Disasters have catastrophic impacts on the populations that are affected by them. In 2020, 9.84 million people were affected by climate-related disasters, and more than 15 000 people lost their lives.Reference UNDRR 1 People affected by disasters encounter life-threatening events or lose a loved one; they also experience stress resulting from migration, substantial reduction of household income, and conflict with receiving disaster compensation. 2 , Reference Lee, Lee and Yoo 3 Stress caused by disasters negatively affects the mental health of impacted individuals by increasing their risk of somnipathy or PTSD.Reference Lee, Lee and Yoo 3 , Reference Kim and Lee 4
It has been reported that those with low socioeconomic status experience higher levels of psychological distress after a disaster.Reference Norris, Friedman and Watson 5 This is because individuals with a low socioeconomic status face a bigger loss of resources from disasters and have difficulty gaining resources for recovery. For instance, in a study of disaster-affected communities in Sri Lanka, the relative level of damage was higher in a low-income class, and their economic recovery took a longer time compared to those in a high-income class.Reference De Silva and Kawasaki 6 Additionally, individuals with low socioeconomic status have relatively little understanding of the process to apply for disaster relief funds. Subsequently, these individuals face significant barriers in obtaining resources for recovery.Reference Rovai 7 , Reference Fothergill and Peek 8 Moreover, people with low socioeconomic status generally have low levels of social support. In a longitudinal study on Mexico’s 1999 flood and landslides, individuals with a lower education level also had a lower level of perceived social support compared to more educated individuals. The gap in social support in relation to education levels increased over time.Reference Norris, Baker and Murphy 9
Meanwhile, research on the mental health of people impacted by disasters has revealed that mental health can be altered in various ways over time.Reference Norris, Tracy and Galea 10 People usually recover their mental health over time from the impact of a disaster, but some individuals may suffer from chronic stress (chronic dysfunction).Reference Norris, Tracy and Galea 10 , Reference Joshi, Cerdá, Ursano, Fullerton, Weisaeth and Raphael 11 Affected people can also experience psychological dysfunction even after psychological recovery (relapsing/remitting) or after a significant amount of time has passed since the disaster (delayed dysfunction).Reference Norris, Tracy and Galea 10
These findings suggest that the adverse effects of disasters on mental health can be stronger and longer-lasting in lower socioeconomic groups. However, a majority of previous research on the mental health of disaster-affected individuals has been conducted in a cross-sectional approach. 12 Therefore, the current study used a longitudinal approach to examine the post-disaster trajectories of mental health based on the income level of affected individuals, and to examine the factors that may influence these trajectories. The findings of this study will allow the identification of areas that require additional support for long-lasting recovery of the mental health of people affected by disasters. Moreover, this research will contribute to the development of health policies to mitigate potential health inequality resulting from disasters.
Methods
Study Participants and Data Collection
Our study used data from the second (2017) to the fourth (2019) “Long-Term Survey on the Change of Life of Disaster Victim” conducted by the National Disaster Management Research Institute (NDMI).Reference Park, Yoon and Lim 13 In our study, disaster-affected people are “those who received a relief fund after encountering [a] disaster.” The survey was conducted every year and covered a wide array of subjects such as physical and mental health, economic status, and social relationships of affected individuals. The samples were allocated and extracted based on the type of disaster, year of disaster, and region. The survey was conducted using Computer Assisted Personal Interviewing (CAPI). Responses from 699 adults aged 18 years and older who faced disaster damage between 2016 and 2017 and completed the entire survey were included in the analysis. Data were provided in anonymized form after approval by NDMI, and data analysis was performed after approval by the Institutional Review Board (IRB) of Seoul National University (SNU 21-09-058).
Measurement
Income level
South Korea is implementing the National Basic Livelihood Security Act to assure a minimum standard of living and self-support for people in need; people whose monthly household income is standard median income or less are selected as recipients of basic living. 14 Setting the median one-person household income from 2017 (approximately 1.65 million won) as a reference, those whose average monthly household income is less than 2 million won were classified as “low-income class,” and those with 2 million won or more were classified as “middle-high-income class.” 15
Post-traumatic stress (PTS)
Impact of Event Scale-Revised Korean version (IES-R-K) was used to measure the post-traumatic stress (PTS) of the disaster-affected people.Reference Eun, Kwon and Lee 16 The IES-R-K includes 22 self-administered questions, and the respondents assess the frequency of symptoms they experienced in the past week related to their trauma using a 5-point Likert scale (0: never to 4: very frequent). The range of scores is 0-88, with higher scores indicating more severe PTS. The reliability of the scale in our study was excellent (Cronbach’s α = 0.977 or higher).
Disaster damage and resources for disaster recovery
According to previous studies, disaster damage was measured by life and health threats, unmet health care needs, damage to houses or places of business, migration and separation, and financial hardship.Reference Lee, Lee and Yoo 3 , Reference Cho 17 Resources for disaster recovery include information on disaster relief services and recovery, government support (financial support, health care support, psychological support, support for environment and facility recovery), and social support.
Control variables included gender, age, marital status, type of disaster, and diagnosed mental disorders before the disaster, which are known to affect the PTS of affected individuals.Reference Norris, Friedman and Watson 5 , Reference Galea, Nandi and Vlahov 18 Detailed variables and measurements are described in Supplementary Table S1.
Analytic Methods
Descriptive analyses were conducted to provide information on participant characteristics and key variables (i.e., disaster damage and resources for disaster recovery). We also performed t-tests and chi-square tests to compare the differences between the low-income class and the middle-high-income class.
To examine the trajectories of mental health according to socioeconomic status, latent growth curve modeling (LGCM) and multigroup analysis (MGA) were conducted in this study. First, the mental health trajectories of the entire participants were analyzed using unconditional LGCM. Based on the estimated unconditional latent growth curve model, differences in trajectories according to income level were analyzed using MGA. Then, the differences in variables affecting the trajectories of each group were examined with the conditional model analysis. Full information maximum likelihood (FIML), which is known to be suitable for handling missing values in structural equation modeling, was used in our study.Reference Enders and Bandalos 19 Frequency analysis and descriptive statistics were used for data analysis with SPSS Statistics version 26.0, and LGCM was performed using SPSS AMOS version 26.0 in this study.
Results
Descriptive Analysis on Participants’ Characteristics and Disaster-Related Factors
The sociodemographic characteristics of the study participants are shown in Table 1. Among 699 participants, 49.4% were men and 50.6% were women. The mean age was 57.3 years. More participants were married (66.5%) than unmarried (33.5%). A small number of people (8.0%) suffered from or had been diagnosed with mental disorders prior to experiencing a disaster.
Table 1. Sociodemographic characteristics of participants
*** p < .001
There was no difference in the degree of disaster damage according to income level. A larger proportion of individuals in the low-income class (48.8%) did not have access to disaster-related information than those in the middle-high-income class (40.8%). Individuals with low incomes are more likely to perceive government-provided health care and psychological support as insufficient compared to those with middle or high incomes (p-value < 0.05). Furthermore, the level of social support was lower in the low-income class (M = 39.31) than in the middle-high-income class (M = 42.55). The PTS was higher in the low-income class than in the middle-high-income class for 3 years (Table 2).
Table 2. Differences in disaster damage, resources for recovery, and post-traumatic stress according to income level
* p < 0.05, *** p < .001
Post-Traumatic Stress Trajectories of Individuals Affected by Disasters According to Their Income Level
Unconditional model
Before MGA, PTS changes of the entire participants were examined through LGCM. To elucidate the optimal trajectories, among the 6 types of growth models presented by Kim (2009), the goodness of fit was compared among the no-growth model, linear growth model, and second-year growth model.Reference Kim 20 The result showed that the second-year growth model (a model that assumes that there is a change in PTS in first year and second year, but no significant change in second year and third year) best described the PTS trajectories of disaster-affected people (Supplementary Table S2). Therefore, the factor loadings of the intercept were fixed at 1, and the factor loadings of the slopes were fixed at 0, 1, and 1.
To determine whether PTS trajectories show different patterns according to income level, a multigroup analysis was conducted. The estimates of the intercept and slope of the PTS growth model between the two groups are shown in Table 3. The mean of the PTS intercept was 27.75 in the low-income class, which was higher than 17.19 in the middle-high-income class. The mean PTS slope was −5.73 in the low-income class and −4.49 in the middle-high-income class, showing a steep decrease in PTS in the low-income class. However, considering that the difference in mean of intercepts between the two groups is larger than the difference in mean of the slope, the mean PTS of the low-income class was still higher than that of the middle-high-income class in the second year (2018) and the third year (2019). Furthermore, all the variances of intercepts and slopes were significant in both groups, which suggests that there are individual differences in the PTS trajectories of the two groups (Figure 1).
Table 3. Results of the unconditional model
* p <.05, **p <.01, ***p <.001.
Figure 1 Post-traumatic stress trajectories by income level.
To determine whether the differences in intercepts and slopes between the two groups are statistically significant, the unconstrained model and the constrained model were compared. The results showed a significant difference in PTS trajectories between the low-income class and the middle-high-income class (
$ \Delta {\chi}^2 $
(df) = 71.336 (2), p-value < 0.001).
Conditional model
To determine whether the influence factors of PTS trajectories vary depending on income level, MGA was performed on the conditional model. Comparison of the unconstrained and constrained models through MGA revealed a statistically significant difference in PTS factor loadings of the low-income and middle-high-income classes (
$ \varDelta {\chi}^2 $
(df) = 61.972(32), p = 0.001). The paths, from predictors to PTS, in which the differences between low-income and middle-high-income classes exist are shown in Table 4.
Table 4. Estimates of relations between PTS trajectories and predictors according to income level
C.R.= Critical ratios for differences between parameters (* p < .05, **p < .01, *** p < .001) PTS: Post-traumatic stress
First, in the disaster damage variables, unmet health care needs and financial hardships were associated with PTS in both low-income and middle-high-income classes. The effect of unmet health care needs on the PTS slope showed a significant difference between the two groups (C.R. = −3.024). Life and health threats caused by the disaster exclusively affected the PTS of the low-income class, and the effect showed a significant difference between the two groups (C.R. = −2.051). Damage to houses or places of business had a significant effect on PTS intercept only in the middle-high-income class.
Among the variables concerning disaster recovery resources, the acquisition of information regarding disaster relief and recovery affected the PTS slope for both low-income and middle- and high-income classes. Additionally, health care support had an impact on the mental health of both groups. The intercept of PTS was lowered when the health care support was perceived to be sufficient. Higher social support was associated with lower initial levels of PTS, but only among those in the middle-high-income class.
Discussion
This study examined the differences in post-traumatic stress trajectories according to the income level of disaster-affected people and identified the factors affecting these trajectories. The study found that PTS decreased over time in both low-income and middle-high-income classes, but it remained consistently elevated in the low-income class compared to the middle-high-income class. Specifically, the PTS level of the low-income class in 2017 exceeded the cut-off value for post-traumatic stress disorder (24/25).Reference Eun, Kwon and Lee 16 These findings highlight that individuals with low socioeconomic status are at greater risk for long-term mental health challenges after experiencing a disaster. Moreover, the negative effects of disaster persist longer in those belonging to the low-income class.
The current study identified a number of factors that exert an influence on the PTS trajectories of the low-income and middle-high-income classes. In both groups, higher initial PTS levels were associated with unmet health care needs, while perceived sufficiency of government-provided health care services was linked to lower PTS levels. These findings indicate that access to health and medical care services following a disaster has an impact on the mental health of affected individuals, regardless of their income status. Similarly, studies conducted on the effects of Hurricane Katrina reported that unmet health care needs were associated with elevated stress levels, aggravation of health status, and daily life disruptions.Reference Foundation 21 The causes of unmet health care needs include financial burden, lack of time, and shortage of medical resources for treatment.Reference Foundation 21 These results show that providing sufficient health and medical services after a disaster is critical for the mental health management of affected people. In other words, securing and managing health resources to prevent unfulfilled health care needs in the aftermath of a disaster is essential in the field of public health. Previous studies pointed out that the disaster medical system, which includes emergency medicine and health care, has not been well established in South Korea.Reference Wang 22 , Reference Kim, Lee and Lee 23 The ways to improve the disaster medical system include, for example, the development and implementation of professional training courses for disaster response. In addition, it is required to establish a communication and cooperation system between health authorities responsible for disaster health care.
Household financial hardship was also associated with PTS intercept in both groups, which is in line with previous research findings showing that post-disaster economic hardship serves as a source of stress.Reference Lee, Lee and Yoo 3 , Reference Bonanno, Galea and Bucciarelli 24 Although this study found no direct impact of financial support on PTS trajectories, other research has shown that financial aid can alleviate mental health strain.Reference Daniel and Michaela 25 Moreover, Hallegatte et al. (2020) pointed out that using traditional metrics (i.e., the average monetary value of the assets) to assess the severity of a disaster can misrepresent the impacts of a disaster on individuals by underestimating the impacts of disasters on poor people.Reference Hallegatte, Vogt-Schilb and Rozenberg 26 As the belief in a just world and relative deprivation are associated with mental health, the discussion of the proper amount of financial compensation and equitable distribution for affected people is necessary.Reference Otto, Boos and Dalbert 27 , Reference Mishra and Carleton 28 Van der Geest proposed that individuals affected by disaster should receive humanitarian and needs-based compensation that considers their relative loss, instead of compensation that is based on absolute monetary value.Reference Van der Geest 29
The low-income class had higher PTS intercept when there was a life threat or experience of damage and disease. Previous research revealed that low-income individuals have lower levels of disaster preparedness, awareness, understanding, and compliance with disaster-warning messages compared to middle- and high-income individuals.Reference Fothergill and Peek 8 , Reference Najafi, Ardalan and Akbarisari 30 This indicates that the low-income class has a high risk of injury or death without being able to quickly evacuate when disasters occur. Hence, disaster-related education and emergency supplies should be offered to low-income classes in advance.
Social support has no effect on the PTS intercept and slope in the low-income class, but it affects the middle-high-income class. Considering that the level of social support was significantly lower in the low-income class than in the middle-high-income class in the current study, social support in the low-income class may not have been sufficient to decrease PTS. In general, people receive social support from their friends, family members, and acquaintances. However, when such networks are insufficient, it should be compensated by social support through local community capitals (e.g., welfare facilities, mental health service centers, etc.).Reference Kaniasty and Norris 31 As people with low socioeconomic status are reported to have low material and informational support after disaster,Reference Suar and Alat 32 , Reference Min, Joo and H-n 33 it is important to provide informational support for recovery and restoration and material support such as necessities and financial aid for them. Furthermore, both quantitative and qualitative aspects of social support should be considered when providing social support to affected people.Reference Shang, Kaniasty and Cowlishaw 34
Limitations
There are limitations to this study. First, the study participants were adults aged 18 and over, with an average age of 57.3 years old. Therefore, the findings in this study may not apply to communities with children and adolescents. Given that younger individuals experience disaster-related stress in a distinct way from adults, research on disaster mental health needs to be performed with more diverse age groups. Second, the participants in this study were recipients of a disaster relief fund, with the level of resources available for disaster recovery assessed through the sufficiency of disaster relief services and levels of social support they received. Subsequent research should encompass individuals who did not receive such support because they were unaware of or ineligible for these disaster relief funds or services. Additionally, disaster relief can be offered by various entities such as private enterprises, volunteer organizations, and civic organizations, which need to be taken into account in further studies. Third, despite the level of influencing factors, such as social support, which can change over time, only the values of predictors from the first year were used in the analysis. Future research should explore variations in these predictors to better understand their relationship with PTS trajectories.
Conclusion
This study examined post-traumatic stress trajectories according to the income level of the disaster-affected people and the factors affecting these trajectories. The results demonstrated a larger, more negative impact of disaster on the mental health of the low-income group. The current study also confirmed the need for improved health care services to prevent and alleviate PTS among individuals affected by disasters. Moreover, stronger social support networks should be established to better support the mental well-being of affected individuals.
Given the results of this study, the following implications were derived regarding health care policies for mental health recovery of disaster-affected communities. First, equity should be a focal point in managing the mental health of individuals affected by disasters. Specifically, resources for disaster recovery, including disaster relief services and compensation, should be appropriately allocated among all populations during the disaster recovery process. This will prevent exacerbation of existing social inequalities during times of disaster. Furthermore, efforts to dismantle inequality must be continued during times free of disaster to work toward a more sustainable and equitable future for all. Second, medical and health services should be provided in a timely and sufficient manner for individuals affected by disasters. Disaster threatens life and safety, the most basic human needs, and causes psychological stress in affected people. Therefore, it is essential to better prepare health care systems dedicated to disaster preparation and recovery. This can be achieved through collaboration and communication among disaster-related health care agencies. Last, local communities should expand social support networks to provide adequate social support for individuals who face difficulties in obtaining support from their existing individual relationships.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/dmp.2024.138.
Acknowledgments
This research was supported by establishment of relief service for disaster victims, National Disaster Management Research Institute (NDMI), Republic of Korea. The authors also express gratitude to Tae-jin Lee and Wonkwang Jo for providing useful feedback.
Author contribution
Yubin Lee: conceptualization, methodology, formal analysis, writing (original draft) Myoungsoon You: writing, review and editing, supervision.
Competing interest
The authors declare none.
Open access
