On April 16, 2014, the Sewol ferry with 476 passengers sank at sea off South Korea (hereafter Korea)’s southern coast of the Yellow Sea. The number of dead and missing from this accident was 304, of whom 250 were high school students on a field trip. This disaster was a tremendous shock to the entire country, leaving indelible emotional scars especially in the community of Ansan, where most victims and survivors as well as their families resided.

Many studies have reported that such national disasters can trigger adverse outcomes not only among victims and their families, relatives, and friends, but also among a broad spectrum of the population exposed to the disaster situation [1-3], including voluntary support workers and community residents who indirectly experienced the disaster through the media [4,5]. In particular, human disasters tend to occur in places with high concentrations of human and material resources, thus involving a high death toll and affecting the entire society [6]. For the purpose of this study, it was hypothesized that accidents involving specific local residents and students would have a greater effect on the community concerned.

The effect of accidents and disasters on health status manifests in various forms, not only psychiatric symptoms such as posttraumatic stress disorder and depression, but also somatic symptoms, substance abuse, interpersonal relationship difficulties, and loss of social network [7]. The mental health response to disaster is an important research area, as shown by a large number of studies on the 9/11 terrorist attack. Higher prevalence of posttraumatic stress was found in New York compared with Washington and other metropolitan areas [4], and the closer to the terrorist attack site the higher the prevalence [1]. In Korea, in the wake of the Hebei Spirit oil spill in 2007, a high prevalence of depression, stress, and suicidal ideation was observed among the residents of the victimized area [8]. A one year follow-up study of community health status in the affected areas reported high burdens of disease through mental disorders [9].

Population groups exposed to a disaster undergo a generally observed process of showing strong emotional and psychological reactions for up to one year, and begin to recover after the anniversary reaction [10]. In this process, adequate interventions can shorten the recovery period or mitigate psychological harm, whereas health problems can persist without such interventions. Although the Trauma Center built in Ansan after the disaster has provided counseling and monitoring for the bereaved families and related individuals [11], the level of support and research provided for community residents is not up to the magnitude of shock suffered by them. The effects of emotional and psychological harm suffered by the community residents as a whole should be analyzed to establish adequate disaster mental health response plan.

The purpose of this study was to determine the effect of a disaster on a local community by evaluating the mental health status of community residents in terms of depression, stress, somatic symptoms, anxiety, and suicidal ideation. Evaluation took place six months after the disaster using a survey-based approach comparing different geographical areas.

The mean age of the subjects (n=7,076) was 49.7 years, with males accounting for 44.9% of the sample. The rural areas of Jindo and Haenam had a higher mean age than the urban areas in Ansan and control cities, as well as higher percentages of female residents and lower education and income levels (Table 1).

Prevalence in Ansan for depression, moderate-to-severe stress, somatic symptoms, anxiety, and suicidal ideation were 11.4%, 8.8%, 11.9%, 5.9%, and 17.7% (crude rate), respectively. The highest levels among the three areas and the regional differences proved statistically significant (p<0.05) (Table 2). After age and sex standardization as well, Ansan’s prevalence rates were highest. Jindo and Haenam showed the lowest post-standardization prevalence rates.

The results for different sex and age groups confirmed the overall tendency of higher prevalence rates in Ansan than the other two areas for all age groups (Figure 2). Females showed a higher prevalence of all psychiatric symptoms irrespective of area. In general, upward trends in depression, anxiety, and suicidal ideation were observed with increasing age. The prevalence of stress in the young and middle-aged groups in urban areas (Ansan and Gyeonggi) were generally high compared with Jindo and Haenam.

A multiple logistic regression analysis was performed to adjust for age, sex, education, and income, followed by inter-area ORs for psychiatric symptoms (Table 3). Compared with the control area (Gyeonggi), Ansan showed significantly higher adjusted ORs for depression (1.66; 95% CI, 1.36 to 2.04), stress (1.37; 95% CI, 1.10 to 1.71), somatic symptoms (1.31; 95% CI, 1.08 to 1.58), anxiety (1.82; 95% CI, 1.39 to 2.39), and suicidal ideation (1.33; 95% CI, 1.13 to 1.56) compared with the other areas. In contrast, Jindo and Haenam showed low ORs in all domains.

In this study, the short-term to mid-term effects of the Sewol ferry disaster on the affected community were evaluated by conducting a survey on the mental health status of residents 4 to 6 months after the disaster. Compared to the control area (Gyeonggi), the Ansan area most heavily affected by the disaster exhibited higher prevalence rates of the psychiatric symptoms assessed, and similarly high ORs were yielded by the multiple logistic regression analysis adjusting for sociodemographic features. Furthermore, Jindo and Haenam near the accident site showed the lowest prevalence of psychiatric symptoms, contrary to expectation, presumably due to its character as a rural area.

The relevance of the Sewol ferry disaster as it relates to the significantly higher prevalence of psychiatric symptoms in the Ansan area may be estimated by investigating existing data. The 2009 KCHS measured depression rates using the CES-D, and suicidal ideation is assessed every other year. In 2009, the age-standardized depression rates in Danwon and Sangnok of the Ansan area were 4.3% and 4.8%, respectively. The rates for Gunpo, Paldal (Suwon), Guri, and Namyangju in Gyeonggi were 6.2%, 8.3%, 7.1%, and 5.9%, respectively; and for Jindo and Haenam rates were 1.7% and 3.5%, respectively [20]. The 2014 depression data gathered in this survey demonstrated considerably higher depression rates in the Ansan area (11.8%) as well as in Jindo and Haenam (8.1%). With a time gap of five years, this increase cannot be exclusively attributed to the effects of the Sewol ferry disaster, but it may be assumed to have certain effects, given that no significant changes took place in the control area (6.9% vs. 5.9 to 8.3%).

With regard to suicidal ideation, after age standardization the affirmative answer accounted for 9.6% (Danwon) and 17.2% (Sangnok) in Ansan in 2013, 7.2% (Gunpo), 10.6% (Paldal/Suwon), 8.1% (Guri), and 13.5%, (Namyangju) in Gyeonggi, and 3.3% and 4.9% in Jindo and Haenam, respectively [20]. This indicates that in one year’s time suicidal ideation increased in all surveyed communities except for Namyangju, with Danwon in Ansan demonstrating the highest increase of 7.6%. However, the higher baseline value in the Ansan area compared with other areas should be taken into account. This may be interpreted as a higher vulnerability of already affected subjects to the effects of the disaster [21].

As indicators allowing for time-series comparison, experience of depressive symptoms and perceived stress questions, one item each, are annually assessed in the KCHS. In Ansan, the age-standardized rate of depressive symptoms increased from 9.8% in 2013 to 13.0% in 2014 in Danwon, and decreased from 12.7% to 8.1% in Sangnok, although higher than the average of 7.0% in Gyeonggi [20]. Danwon showed the highest depression rate of the 254 communities across the country. This is consistent with our survey results using the CES-D and indicative of the effects of the Sewol ferry disaster. However, it should be taken into account that the depression rate in Ansan was also high in 2013 and over the 75% quantile. Perceived stress slightly decreased from 33.3% to 31.7% in Danwon and from 35.6% to 26.2% in Sangnok, without showing any noticeable differences from the Gyeonggi average of 30.2% (95% CI, 29.7% to 30.7%) [20]. These findings deviate slightly from what was found in this study for stress.

Somatic symptoms and anxiety were also measured in this study. The ORs in the Ansan area were higher than in Gyeonggi. This result is consistent with the literature that indicates unexpected disaster gives rise to anxiety through uncertainty about danger [22], and triggers various nonspecific symptoms inexplicable with physical disorders [23]. Even though anxiety is a normal reaction to disaster in the short term, a prolonged period of anxiety can develop into a chronic state of anxiety and trigger psychosomatic problems [24].

The age-dependent prevalence of depression shown in this study is similar to the study of Oh et al. [25] conducted in 2009, in which the CES-D results from the community health survey were analyzed and mapped by age. Subjects in their 30s showed the lowest depression prevalence, which increased with age, whereas those aged 19 to 29 showed a slightly higher depression rate than people in their 30s. The effect of the disaster on children and adolescents could not be assessed because this study only targeted adults. This aspect should be investigated in future research, given that early detection of and intervention for psychological injury in children and adolescents still in the formative years of their lives are of vital importance [26].

Most of the mental health research on large-scale disasters has been conducted focusing on survivors or bereaved families, and only a small number of studies have examined the effects of disaster on the general population at the community level. There are a few community-based studies in relation to the ocean contamination by the 2007 Hebei Spirit oil spill, but most participants were individuals directly involved through cleanup works or residents who suffered financial losses [8,9,27]. In other countries, too, most studies targeted eyewitnesses or residents within the areas affected by natural disasters, such as typhoons and earthquakes [5,28]. In contrast, the Sewol ferry disaster did not take place in a physical space of residence of the victims, and residents experienced the accident only indirectly via media or acquaintances. It is also distinct from other disasters in that the main victims were adolescents from the communities concerned, thus bringing a tremendous shock and severe emotional and psychological trauma to community members of all ages. Taking into account such distinct features and the implications of this study, comprehensive mental health monitoring and interventions are necessary for the community members.

Psychological and emotional tension caused by a disaster escalates the collective stress of the whole community. Fears of other disasters and worries about impending crises are ignited. Such a changed atmosphere and loss of human and material resources caused by the disaster can create long-term stress [29]. Although psychological damage may mitigate over time, if left uncared for or complicated by other negative factors long-term problems can develop. This study was conducted 4 to 6 months after the disaster, and a Taiwanese study conducted at the same point in time similarly reported symptoms of posttraumatic stress disorder and severe depression after a major disaster [28]. Such symptoms have also been reported as much as seven years after the disaster [5,29]. A follow-up survey in relation to the 9/11 terrorist attack noted that persisting long-term psychological symptoms were more associated with absence of timely interventions than with direct exposure to the disaster [2]. Therefore, interventions should be administered timely.

One of the limitations of this study is the selection of control group. Four regions in Gyeonggi, the province where Ansan is located, were selected as control communities to minimize regional deviations, but even after adjusting for sociodemographic features there were inter-region deviations. Although Ansan showed a higher prevalence rate in 2014, the health index related to mental health status showed an elevated level in a prior survey. Jindo and Haenam had low prevalence rates in the present and prior surveys but their geographical features are distinct from Gyeonggi, which did not allow for direct comparison with the control communities. Moreover, the Sewol ferry disaster was a national tragedy that was reported in real time via media, leaving strong emotional imprints in all Koreans. As such, there may have been an underestimation of the effects of the disaster on the accident region.

Another limitation was the lack of data and the impossibility of time-series observation. However, meaningful observations could be made using different forms of variables and past data. As for depression and suicidal ideation, different reactions of respondents between a face-to-face interview and self-administered questionnaire should also be taken into account, because sensitive questions may have been answered more honestly in the latter such that underestimation of past prevalence rates cannot be ruled out.

As a third limitation, it should be pointed out that individuals directly involved in the disaster, such as bereaved families and relatives, could not be targeted for sampling due to the nature of the community-based sampling system of this study. Furthermore, items evaluating the direct relevance of the Sewol ferry disaster were excluded from the questionnaire to forestall prejudices related to the disaster. This did not allow for the application of detailed risk classification models in accordance with the degree of exposure. Nor could we apply the survey instrument for posttraumatic stress disorder because most community samples did not directly experience the disaster. However, some studies showed that residents who had vicariously experienced a national disaster developed posttraumatic stress disorder [3]. It is therefore considered necessary to develop a screening tool for evaluating disaster-related stress in the general population.

Despite these limitations, various aspects of mental health were evaluated using screening tools with recognized efficiencies. A salient feature of this study is its representativeness, in which systematically extracted samples were used from the existing community health survey system. It is also significant that this study could evaluate the mental health status of communities within six months after the disaster. Given the results of this cross-sectional survey, it is considered necessary to provide adequate interventions to actively help residents of victimized communities recover their mental health status. A follow-up study to observe the change and recovery process in the affected communities will be significant from both academic and social perspectives.