and Tsunami. Soc Sci Med. 2020;257:111981. 10.1016/j.socscimed.2018.09.057 [DOI] [PubMed] [Google Scholar]
- 30. Tavakol M, Dennick R. Making sense of Cronbach’s alpha. Int. J Med Educ. 2011;2:53-55. 10.5116/ijme.4dfb.8dfd [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
All data needed to evaluate this study’s conclusions are present in the article and/or the Supplementary Materials. The JAGES data used in this study will be made available upon request, as per NIH data access policies. The authors require the applicant to submit an analysis proposal to be reviewed by an internal JAGES committee to avoid duplication. Confidentiality concerns prevent us from depositing our data in a public repository. Authors requesting access to the Iwanuma data need to contact the principal investigator of the parent cohort (K.K.) and the Iwanuma substudy principal investigator (I.K.) in writing. Proposals submitted by outside investigators will be discussed during the monthly investigators’ meeting to ensure that there is no overlap with ongoing analyses. If approval to access the data is granted, the JAGES researchers will request the outside investigator to help financially support our data manager’s time to prepare the data for outside use.