Ethics statement
This study was approved by the Ethics Committee of International Research Institute of Disaster Science, Tohoku University (ID: 2020-040), and research was performed in accordance with relevant guidelines/regulations. All study participants were informed that they would be conducting a disaster-related study and gave their voluntary informed consent to participate in the study. The study adhered to ethical standards for research on human subjects (informed consent and right to information, protection of personal information and guarantee of privacy, non-discrimination, rewards, and the ability to withdraw from the study at any stage).
Survey period
We conducted three surveys: June 25–26, 2020; September 25–26, 2020; February 10–12, 2021. A combined graph displaying the spread of COVID-19 infections in Japan and the survey period is shown in Fig. 1 A.
Participants
Overall, 1,000 Japanese participants were included (500 men and 500 women). In the first survey, their ages ranged from 20 to 89 years, with a mean age of 45.5 years (standard deviation: 15.02) and a median of 46.0 years (25th percentile: 32 and 75th percentile: 57). In the second survey, their ages ranged from 21 to 86 years, with a mean age of 54.2 years (standard deviation: 12.95) and a median of 54.2 years (25th percentile: 44 and 75th percentile: 64). In the third survey, their ages ranged from 20 to 86 years, with a mean age of 45.2 years (standard deviation: 14.71) and a median of 46 years (25th percentile: 33 and 75th percentile: 57).
Of the total participants in Japan, 500 (50%) experienced a natural disaster prior to the COVID-19 pandemic and 500 (50%) did not. Of the total participants in the US, 127 (63%) experienced a natural disaster prior to the COVID-19 pandemic and 73 (37%) did not. The natural disasters experienced are shown in Figure 3. The participant profile data are shown in Supplementary Figure S1.
Overall, 200 participants were included (100 men and 100 women) as the US sample. Their ages ranged from 23 to 70 years, with a mean age of 43.2 years (standard deviation: 12.7) and a median of 40.0 years (25th percentile: 33 and 75th percentile: 52).
Assessments
Depression Anxiety Stress Scales-21 (DASS-21)
The DASS-21 [7] is a measure of distress commonly caused by anxiety, depression, and stress, and characterised as a distinct syndrome. As the measure is not specific to diagnosis, it is appropriate for use in broad clinical populations. This measure is suitable in a wide range of clinical and research settings where it is necessary to measure the interplay of different forms of emotional distress [8]. In this study, participants were asked to choose how each of the items on the DASS-21 applied to them over the past week. The items are scored on a 4-point scale ranging from 0 (did not apply to me at all) to 3 (applied to me very much or most of the time). Higher scores indicate more frequent symptomatology. Seven items are included in each scale: depression (such as, “I couldn't seem to experience any positive feeling at all”), anxiety (such as, “I experienced breathing difficulty”), and stress (such as, “I found it hard to wind down”) [7].
Impact of Event Scale-Revised (IES-R)
The IES-R is a 22-item scale that measures how distressing each item was during the past week. It is rated from 0 (not at all) to 4 (extremely). The scale has three subscales: intrusion, avoidance, and hyper-arousal. The Japanese-translated version, which has been evaluated among different populations, has sensitivity and specificity values that range 0.75–0.89 and 0.71–0.93, respectively, at a cut-off of 25 for partial PTSD diagnosis [9]. In this study, this value was used to define participants who were significantly symptomatic from disasters they experienced in the past.
Lubben Social Network Scale–6 (LSNS-6)
Over the past 30 years, the LSNS has gained popularity as a valid and efficient tool for assessing social networks and isolation risks (https://www.bc.edu/content/bcweb/schools/ssw/sites/lubben/description.html). The LSNS-6 has been a viable tool for the assessment of social networks [10].
Posttraumatic Growth Inventory Short-form Japanese version (PTG-SF-J)
The Posttraumatic Growth Inventory [9] was developed to assess positive psychological changes reported by persons who have experienced traumatic events. This scale includes the following factors: New Possibilities, Relating to Others, Personal Strength, Spiritual Change, and Appreciation of Life. The Posttraumatic Growth Inventory is modestly related to optimism and extroversion. This scale is useful to determine how individuals perceive their own growth when reflecting on past trauma, which includes reconstructing or strengthening their perceptions of self, others, and the meaning of events [11]. The translation and validation were conducted by Taku et al. [12] Responses are rated on a 6-point Likert-type scale ranging from 0 (did not experience) to 5 (to a very great degree) [11]. Owing to qualitative differences among the subscales [11], subscale scores were used in the analyses.
In the present study, we used a form of this scale with fewer question items, the PTGI-SF-J [9]. We asked participants, “To what extent have the following changes occurred to your own way of life (or values) as a result of the new coronavirus epidemic? Please choose the option that best applies to you.” We also asked, “To what extent has your life (or values) changed as a result of the epidemic?”
Correlation between PTGI-SF scores and other factors
Supplementary Figure S2a shows Spearman’s correlation analysis of age and PTG score, which was R2 = 0.0003 in the first study. In the second survey, the correlation between age and PTG score was R2 = 0.0017. In the third survey, the correlation was R2 = 0.0011.
The correlation coefficients were low and not significant. Supplementary Figure S2b shows the correlation between social ties and PTG by LSNS-6 score in all three surveys. Spearman’s correlation analysis showed that the correlation between LSNS-6 score and PTGI-SF score was R2 = 0.0498 in the first survey, R2 = 0.0632 in the second survey, and R2 = 0.0777 in the third survey. The correlation coefficients were also low and not significant.
Supplementary Figures S2c and S2d show the relationship between the number of family members living together and the PTGI-SF scores. Although there was no significant difference between the first and second surveys, the PTGI-SF scores seemed to increase with the number of family members living together. In the third survey, the PTGI-SF score was significantly higher when participants lived by themselves or when the number of people living with them was six or more rather than three to five. It is unclear, however, whether the increase in number of family members directly caused the increase in PTG or whether it caused the increase in depression and anxiety, which in turn caused the increase in PTG, as indicated by the Kruskal-Wallis test. Further, it is unclear whether PTG increased directly because of the increase in number of family members or secondarily because of the increase in depression and anxiety.
Supplementary Figures S2c and S2d show the results of correlations among the study variables based on the US sample, suggesting that PTG is reported higher as participants are younger, need more support, and live with 3-5 family members rather than alone.
Statistics
Non-parametric testing was used for comparison. Prediction One (Sony Network Communications Inc. Japan) was used to create a predictive model of PTG scores using all variables examined in the web survey. First, the entire data set was read and randomly split in half as an internal training and cross-validation data set. Internal cross-validation was performed to create the best predictive model.