Oil spills are man-made disasters that can seriously affect both ecosystems and human health and life in a widespread area over a long-term period.Reference Ha, Kwon and Cheong 1 The impact of an oil spill on the residents in contaminated areas may be intense, prolonged, and complex in nature. Oil spills can lead to not only physical health, economic, and toxicological effects, but also mental health effects.Reference Osofsky, Osofsky and Hansel 2 Oil spills have been known to cause large-scale and personal economic damages, especially in the commercial fishery industry,Reference Arata, Picou, Johnson and McNally 3 which subsequently causes psychological stress in local communitiesReference Grattan, Roberts, Mahan, McLaughlin, Otwell and Morris 4 and disruptions of community bonds.Reference Galea, Tracy, Norris and Coffey 5 Research on the psychological and behavioral health effects of oil spill disasters has been recommended as a top priority.Reference Palinkas 6 Susceptible biological factors such as sex and adaptive capacity such as family support have been reported as important determinants of suicide risk.Reference Tang, Yen and Cheng 7
The Hebei Spirit oil spill occurred on December 7, 2007, in the West Sea off the coast of the Taean peninsula of South Korea. The magnitude of the oil spill was approximately 10,900 tons of crude oil, which was approximately one-third the size of the Exxon Valdez oil spill. The Hebei Spirit oil spill was the largest oil spill in South Korea, and contaminated 1052 km of the country’s western coastline.Reference Cheong, Ha and Lee 8 Subsequent to the spill, 4 residents from the affected communities committed suicide from January 2008 to February 2010. Most of the cleanup workers were local residents in the villages near the coastline who continuously participated in the cleanup effort over several months along the contaminated coastline, mostly as a means of temporary livelihood. In January 2013, a local court assessed $693.2 million for total damages including compensations for economic damage to local residents, expenses for cleanup work, and expenses for marine ecosystem restoration by local government agencies, of which $390.7 million was awarded to the local residents of Taean District. 9 However, those amounts were far below the amounts that the affected community expected and for which it petitioned.
Previous studies have been conducted on mental health in relation to oil spills, including the Sea Empress,Reference Lyons, Temple and Evans 10 Exxon Valdez,Reference Arata, Picou, Johnson and McNally 3 and Prestige Reference Sabucedo, Arce and Senra 11 oil spills and the most recent large accident, the Deepwater Horizon oil spill.Reference Osofsky, Osofsky and Hansel 12 , Reference Buttke, Vagi and Bayleyegn 13 A previously published study on the Hebei Spirit oil spill was a brief report on acute stress in a selected small group of victims.Reference Song, Hong and Cheong 14 In the present study, therefore, we aimed to investigate the prevalence of psychological symptoms in a representative sample of residents and to examine the determinants of psychological vulnerability following the Hebei Spirit oil spill for assessment of needs for public mental health intervention in the affected area.
METHODS
Study Subjects
The study subjects were residents who lived in the Taean District from July 7, 2008, to August 1, 2008. A stratified sampling method was used that was based on the distance to the oil band and contaminated coast. The communities (myeons and eups, administrative districts equivalent to towns and townships, respectively) facing the shoreline were categorized into four regions: I, closest to the spill point (less than 15 km) and directly affected by the thick oil band (Sowon and Wonbuk); II, from 15 km to 25 km from the spill site (Geunheung and Iwon); III, remote from the spill site, more than 25 km from the spill site (Anmyeon, Gonam, and Nam); and IV, inland villages that were less dependent on maritime activities (Taean-eup, downtown Taean) (Figure 1). The sampling unit was ri, the smallest administrative district equivalent to a village, and the study ri were randomly selected among the lists of ri in each classified region. Sampling rates were 4% (452 of 11,313), 3% (260 of 8672), 2% (346 of 17,341), and 1% of the population (266 of 26,584) of regions I, II, III, and IV, respectively. 15
Map Showing the Locations of the Study Areas and the Oil Band of the Hebei Spirit Oil Spill in Taean, Korea, 2007.
Each interviewer was trained on the survey, visiting homes, informing subjects and obtaining consent, conducting the interview, and recording methods for questionnaires, which included questions on sociodemographic factors and scales for psychological symptoms. Five such trained interviewers visited all houses in the sampled villages and interviewed one person per household. The house visits were supported by the Taean District administrative office and corresponding village social workers. Among 1324 interviewees, a total of 993 subjects were included for statistical analysis after excluding 127 incomplete samples and 204 subjects younger than 19 years of age.
This study was approved by the institutional review board at the Dankook University Hospital (Cheonan, Republic of Korea). Written informed consent was obtained from each subject prior to the interview.
Geographical Variables of Oil Spill Exposure
Besides the 4 classified regions described above, we used geographical variables to represent the degree of oil spill exposure, inland or coastal area, and the distance to the oil band (Figure 1). An area was considered to be either a coastal area, if the villages (ri’s) directly bordered water, or an inland area, if the villages did not border the water. The straight-line distance to the oil band from the residential location of each participant was measured by using a global positioning system on an Internet map application. 16 For 114 subjects for whom no address was available for the village (ri), this distance was replaced by the median distance of subjects who resided in the same town or township (myeon or eup).
Psychological Symptoms
Post-Traumatic Stress
Post-traumatic stress (PTS) was assessed by use of the post-traumatic diagnostic scale (PDS), which was originally developed by Foa et alReference Foa, Cashman and Perry 17 and was subsequently translated into the Korean language and validated.Reference Nam, Kwon and Kwon 18 The Korean version of the PDS is composed of 17 items of PTS-related symptoms, with each item scored from 0 to 3 points. The individual scores are then summed with a possible total score from 0 to 51 points. This scale also measures the frequency of a traumatic event. A subject with a total score above 15 points was considered to be experiencing PTS symptoms.Reference Sheeran and Zimmerman 19 The Cronbach’s α of the Korean version of the PDS was 0.95 in this study.
Depression
The Korean version of the Center for Epidemiologic Studies Depression Scale (CES-D)Reference Cho and Kim 20 was used to assess depressive symptoms experienced during the week leading up to the interview. This scale contains 20 items that measure feelings of depression, with each item having a possible score from 0 (rarely or none [less than 1 day]) to 3 (most or all [5-7 days]). The cutoff point for a clinically depressed patient is 25 and that for a high-risk group in the community is 21.Reference Radloff 21 The Cronbach’s α of the Korean version of the CES-D was 0.90 in this study.
Suicidal Ideation
Suicidal ideation was measured by using the Korean version of the Scale for Suicidal Ideation (SSI).Reference Park and Shin 22 Each of the 19 items was scored from 0 (none or rarely) to 2 points (every time). A high-risk group was defined by a score greater than 16 out of 38 possible total points. The Cronbach’s α was 0.93 in this study.
Anxiety
The Korean version of the Symptom Checklist–90-Revision (SCL-90-R) was used to assess anxious symptoms.Reference Kim, Kim and Won 23 Possible scores were from 0 (none) to 4 (very severe) points for each of the 10 items. The cutoff point was a T score of 70 on the standardized distribution of the Korean population for a high-risk group. The Cronbach’s α was 0.95 in this study.
Modifiers, Confounders, or Other Covariates
Information on other covariates was obtained during the house visit interview. Occupation was classified into 2 categories (fishery: culturist, fishery, and shell-fishing, and nonfishery: services, agriculture, government officer, part-time worker, business professional, and others). Educational attainment was also classified into 2 categories (<12 years: less than high school graduation, and ≥12 years: high school graduation or more). Participation in the cleanup work (yes or no), demographic factors such as age, sex, household income (≤999 KRW/month, 1000-2999 KRW/month, and ≥3000 KRW/month), and marital status (married, never married, or divorced/separated) were considered in the analyses.
Statistical Analysis
The finite population correction (FPC) and sampling weights using the 2008 population numbers of each region (I to IV) of Taean District 15 were calculated, which were used in the survey analysis methods. The formula was as followsReference Bove 24 :
$$\eqalign{FPC=\,&{\rm square\ root [(population\ size}{\minus}{\rm sample\ size)}/\cr &\quad\quad\quad\quad\ \ {\rm (population\ size}{\minus}{\rm 1)]}$$
The FPC was used for prevalence only because it is used to reduce the variance when a substantial fraction of the total population of interest has been sampled. The FPC was not used for evaluation of psychological symptoms because it may not be appropriate if the target of inference is the process generating the data rather than the statistics of a particular finite population.Reference Lumley 25 Sampling weights were determined by using the population number of a given region, divided by the sample size of the same region.
Symptom prevalence and 95% confidence intervals (CIs) were calculated by using the survey analysis methods based on the weighted population of the sampled area.Reference Lumley 25 To examine the association between oil spill exposure and psychological symptoms, the odds ratio (OR) and 95% CI were estimated by using the survey logistic regression model adjusted for several covariates, including age, sex, marital status, educational attainment, cleanup work, and occupation. P for trend was calculated by using the ordinary or continuous scales of the variables in the corresponding survey logistic model. To examine the modifying effects of occupation on the association between oil spill exposure and psychological symptoms, we tested the significance of the interaction. To test the statistical significance of the multiplicative interaction, the log likelihood ratio test was conducted between with and without the interaction term in the corresponding survey logistic model.Reference Stevenson, Nunes and Sanchez 26 To test the additive interaction, the relative excess risk due to interaction and 95% CI was estimated by using the epi.interaction() function in the R package epiR.Reference Stevenson, Nunes and Sanchez 26 All analysis was conducted by using R version 2.13.1 27 with the significance level set at P<0.05.
RESULTS
Prevalence of Psychological Symptoms of the Study Subjects
The overall prevalences of post-traumatic stress disorder (PTSD), depression, suicidal ideation, and anxiety symptoms were 19.5%, 22.0%, 2.3%, and 4.2%, respectively. The psychological symptoms of PTSD, depression, and suicidal ideation were the most prevalent (49.0%, 39.9%, and 5.1%, respectively) in the closest category (the lowest quartile) of the distance from residential area to the oil band and the second most prevalent in the fishery or related occupational group (45.0%, 37.7%, and 4.7%, respectively). The anxiety symptom was the highest in the fishery or related occupational group, followed by the closest category of the distance to the oil band (11.3% and 9.5%). Prevalence of psychological symptoms except for suicidal ideation was higher in subjects who were aged 40 years or older, married, less educated, of middle or lower household income, residing in regions exposed to oil (from IV to I regions in decreasing order), residing closer to the oil band (from Q1 to Q4 distance in order), and participating in cleanup work. Contrary to this, suicidal ideation prevalence was higher among males, people younger than 39 years, those who were highly educated (12 or more years of education), and those of middle or higher household income (Table 1), but the significances of difference in sex, age, education, and income disappeared in the multivariate model adjusted for occupation (data not shown).
Prevalence of Psychological Symptoms of the Study Subjects in Areas Affected by the Hebei Spirit Oil Spill, Taean, Korea, 2008Footnote a
a Abbreviations: CI, confidence interval; Prev, prevalence. Prevalence (%) and 95% CIs were estimated by using survey data analysis based on the weighted population of the sampled area.
b 1000 KRW=0.96 USD (30 June 2008); GNI (Korea, 2008)=19,231 USD.
c The fishery and related occupations included culturist, fishery and shell-fishing; non-sea-related occupations included services, agriculture, officer, part-time worker, professional, and others.
d The straight line of the distance from the subject’s residential house to the nearest oil band in the coastline. Numbers are not always the same total because of missing values.
Risk of Psychological Symptoms According to Geographical Distance to Oil Band and Occupation
When analyses were performed separately with distance and occupation in association with psychological symptoms, all types of symptoms showed a significant association with closer distance and fishery or related occupation. After mutual adjustment for distance and occupation, PTS and depression remained significantly associated with both distance and occupation, whereas suicidal ideation and anxiety remained significantly associated only with occupation. Thus, a fishery or related occupation had a stronger impact on suicidal ideation and anxiety than did living closer to an oil band (Table 2).
Risk of Psychological Symptoms in Association With Geographical Distance of Residential Site to Oil Spill Band and Occupation of Study Subjects in Area Affected by the Hebei Spirit Oil Spill, Taean, Korea, 2008Footnote a
a Abbreviations: CI, confidence interval; OR, odds ratio; ref, reference. Model 1: ORs and 95% CIs were estimated by using the survey logistic regression model adjusted for age, sex, marital status, educational attainment, and cleanup work. Model 2: additionally to model 1, distance to oil band and occupation were mutually adjusted in one model. P for trend was calculated by using the continuous scale of the exposure variable in the corresponding model.
b The straight line of the distance from the subject’s residential house to the nearest oil band in the coastline.
c The fishery and related occupations included culturist, fishery, and shell-fishing; non-fishery occupations included services, agriculture, officer, part-time worker, professional, and others.
Synergistic Effect of Geographical Distance to the Oil band and Occupation
The interactive effects of distance and occupation by type of psychological symptoms are illustrated in Figure 2. PTS was significantly associated with distance and occupation, with a significant additive synergism (P value for additive interaction=0.03). Depression showed a similar pattern but without a significant interaction effect (Table S1). A significantly increased risk of suicidal ideation was shown only in people with a fishery or related occupation. Anxiety showed a higher risk among people with a fishery or related occupation regardless of distance than among people living distant from the oil band and with a non-sea-related occupation (P value for multiplicative interaction=0.06).
ORs and 95% CIs of Psychological Symptoms for Distance to the Oil Band and Occupation of the Study Subjects in the Hebei Spirit Oil Spill, Taean, Korea, 2008. Abbreviations: CI, confidence interval; OR, odds ratio. ORs and CIs were estimated by using the survey logistic regression model adjusted for age, sex, marital status, educational attainment, and cleanup work. Geographical distance: the straight line of distance from the subject’s residential house to the nearest oil band in the coastline (km). Q3-Q4, 11.10-45.20 km; Q1, 0.17-3.84 km. The fishery and related occupations included culturist, fishery, and shell-fishing; non-fishery-related occupations included services, agriculture, officer, part-time worker, professional, and others. *P value for multiplicative interaction <0.1: calculated from the P value for interaction term in the survey logistic model. +P value for additive interaction <0.05: calculated from the P value for relative excess risk due to interaction.Reference Hosmer and Lemeshow 28
DISCUSSION
We found that symptoms of PTSD, depression, suicidal ideation, and anxiety were prevalent among residents of the coastal communities affected by the Hebei Spirit oil spill. These prevalences increased significantly along with increasing severity of oil exposure and were higher in people with a fishery or related occupation than in those with other livelihoods. People with a fishery or related occupation and people residing in the vicinity of the oil band in the contaminated coastline showed additively increased symptom risks of PTSD. Risk of suicidal ideation was predominantly increased in people with a fishery or related occupation. Risk of anxiety was higher among people with a fishery or related occupation regardless of distance as well as among those residing within a closer distance with a non-sea-related occupation.
We found increased psychological symptoms according to exposure level or proximity to the oil spill, which was described as a direct effect due to the disaster itself, as well as a significant association with occupation, which was described as an indirect effect or economic damage/income loss as a result of the disaster.Reference Palinkas 6 These findings are consistent with previous reports on the associations between oil spill accidents and the mental health of people affected by the Exxon Valdez,Reference Palinkas, Petterson, Russell and Downs 29 Sea Empress,Reference Lyons, Temple and Evans 10 and Prestige Reference Carrasco, Pérez-Gómez, García-Mendizábal, Lope, Aragonés, Forjaz, Guallar_Castillón, López_Abente, Rodríguez-Artalejo and Pollán 30 accidents.
Although it may not be appropriate to directly compare the results of the present study with those reported in previous studies because of the different measurement tools used, the PTSD symptom prevalence in the present study is compatible with that for survivors 6 months after the Oklahoma city bombing (36%),Reference North, Nixon and Shariat 31 for the US population outside of New York City 2 months after the 9/11 terrorist attacks on the Twin Towers in Manhattan (17%),Reference Silver, Holman and McIntosh 32 and for residents 1 year after experiencing the Exxon Valdez oil spill (17%).Reference Palinkas, Petterson, Russell and Downs 29
In terms of depression, the prevalence in the present study is consistent with the 23% prevalence in survivors of the Oklahoma city bombing,Reference North, Nixon and Shariat 31 the 20.2% prevalence in residents of the communities exposed to the Exxon Valdez oil spill,Reference Palinkas, Petterson, Russell and Downs 29 and the 35% prevalence in residents directly affected by the Gulf of Mexico oil spill,Reference Osofsky, Osofsky and Hansel 2 although the measurement tools were all different. However, the prevalence had decreased from 62.7% at 8 weeks after the Hebei Spirit oil spill by use of the same scale, the CES-D.Reference Song, Hong and Cheong 14
The prevalence of suicidal ideation was 4.0% in subjects from region I towns (Sowon and Wonbuk) in this study. This prevalence was much decreased from the 18.3% at 8 weeks after the Hebei Spirit oil spill accident.Reference Song, Hong and Cheong 14
The anxiety symptom prevalence in this study was lower than the 34.0% prevalence in residents 1 year after the Exxon Valdez oil spill,Reference Palinkas, Petterson, Russell and Downs 29 the 11.0% prevalence in residents at 14 months after the Prestige oil spill accident,Reference Carrasco, Pérez-Gómez, García-Mendizábal, Lope, Aragonés, Forjaz, Guallar_Castillón, López_Abente, Rodríguez-Artalejo and Pollán 30 and the 48% prevalence in residents directly affected by the Gulf of Mexico oil spill.Reference Osofsky, Osofsky and Hansel 2 Increased symptom risk of anxiety in people with a fishery or related occupation regardless of distance may show an indirect exposure as a major determinant. However, the increased risk among people with a non-sea-related occupation but who resided within the vicinity of the oil band may reflect general anxiety due to being a visual witness to the disaster. The findings of a synergistically increased risk of PTS among people with a fishery or related occupation and living closer to the oil band and the predominantly strong impact of occupation on suicidal ideation imply a target group on whom future planned public intervention should be focused. Furthermore, community mental health programs as well as programs of socioeconomic compensation should consider all people who have worked in fisheries or in a related occupation beyond the physical border of oil contamination.
Limitations
The present study had some limitations. First, we did not perform a comparison of the study findings with findings for an unexposed area. Instead, we compared groups at different levels of oil exposure, ie, internal comparisons, which may have reduced the possibility of reporting bias due to exaggerated responses on subjective symptoms among people in the exposed area compared with unexposed areas. Although the extent to which a psychological measurement is biased is unknown,Reference Walter 33 comparisons within exposed areas might be less biased than comparisons between exposed and unexposed areas in terms of reporting. On the other hand, mental health effects due to an oil spill disaster have been shown to reach far beyond the oil band in comparison with physical health effects.Reference Grattan, Roberts, Mahan, McLaughlin, Otwell and Morris 4 , Reference Fan, Prescott and Zhao 34 Therefore, a survey confined to communities severely affected by the oil spill as in the present study may not grasp the whole magnitude of the mental health effects. Second, differences might exist between responders and nonresponders to the questionnaire interview. However, the nonresponse rate of this survey was 9.6%, and the potential selection bias due to nonresponse should be negligible. Third, psychological symptoms in this study were not clinically diagnosed but were assessed by using existing questionnaire scales. Therefore, a potential misclassification bias in the psychological outcomes may have existed, but it should not be differential regarding the exposure status. Last, the cross-sectional design limits establishing a causal relationship between the oil spill and mental distress in the affected community members.
CONCLUSIONS
Although both direct and indirect exposure to an oil spill increased the risk of PTS, depression, suicidal ideation, and anxiety symptoms, indirect exposure, ie, economic impacts, was a stronger determinant for suicidal ideation. Longer follow-up study and social supports including a compensation program for income loss and community mental health programs are needed for residents in the area affected by the Hebei Spirit oil spill.
Acknowledgments
This study was financially supported by the Ministry of Environment, Republic of Korea.
Conflict of interest
The authors declared no conflicts of interest.
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/dmp.2015.68.
