For Peer Review Only

BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

Burden of disease attributable to the Hebei Spirit oil spill in Taean, Korea

ForJournal: peerBMJ Open review only Manuscript ID: bmjopen-2013-003334

Article Type: Research

Date Submitted by the Author: 06-Jun-2013

Complete List of Authors: Kim, Young-Min; Rollins School of Public Health, Emory University, Department of Environmental Health Park, Jae-Hyun; Sungkyunkwan University School of Medicine, Department of Social and Preventive Medicine Choi, Kyusik; Sungkyunkwan University School of Medicine, Department of Social and Preventive Medicine Noh, Su Ryeon; Taean Environmental Health Center, Choi, Young-Hyun; Taean Environmental Health Center, Cheong, Hae-Kwan; Sungkyunkwan University School of Medicine, Department of Social and Preventive Medicine

Primary Subject Public health Heading:

Secondary Subject Heading: Epidemiology, Health economics, Occupational and environmental medicine

EPIDEMIOLOGY, HEALTH ECONOMICS, Epidemiology < THORACIC http://bmjopen.bmj.com/ Keywords: MEDICINE, MENTAL HEALTH, TOXICOLOGY

on October 1, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 Burden of disease attributable to the Hebei Spirit oil spill in 5 6 7 Taean, Korea 8 9 10 Young-Min Kim1 , Jae-Hyun Park2, Kyusik Choi2, Su Ryeon Noh3, Young-Hyun Choi3, Hae- 11 12 Kwan Cheong2 13 14 1Department of Environmental Health, Rollins School of Public Health, Emory University, 15 For peer review only 16 17 1518 Clifton Road NE, Atlanta, GA, USA 18 19 2Department of Social and Preventive Medicine, Sungkyunkwan University School of 20 21 Medicine, 2066 Seobu-ro, Jangan-gu, Suwon, 440-746, Republic of Korea 22 23 3 24 Taean Environmental Health Center, 1952-16 Seohae-ro, Taean, Chungnam, 357-902, Re 25 26 public of Korea 27 28 29 30 * Corresponding author: Hae-Kwan Cheong 31 32 2066 Seobu-ro, Jangan-gu, Suwon, Gyeonggi-do, 440-746, Korea 33 34

35 Tel: +82-31-299-6300, Fax: +82-031-299-6299 http://bmjopen.bmj.com/ 36 37 E-mail: [email protected] 38 39 40 41 * Keywords: oil spill, disability-adjusted life-year (DALY), polycyclic aromatic 42

43 on October 1, 2021 by guest. Protected copyright. 44 hydrocarbons (PAHs), post-traumatic stress disorder (PTSD), disability weight 45 46 47 48 Word count: 49 Abstract: 257 words 50 51 Main text: 2915 words 52 No. of tables: 3 53 No. of Figures: 2 54 No. of references: 27 55 56 57 1 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 25 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 ARTICLE SUMMARY 5 6 7 Article focus 8 ▶ This study aimed at assessing the burden of disease (BOD) of the residents of 9 10 the contaminated coastal area 11 ▶ and at analyzing the BOD attributable to the oil spill by disease, group, and 12 region. 13 14 15 Key messagesFor peer review only 16 17 ▶ This study provided an opportunity to assess the disease burden of the 18 residents integrating mental and physical symptoms. 19 20 ▶ Among diseases, years lost due to disabilities (YLDs) for mental disease such 21 as PTSD and depression attributable to the oil spill are higher in males than 22 23 females and those for asthma and allergies are higher in females than males. 24 ▶ Oil spills near coastlines can cause considerable adverse health effects and that 25 26 the summation of disease burden is not negligible, even though the size of the 27 population affected may be small. 28 29 Strength and limitations 30 31 ▶ This is the first study to quantify the BOD of a single environmental disaster. 32 33 ▶ This study demonstrates that BOD is an objective and comprehensive metric for 34 estimating and comparing the health effects of environmental hazards and

35 disasters across different regions and time periods. http://bmjopen.bmj.com/ 36 Long-term health outcomes such as cancers and chronic degenerative diseases 37 ▶ 38 could not be estimated. 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 2 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8

9 10 11 ABSTRACT 12 13 Objectives: We aimed at assessing the burden of disease (BOD) of the residents of the 14 15 contaminatedFor coastal area peer and at analyzing review the BOD attributable only to the oil spill by disease, 16 17 group, and region. 18 19 20 Design: Health impact assessment study of a specific environmental exposure from oil spill 21 22 Setting: Use of secondary health outcome data. A whole population of a community affected 23 24 by an anthropogenic environmental disaster. 25 26 Participants: 63,053 individuals (male: 31,636; female: 31,417) and 3,420 individuals (male: 27 28 1,805; female: 1,615) from two counties, respectively 29 30 31 Interventions: None. Observational study on the effect of a specific environmental health 32 33 hazard. 34

35 Primary and secondary outcome measures: BOD was measured using disability-adjusted http://bmjopen.bmj.com/ 36 37 life-year (DALY) method based on the incidence and prevalence of the health outcome 38 39 40 related to the oil spill. 41 42 Results: Years lost due to disabilities (YLD) attributable to the oil spill were estimated to be

43 on October 1, 2021 by guest. Protected copyright. 44 14,724 DALYs (male: 7,425 DALYs; female: 7,299 DALYs) for the year 2008. The YLD for 45 46 males for mental diseases were higher than for females, and the YLD for females was higher 47 48 in asthma and allergies than for males. The effects of asthma and allergies were the greatest 49 50 51 for people in their 40s, with the burden of mental illness being the greatest for those in their 52 53 20s. Proximity to the spill site was associated with increased burden of disease. 54 55 Conclusions: An oil spill near a coastline can cause substantial adverse health effects. As the 56 57 3 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 25 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 health effects of hazardous pollutants from oil spills are long-lasting, close follow-up studies 5 6 are required to identify chronic health effects. 7 8 Trial registration: Not applicable 9 10 11 12 INTRODUCTION 13 14 On December 7, 2007, the Hebei Spirit oil tanker spilled 12,547 kl (10,900 tons) of crude 15 For peer review only 16 oil into the Yellow Sea, approximately 8 km away from the coast of Taean, Korea. This was 17 18 19 the largest oil spill in the history of Korea. Spilled oil contaminated over 1,000 km of the 20 21 western coast of Korea including the Taean Coast National Park.[1] Crude oil contains 22 23 various volatile organic compounds (VOCs) such as benzene, toluene, ethylbenzene, xylenes, 24 25 and polycyclic aromatic hydrocarbons (PAHs),[2] all of which are potential air-borne toxins. 26 27 Certain VOCs, benzene in particular, are carcinogenic to humans and are associated with 28 29 30 haematologic cancer.[3] Previous studies [4-11] have reported that oil spills have caused 31 32 adverse health effects for residents of contaminated areas and clean-up workers. Risk 33 34 analyses revealed that more frequent clean-ups and greater exposure to oil are strongly

35 http://bmjopen.bmj.com/ 36 associated with higher frequencies of psychological symptoms. [8-11] Rodriguez-Trigo [6] 37 38 39 reported that participation in a major oil spill clean-up is associated with chromosomal 40 41 damage, although they were unable to determine whether oil exposure itself caused the 42

43 abnormalities. on October 1, 2021 by guest. Protected copyright. 44 45 After the Hebei Spirit oil spill, more than 2,000,000 people (person-days), including over 46 47 1,200,000 person-days by volunteers, participated in clean-up activities during the first seven 48 49 50 months.[1] Among them were residents of Taean, a highly exposed group that experienced 51 52 adverse health effects, including both physical and psychological symptoms.[4, 12] Although 53 54 many studies have reported positive relationships between oil spills and acute and chronic 55 56 57 4 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 health impacts, there have been no studies quantifying the burden of disease (BOD) due to oil 5 6 spills, which are necessary to assess the scale of health damage at the population level as well 7 8 as the associated compensation costs. 9 10 11 Therefore, we sought to quantify the adverse health impacts of exposure to oil spills by 12 13 applying the disability-adjusted life-year (DALY) measure developed by the World Health 14 15 OrganizationFor (WHO).[13] peer This approach review is practical because only the estimated adverse health 16 17 burden attributable to an oil spill can be compared with other BOD, including environmental 18 19 20 burdens of disease (EBD). Thus, the analyses of disease burdens by disease and region can 21 22 inform policy prioritization to mitigate adverse health effects for specific exposed 23 24 populations. We aimed at assessing the BOD of the residents of the contaminated Taean 25 26 coastal area and at analyzing the BOD attributable to the oil spill by disease, group, and 27 28 region. 29 30 31 32 33 METHODS 34

35 http://bmjopen.bmj.com/ 36 Study area and design 37 38 In September 2008, the Taean Environmental Health Center (TEHC) was established 39 40 under the support of the Ministry of Environment to monitor the effects of the Hebei Spirit oil 41 42 spill on the health of the residents of the Taean coastal area. The TEHC first conducted health

43 on October 1, 2021 by guest. Protected copyright. 44 45 surveys on the residents, including the referent population, one and a half years after the spill, 46 47 establishing a cohort. Follow-up surveys were conducted every two years on the 1,900 48 49 individuals highly exposed to oil and its remnants, including residents in the contaminated 50 51 coastal area, public services persons who worked on the clean-up, and susceptible 52 53 54 populations, including school children and pregnant women. The survey included a 55 56 57 5 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 25 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 comprehensive questionnaire gathering personal history on exposure and medical problems, 5 6 and a structured questionnaire assessing psychological health and asthma, and physical and 7 8 laboratory examinations of respiratory, cardiovascular, neurologic, and psychological systems. 9 10 11 The TEHC published its first report in 2010 after analysing the prolonged health effects 12 13 for 1.5 years, and demonstrated significant relationships between exposure concentrations 14 15 and adverse Forhealth outcomes.[14] peer To estimate review the BOD caused only by the Hebei Spirit oil spill, 16 17 we obtained the survey data from the TEHC, which included disease, residence, and 18 19 20 demographic information. Residents in Taean County and some islands in neighboring 21 22 Boryeong County were included in the survey. The Boryeong islands were expected to be 23 24 affected by the oil spill because the islands are located in the affected area based on the wind 25 26 direction and tidal current at the time of the spill.[15] Kim et al.[15] reported that the VOCs 27 28 moved in a southeasterly direction, following the coastline from the accident point. 29 30 31 We classified the residents of Taean County and the Boryeong islands into five regions 32 33 based on both distance from the contaminated coastline and contamination level. Area 1 is 34

35 less than 1.1 km away from the coastline adjacent to the accident location where the coastline http://bmjopen.bmj.com/ 36 37 was contaminated at a high level. Area 2 is 1.1-4.2 km away with no coastline, but was 38 39 40 directly affected by atmospheric VOCs and PAHs due to the wind direction. Area 3 is 4.2-23 41 42 km away with a moderately contaminated coastline, and Area 4 is more than 23 km from the

43 on October 1, 2021 by guest. Protected copyright. 44 contaminated coastline. Area 4 was assumed to be the reference area, not only because the 45 46 area might not be impacted by the oil spill directly as it is far from the contaminated coastline, 47 48 but also because residents in Area 4 rarely participate in the clean-up efforts. We classified 49 50 51 the Boryeong Islands as a separate area. Figure 1 shows the locations of the study areas and 52 53 five classified regions. 54 55 Figure 1 56 57 6 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 We assessed health outcome data regarding asthma, allergy (rhinitis, dermatitis, 7 8 conjunctivitis), and mental disorders (post-traumatic stress disorder (PTSD) and depression), 9 10 11 conditions that were well-documented by the TEHC. We focused on subchronic (duration of a 12 13 few months) and chronic (over one year duration) adverse health effects of the Hebei Spirit 14 15 oil spill, assessingFor prolonged peer health effects review over a period of 1.5 onlyyears. 16 17 18 19 20 Estimation of burden of disease attributable to oil spill 21 22 To estimate the BOD attributable to the oil spill, we calculated years lost due to disability 23 24 (YLD) [13] using disease incidence and prevalence, onset age and duration of disability, and 25 26 disability weight (DW). Because there was no prevalence data recorded prior to the accident, 27 28 we estimated excess incidences using prevalence data from the TEHC under the following 29 30 31 assumptions. We first assumed that residents living in Area 4 had no negative health 32 33 consequences due to the oil spill and used this as the baseline prevalence. We then estimated 34

35 excess incidences using prevalence differences between Area 4 and the other areas stratified http://bmjopen.bmj.com/ 36 37 by area, age, disease, and sex. 38 39 40 The duration of disability and the average onset age were calculated using the DISMOD 41 42 II model [16] for PTSD, rhinitis, dermatitis, and conjunctivitis based on prevalence data from

43 on October 1, 2021 by guest. Protected copyright. 44 the TEHC and the mortality rate of South Korea [17] by age, sex, and disease. For asthma, 45 46 the results of the Victoria, Australia study [18] were used. 47 48 DWs for each disease were measured following a method previously used by the Global 49 50 51 Burden of Disease group [13] and the Dutch Disability Weights Group.[19] A panel that 52 53 included 12 health and medicine specialists was convened. Eleven indicator conditions were 54 55 measured using the person trade-off (PTO) method. A facilitator first explained the method 56 57 7 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 25 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 and then led the discussion to determine the PTO value of each participant. The discussion 5 6 was repeated twice. Then, the visual analogue scale (VAS) method was applied to obtain 7 8 DWs for the target conditions of this study. Inter-method reliability was confirmed between 9 10 11 PTO and VAS using 11 indicator conditions, which showed fairly high consistency 12 13 (Spearman’s correlation coefficient 0.966, P-value<0.001). Reliability was confirmed by test- 14 15 retest after threeFor weeks bypeer seven specialists review and obtained fairly only high consistency (Spearman’s 16 17 correlation coefficient 0.973, P-value<0.001). 18 19 20 Finally, we estimated YLD caused by the Hebei Spirit oil spill utilizing the estimated 21 22 excess incidence attributable to the oil spill, DW, and the onset age and duration of disability 23 24 by disease, age, sex, and area. 25 26 27 28 29 RESULTS 30 31

32 33 34 Summary of the characteristics of the study area

35 http://bmjopen.bmj.com/ 36 The populations of Taean County and the Boryeong islands were 63,053 individuals 37 38 (male: 31,636; female: 31,417) and 3,420 individuals (male: 1,805; female: 1,615) in 2008, 39 40 respectively. Broken down by area, 8,348 (13%), 11,272 (17%), 4,597 (7%), 38,836 (58%), 41 42 and 3,420 (5%) residents lived in Areas 1, 2, 3, 4, and the Boryeong islands, respectively.

43 on October 1, 2021 by guest. Protected copyright. 44 45 Area 4, the reference area, was the most populous because it includes a city center. The 46 47 largest age group consisted of subjects in their 50s, with individuals in their 40s, 50s, and 60s 48 49 comprising 45% of the total population (data presented in the supplement Table S1). 50 51 A total of 10,171 residents living in the study area participated in the TEHC survey. 52 53 54 Among them, 4,354 (42.8%) and 5,817 (57.2%) residents were male and female, respectively. 55 56 57 8 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 Individuals under 15 years of age, 15-65, and over 65 years of age comprised 7.2%, 48.0%, 5 6 and 44.8% of the population, respectively. Residents of Area 1 participated in clean-up work 7 8 for an average of 122 days, and residents of Areas 2, 3, and the Boryeong islands participated 9 10 11 for averages of 50, 36, and 100 days, respectively.[14] Table 1 shows the prevalence of 12 13 disease related to the current study by disease, sex, and area derived from a survey of 9,246 14 15 adult (over 19For years old) peer residents. For review young people under only 19 years old, we used results 16 17 from 925 school students (male: 505; female: 420) living in the study area. 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 9 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from Page 10 of 25 42.6 10.1 40.2 17.0 15.0 24.5 10.8 21.3 13.5 25.2 23.8 27.9 Male Female Boryeong Boryeong Islands

Area 4 2.4 9.2 4.4 10.0 4.6 5.3 3.2 11.6 8.7 7.6 20.2 15.9 Male Female

3.1 10 10 Area 3

BMJ Open 9.4 4.7 2.9 8.8 14.0 9.6 10.4 23.9 10.6 13.2 9.0 Male Female

http://bmjopen.bmj.com/

7.1 Area 2

5.1 6.1 7.0 8.8 14.8 25.2 20.2 12.6 12.6 15.4 11.7 Male Female on October 1, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

a) For peer review only Area 1 22.0 20.4 32.9 34.6 Male Male Female . Prevalence of diseases related to this study (in study %). this related to . of diseases Prevalence b) Rhinitis 41.4 34.6 Dermatitis Conjunctivitis 19.6 32.0 25.0 38.7 Depression 23.3 29.9 PTSD Table 1 Table a) Area 1: less than 1.1 km away; Area 2: 1.1-4.2 km away; Area 3: 4.2-23 km away; Area 4: more than 23 km from the contaminated contaminated the km from 23 than 4: more Area away; km 3: 4.2-23 Area away; km 2: 1.1-4.2 Area away; km 1.1 1: less than Area a) Allergy Allergy Asthma Disease coastline; b) PTSD: posttraumatic stress stress disorder. coastline; b) PTSD: posttraumatic Mental Mental disease 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 11 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 Estimated burden of disease due to oil spill 5 6 Excess incidence attributable to the oil spill 7 8 The excess incidence of disease caused by the oil spill varied by area, sex, and disease 9 10 11 (Table 2). There were additional incidences of asthma (2,088: 989 male; 1,099 female), 12 13 rhinitis (3,625: 1,855 male; 1,770 female), dermatitis (1,976: 962 male; 1,014 female), 14 15 conjunctivitisFor (2,992: 1,365 peer male; 1,627 review female), PTSD (2,681:1,300 only male; 1,381 female), 16 17 and depression (2,326: 1,370 male; 956 female) in Areas 1,2,3 and the Boryeong islands. 18 19 20 Area 1 had the highest additional incidences for all diseases, while Area 3, a relatively 21 22 less affected area than Areas 1 and 2, showed the lowest additional incidences for all diseases. 23 24 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 11 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from Page 12 of 25

Total Total

989 1,099 962 1,014 Male Female 1,370 956 1,300 1,381 1,855 1,770 1,365 1,627

76 76 106 113 113 190 116 116 138 201 201 149 353 353 354 222 222 322 Male Female Boryeong Boryeong Islands

0 Area 3 Area 3 12 12 c

0 65 65 57 53 53 97 94 94 32 209 207 101 183 Male Male Female BMJ Open

http://bmjopen.bmj.com/

Area 2

266 169 133 222 453 481 241 231 294 214 146 158 Male Female

on October 1, 2021 by guest. Protected copyright.

a For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Area 1

839 839 729 473 473 449 838 838 581

Male For Female peer review only 1,092 1,054 Excess incidence caused by the oil spill (Unit: number of persons) persons) of number spill (Unit: by the oil caused incidence Excess Rhinitis Conjunctivitis 857 1,040 Dermatitis 606 687 b

Table 2. 2. Table a: Area 1, less than 1.1 km away; Area 2, 1.1-4.2 km away; Area 3, 4.2-23 km away; Area 4, more than 23 km from the contaminated coastline; coastline; contaminated the km from 23 than more 4, away; Area km 4.2-23 3, away; Area km 1.1-4.2 2, Area away; than km 1.1 less 1, a: Area area. and reference 3 Area between c: difference stress disorder; No b: post-traumatic Allergy Allergy Asthma Disease Depression PTSD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 13 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 Disability weight 5 6 The VAS results showed that the DW for stomach cancer (stage 4) was the highest at 7 8 0.909, and that the DW for upper respiratory infection was the lowest at 0.065, out of a total 9 10 11 of 118 conditions surveyed. Among the 118 conditions, DWs for diseases examined in this 12 13 study are shown in Table 3. We used the DW of mild intermittent asthma, 0.222, to represent 14 15 asthma, becauseFor the prevalence peer of asthma review in this study area only was estimated using survey 16 17 results following the modified International Study of Asthma and Allergies in Childhood 18 19 20 (ISSAC), and most cases of asthma were not severe.[14] 21 22 23 24 Table 3. Disability weights used in the current study 25 26 Disability 27 Disease Stage

28 weight Asthma 29 Mild intermittent 0.222 30 Mild persistent 0.291 31 Moderate persistent 0.510 32 Severe persistent 0.715 33 34 Very severe persistent 0.836

35 Allergy Rhinitis 0.205 http://bmjopen.bmj.com/ 36 Dermatitis 0.132 37 38 Conjunctivitis 0.212

39 a 40 PTSD 0.533 41 Depression 0.219 42 a: post-traumatic stress disorder 43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 Burden of disease due to oil spill 48 49 Figure 2 presents the YLD according to disease, sex, and age group. The YLD for 50 51 52 asthma, allergic rhinitis, dermatitis, conjunctivitis, PTSD, and depression were approximately 53 54 6,138 DALYs, 2,420 DALYs, 1,417 DALYs, 18 DALYs, 2,869 DALYs, and 1,862 DALYs, 55 56 57 13 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 25 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 respectively. The total YLD attributable to the Hebei Spirit oil spill for the year 2008 was 5 6 14,724 DALYs. The YLD for males was 7,425 DALYs and for females was 7,290 DALYs. 7 8 The YLD for males for PTSD and depression were higher than for females, and the YLD for 9 10 11 females was higher in asthma and allergies than for males (More details are shown in data 12 13 supplement Table S2). Among the included diseases, the YLD for asthma was the highest. 14 15 The YLD forFor asthma and peer allergies was thereview highest for residents only in their 40s, and the YLD for 16 17 mental disease was the highest for residents in their 20s. 18 19 20 Figure 2 21 22 23 24 DISCUSSION 25 26 27 28 29 The total YLD attributable to the Hebei Spirit oil spill was estimated at 14,724 DALYs 30 31 for 2008. Asthma was found to represent the most prominent disease burden (6,138 DALYs) 32 33 34 among the seven diseases, followed by PTSD and rhinitis. The asthma burden of 6,138

35 http://bmjopen.bmj.com/ 36 DALYs in the study area can be translated to 9,233 DALYs per 100,000 individuals, or 6.5 37 38 times higher in contrast to the total burden of asthma for South Korea of 1,418 DALYs per 39 40 100,000 individuals.[20] 41 42 Kim et al.[21] estimated the prevalence and economic costs of allergic rhinitis in South

43 on October 1, 2021 by guest. Protected copyright. 44 45 Korea for 2007 and reported that a total of 4,068,517 people were identified as having 46 47 allergic rhinitis. The direct cost was $223.68 million and the lost productivity was estimated 48 49 to be $49.25 million, for a total economic burden of $272.9 million in 2007. Applying this 50 51 method to the current study, with an excess incidence of 3,625 persons, the cost of additional 52 53 54 allergic rhinitis due to the oil spill is roughly estimated to be $243,170 for one year (2008). 55 56 57 14 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 These results indicate that oil spills near coastlines can cause considerable adverse health 5 6 effects and that the summation of disease burden should not be neglected, even though the 7 8 size of the population affected may be small. 9 10 11 The health impacts of oil spills on residents of contaminated areas vary by age, sex, and 12 13 region. Approximately one year after the accident, the burden of mental disease attributable 14 15 to the oil spillFor was higher peer in men than inreview women, whereas the onlyburden of allergies was higher 16 17 in women than men. The most heavily impacted area, Area 1, comprised the majority of the 18 19 20 BOD attributable to the oil spill. For example, about 60% of the additional incidences of 21 22 disease occurred in Area 1, less than 1.1 km from the contaminated coast, with 22% in Area 2 23 24 and 6% in Area 3. Critical health effects of oil spills are more likely to affect clean-up 25 26 workers due to higher and repeated exposures to hazardous pollutants. Residents of Area 1 27 28 participated in clean-up work for an average of 122 days, and residents of Areas 2 and 3 29 30 31 participated for averages of 50 and 36 days. Previous studies on the acute health impact of oil 32 33 spills [4-6] also reported that more frequent and greater exposure to clean-up activities was 34

35 strongly associated with a higher occurrence of symptoms, including both acute and chronic http://bmjopen.bmj.com/ 36 37 health outcomes. 38 39 40 We estimated the excess incidences of illnesses caused by an oil spill using prevalence data 41 42 classified by distance from the contaminated coast, under the assumption that the area farthest

43 on October 1, 2021 by guest. Protected copyright. 44 distant, Area 4, was not affected by the oil spill, and that the prevalence of diseases was 45 46 similar in all study areas before the accident. These assumptions were used because real data 47 48 describing disease prevalence before and after the oil spill were not available. Residents in 49 50 51 Area 4, the reference area, had very little participation in the clean-up efforts, which could 52 53 account for differences in disease prevalence between Area 4 and other areas. In addition, 54 55 seventy-two hours after the oil spill, the downwind direction changed from the southeast to 56 57 15 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 25 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 the west and as a result, volatile components did not move further inland.[15] The benzene 5 6 concentrations in the atmosphere after the Hebei Spirit oil spill simulated in our previous 7 8 study [22] also showed that concentrations reduced rapidly with distance from the 9 10 11 contaminated coastline. Benzene concentrations over 0.1 ppm were not detected at the area 12 13 more than 13 km from the spill site. Therefore, the assumption that Area 4 was not affected 14 15 by the oil spillFor is reasonable. peer Kim et review al.[23] reported that only the prevalence of rhinitis in 16 17 Korean population was 16.4%, 24.7%, and 21.7% in 20-35, 36-50, and older than 50 age 18 19 20 group, respectively, while that of Area 4 adults was 15.8% and 17.9% in 19-49 and over 50 21 22 age group, respectively, showing slightly lower prevalence than whole Korean population. 23 24 It is possible that PTSD and depression may overlap. According to a previous cross- 25 26 sectional study of psychological disease in Korea, about 8.5% of depression is accompanied 27 28 by PTSD.[24] This indicates that the estimated disease burden may have been overestimated. 29 30 31 We did not include anxiety in our analyses, although prevalence data for anxiety were 32 33 available. 34

35 Although the PAHs and VOCs in spilled oil can cause haematologic cancer, lung cancer, http://bmjopen.bmj.com/ 36 37 and stomach cancer,[6] we did not assess the incidences of any cancers due to the lack of data. 38 39 40 The TEHC analysed the urinary malondialdehyde (MDA) and 8-hydroxy-2 -deoxyguanosine 41 42 (8-OHdG) levels of residents in the contaminated area, as oxidative stress indices and the

43 on October 1, 2021 by guest. Protected copyright. 44 TEHC found that the urinary MDA and 8-OHdG concentrations of the residents living in 45 46 Area 1 were much higher than in other areas. The average urinary MDA concentration of 47 48 residents in Area 1 was 5.42 (SD=3.14) µmol/g Cr, whereas that of other areas was 3.83 49 50 51 (SD=2.58) µmol/g Cr.[14] The average 8-OHdG concentration for Area 1 was 6.84 (SD=3.84) 52 53 µg/g Cr, while that of other areas was 4.67 (SD=3.00) µg/g Cr. Previous studies reported that 54 55 oxidative DNA damage is associated with increased risk of cancer, including haepatocellular 56 57 16 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 carcinoma, lung cancer, and breast cancer, and thus the MDA and haepatic 8-OHdG levels are 5 6 useful as markers to identify subjects at extremely high-risk.[25-27] The increased urinary 7 8 MDA and 8-OHdG concentrations in the residents in this study area may cause cancers in the 9 10 11 future. By monitoring more long-term effects, the BOD imposed by these diseases may be 12 13 quantified and added to the results of this study. 14 15 This is theFor first study peer to quantify the review BOD of a single environmental only disaster. The EBD 16 17 of the current study was estimated for a highly exposed population after a single event. This 18 19 20 study demonstrates that BOD is an objective and comprehensive metric for estimating and 21 22 comparing the health effects of environmental hazards and disasters across different regions 23 24 and time periods. 25 26 Oil spills near coastlines can cause considerable adverse health effects. The BOD for one 27 28 year for the residents living near contaminated coastal areas is significant and is related to 29 30 31 proximity to the spill as well as participation in clean-up efforts. Asthma and PTSD were 32 33 found to comprise the most prominent disease burden in the contaminated areas. The health 34

35 impacts of oil spills on residents vary with age, sex, and region. The closer the population is http://bmjopen.bmj.com/ 36 37 to the accident site, the greater the BOD is for that population and accident. As the health 38 39 40 effects of hazardous pollutants from oil spills are long-lasting, close follow-up studies are 41 42 required to estimate the cumulative disease burden of oil spills based on the identification of

43 on October 1, 2021 by guest. Protected copyright. 44 long-term health effects. 45 46 47 48 49 50 51 52 53 54 55 56 57 17 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 25 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 Acknowledgements 5 6 This work was supported by the Taean Environmental Health Center. The authors would like 7 8 to thank Woo-Cheol Jung and Jongil Hur for initiating the study and providing the funding 9 10 and data. 11 12 13 14 Competing interests 15 For peer review only 16 17 The authors have no competing interests. 18 19 20 21 22 Funding 23 24 This work was supported through the Taean Environmental Health Center by the Ministry of 25 26 Environment, Republic of Korea. 27 28

29 30 31 Data sharing 32 33 This study was conducted as a part of a project on longterm health effects of oil spill on 34

35 http://bmjopen.bmj.com/ 36 human health granted by Ministry of Environment, Republic of Korea, under the custody of 37 38 Taean Environmental Health Center. Data of the health outcome may be shared for the 39 40 researchers if their proposal meets the aim of the project. 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 Contributorship 46 47 Young-Min Kim: Prepared a manuscript of this paper and conducted statistical analysis. 48 49 Jae-Hyun Park: Supervised the technical aspect of estimation of burden of disease. 50 51 Conducted a survey on the disability weight. 52 53 54 Kyusik Choi: Conducted statistical analysis and participated in the discussion section. 55 56 57 18 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 Su-Ryeon Noh: Provided data and participated in the discussion of the outcome. 5 6 Young-Hyun Choi: Provided a framework of interpretation of findings and participated in the 7 8 discussion of the outcome. 9 10 11 Hae-Kwan Cheong: Designed the study. Supervised the estimation of exposure and health 12 13 outcome. Edited and finalized the manuscript. 14 15 For peer review only 16 17 18 References 19 20 1. County chief of Taean. Taean Tonggyeyeongam, No.18. Taean: Planning and Inspection 21 22 Office of Taean County 2008. http://law.taean.go.kr/asp/tong/tong_file/2007/02.pdf 23 24 (date accessed 5 Oct 2010). 25 26 27 2. Ji K, Seo J, Liu X, et al. Genotoxicity and endocrine-disruption potentials of sediment 28 29 near an oil spill site: two years after the Hebei Spirit oil spill. Environ Sci Technol 30 31 2011;45:7481-7488. 32 33 3. World Health Organization, International Agency for Research on Cancer. IARC 34

35 http://bmjopen.bmj.com/ 36 Monographs on the Evaluation of Carcinogenic Risks to Humans overall Evaluations of 37 38 Carcinogenicity: An Updating of IARC Monographs Volumes 1 to 42 (Supplement 7). 39 40 Lyon: International Agency for Research on Cancer 1998. 41 42 4. Sim MS, Jo IJ, Song HG. Acute health problems related to the operation mounted to clean

43 on October 1, 2021 by guest. Protected copyright. 44 the Hebei Spirit oil spill in Taean, Korea. Mar Pollut Bull 2010;60:51-57. 45 46 47 5. Zock JP, Rodrı´guez-Trigo G, Pozo-Rodrı´guez F, et al. SEPAR-Prestige Study Group: 48 49 prolonged respiratory symptoms in clean-up workers of the Prestige oil spill. Am J Resp 50 51 52 Crit Care Med 2007;176:610-616. 53 54 55 56 57 19 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 25 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 6. Rodrı´guez-Trigo G, Zock JP, Pozo-Rodrı´guez F, et al. SEPAR-Prestige Study Group: 5 6 Health changes in fishermen 2 years after clean-up of the Prestige oil spill. Ann Intern 7 8 Med 2010;153:489-498. 9 10 11 7. Lyons RA, Temple JM, Evans D, et al. Acute health effects of the Sea Empress oil spill. J 12 13 Epidemiol Community Health 1999;53:306-331. 14 15 8. Song M,For Hong YC , Cheongpeer HK, et reviewal. Psychological health only in residents participating in 16 17 clean-up works of Hebei Spirit oil spill. J Prev Med Public Health 2009;42:82-88. 18 19 20 [Korean] 21 22 9. Palinkas LA, Russell J, Downs MA, et al. Ethnic differences in stress, coping, and 23 24 depressive symptoms after the Exxon Valdez oil spill. J Nerv Ment Dis 1992;180:287- 25 26 295. 27 28 10. Palinkas LA, Petterson JS, Russell J, et al. Community patterns of psychiatric disorders 29 30 31 after the Exxon Valdez oil spill. Am J Psychiatr 1993;150:1517-1523. 32 33 11. Palinkas LA, Petterson JS, Russell JC, et al. Ethnic differences in symptoms of 34

35 posttraumatic stress after the Exxon Valdez oil spill. Prehosp Disaster Med 2004;19:102- http://bmjopen.bmj.com/ 36 37 112. 38 39 40 12. Lee CH, Kang YA, Chang KJ, et al. Acute health effects of the Hebei oil spill on the 41 42 residents of Taean, Korea. J Prev Med Public Health 2010;43:166-73. [Korean]

43 on October 1, 2021 by guest. Protected copyright. 44 13. Murray CJ, Lopez AD. The Global Burden of Disease: A Comprehensive Assessment of 45 46 Mortality and Disability from Disease, Injuries and Risk Factors in 1990 and Projected to 47 48 2020. Cambridge, MA: Harvard University Press1996. 49 50 51 14. Taean Environmental Health Center, Ministry of Environment, Republic of Korea. Mid 52 53 and Long Term Health Effect of Hebei Spirit Oil Spill: First Year Survey Report. Taean: 54 55 Taean Environmental Health Center 2011.[Korean] 56 57 20 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 15. Kim JH, Kwak BK, Ha M, et al. Modeling human exposure levels to airborne volatile 5 6 organic compounds the by Hebei Spirit oil spill. Environ Health Toxicol 7 8 2012;27:e2011006. 9 10 11 16. Barendregt JJ, Oortmarssen GJ, Vos T, et al. A generic model for the assessment of 12 13 disease epidemiology: the computational basis of DisMod II. Popul Health Metr 14 15 2003;1:4-10.For peer review only 16 17 17. Statistics Korea. Annual Report on the Cause of Death Statistics (2008). Seoul: Statistics 18 19 20 Korea 2009. 21 22 18. Public Health Division, Department of Human Services, Australia. Victorian Burden of 23 24 Disease Study: Morbidity. Melbourne: Public Health Division 1999. 25 26 19. Stoudhard MEA. Disability weights for diseases: a modified protocol and results for a 27 28 Western European region. Eur J Public Health 2000;10:24-30. 29 30 31 20. Yoon SJ, Bae SC, Lee SI, et al. Measuring the burden of disease in Korea. J Korean Med 32 33 Sci 2007;22:518-523. 34

35 21. Kim SY, Yoon SJ, Jo MW, et al. Economic burden of allergic rhinitis in Korea, Am J http://bmjopen.bmj.com/ 36 37 Rhinol Allergy 2010;24:e110-e113. 38 39 40 22. Kim YM, Cheong HK, Kim J, et al. Scientific basis of environmental health contingency 41 42 planning for a coastal oil spill. J Prev Med Public Health 2009;42:73-81. [Korean]

43 on October 1, 2021 by guest. Protected copyright. 44 23. Kim TB, Kim YK, Chang YS, et al. Association between sensitization to outdoor spider 45 46 mites and clinical manifestations of asthma and rhinitis in the general population of 47 48 adults. J Korean Med Sci 2006;21:247-52. 49 50 51 24. Ministry of Health and Welfare, College of Medicine, Seoul National University. The 52 53 Epidemiological Survey of Psychiatric Illnesses in Korea. Seoul: Ministry of Health and 54 55 Welfare 2006. [Korean] 56 57 21 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 25 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 25. Gonenc A, Oezkan Y, Torun M, et al. Plasma malondialdehyde (MDA) levels in breast 5 6 and lung cancer patients. J Clin Pharm Ther 2001;26:141-144. 7 8 26. Suzana S, Normah H, Fatimah A, et al. Antioxidant intake and status, and oxidative 9 10 11 stress in relation to breast cancer risk: a case-control study. Asian Pacific J Cancer Prev 12 13 2008;9:343-350. 14 15 27. Tanaka H,For Fujita N, Sugimotopeer R, et al.review Hepatic oxidative DNAonly damage is associated with 16 17 increased risk for hepatocellular carcinoma in chronic hepatitis C. Br J Cancer 18 19 20 2008;98:580-586. 21 22 23 24 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 22 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 Figures 5 6 7 8 Figure 1. Study area and classification of affected regions. 9 10 ★ 11 : accident site, about 8 km away from the Taean coastline. 12 13 14 15 For peer review only 16 Figure 2. Burden of disease due to the Hebei Spirit oil spill in Taean County and the 17 18 Boryeong Islands, Korea. 19 20 21 YLD: years lived with disability; Mental disease includes post-traumatic stress disorder 22 23 (PTSD) and depression; Allergy includes rhinitis, dermatitis, and conjunctivitis. 24 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 23 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 25 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 254x191mm (300 x 300 DPI) 34 35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 25 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright.

44 395x435mm (300 x 300 DPI) 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

Burden of disease attributable to the Hebei Spirit oil spill in Taean, Korea

ForJournal: peerBMJ Open review only Manuscript ID: bmjopen-2013-003334.R1

Article Type: Research

Date Submitted by the Author: 13-Jul-2013

Primary Subject Public health Heading:

Secondary Subject Heading: Epidemiology, Health economics, Occupational and environmental medicine

EPIDEMIOLOGY, HEALTH ECONOMICS, Epidemiology < THORACIC http://bmjopen.bmj.com/ Keywords: MEDICINE, MENTAL HEALTH, TOXICOLOGY

Note: The following files were submitted by the author for peer review, but cannot be converted to PDF. You must view these files (e.g. movies) online.

BOD of Taean_BMJ Open_revised.docx on October 1, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 Burden of disease attributable to the Hebei Spirit oil spill in 5 6 7 2 Taean, Korea 8 9 1 2 2 3 3 10 3 Young-Min Kim , Jae-Hyun Park , Kyusik Choi , Su Ryeon Noh , Young-Hyun Choi , Hae- 11 2 12 4 Kwan Cheong 13 14 1Department of Environmental Health, Rollins School of Public Health, Emory University, 15 5 For peer review only 16 17 6 1518 Clifton Road NE, Atlanta, GA, USA 18 2 19 7 Department of Social and Preventive Medicine, Sungkyunkwan University School of 20 21 8 Medicine, 2066 Seobu-ro, Jangan-gu, Suwon, 440-746, Republic of Korea 22 23 3 24 9 Taean Environmental Health Center, 1952-16 Seohae-ro, Taean, Chungnam, 357-902, Re 25 26 10 public of Korea 27 28 11 29 30 12 * Corresponding author: Hae-Kwan Cheong 31 32 2066 Seobu-ro, Jangan-gu, Suwon, Gyeonggi-do, 440-746, Korea 33 13 34

35 14 Tel: +82-31-299-6300, Fax: +82-031-299-6299 http://bmjopen.bmj.com/ 36 37 15 E-mail: [email protected] 38 39 16 40 41 17 * Keywords: oil spill, years lived with disability (YLD), disability-adjusted life-year (DALY), 42

43 on October 1, 2021 by guest. Protected copyright. 44 18 polycyclic aromatic hydrocarbons (PAHs), post-traumatic stress disorder (PTSD), disability 45 46 19 weight 47 48 20 Word count: 49 21 Abstract: 297 words 50 51 22 Main text: 3352 words 52 23 No. of tables: 3 53 24 No. of Figures: 2 54 25 No. of references: 29 55 56 57 1 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 ABSTRACT 5 6 2 Objectives: We aimed at assessing the burden of disease (BOD) of the residents living in 7 8 contaminated coastal area with oil spill and also analyzed the BOD attributable to the oil spill 9 3 10 11 4 by disease, age, sex, and sub-region. 12 13 5 Design: Health impact assessment by measuring years lived with disability (YLD) due to an 14 15 6 oil spill. For peer review only 16 17 18 7 Setting: A whole population of a community affected by an anthropogenic environmental 19 20 8 disaster and secondary health outcome data. 21 22 23 9 Participants: For the health outcome survey, 10,171 individuals (male: 4,354; female: 5,817) 24 25 10 participated, covering 66,473 populations (male: 33,441; female: 33,032) living in the 26 27 11 contaminated coastal area. 28 29 12 Interventions: None. Observational study on the effect of a specific environmental health 30 31 32 13 hazard. 33 34 14 Primary and secondary outcome measures: Using YLD method, BOD including both

35 http://bmjopen.bmj.com/ 36 15 physical and mental diseases was measured. For the YLD measurement, excess incidences of 37 38 16 illnesses related to oil spill were estimated from the comparison of prevalence of the health 39 40 17 outcomes between contaminated areas and reference area without contamination. 41 42 18 Results: YLD attributable to the oil spill were estimated to be 14,724 DALYs (male: 7,425

43 on October 1, 2021 by guest. Protected copyright. 44 45 19 DALYs; female: 7,299 DALYs) for the year 2008. The YLD of mental diseases including 46 47 20 post-traumatic stress disorder (PTSD) and depression for males were higher than that for 48 49 21 females. The YLD for females was higher in asthma and allergies (rhinitis, dermatitis, 50 51 52 22 conjunctivitis) than that for males. The effects of asthma and allergies were the greatest for 53 54 23 people in their 40s, with the burden of mental illness being the greatest for those in their 20s. 55 56 57 2 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 Proximity to the spill site was associated with increased burden of disease. 5 6 2 Conclusions: An oil spill near a coastline can cause substantial adverse health effects. As the 7 8 health effects of hazardous pollutants from oil spills are long-lasting, close follow-up studies 9 3 10 11 4 are required to identify chronic health effects. 12 13 5 Trial registration: Not applicable 14 15 For peer review only 16 6 17 18 19 7 INTRODUCTION 20 21 8 On December 7, 2007, the Hebei Spirit oil tanker spilled 12,547 kl (10,900 tons) of crude 22 23 9 oil into the Yellow Sea, approximately 8 km away from the coast of Taean, Korea. This was 24 25 26 10 the largest oil spill in the history of Korea. Spilled oil contaminated over 1,000 km of the 27 28 11 western coast of Korea including the Taean Coast National Park.[1] Crude oil contains 29 30 12 various volatile organic compounds (VOCs) such as benzene, toluene, ethylbenzene, xylenes, 31 32 13 and polycyclic aromatic hydrocarbons (PAHs),[2] all of which are potential air-borne toxins. 33 34 14 Certain VOCs, benzene in particular, are carcinogenic to humans and are associated with 35 http://bmjopen.bmj.com/ 36 37 15 haematologic cancer.[3] Previous studies [4-11] have reported that oil spills have caused 38 39 16 adverse health effects for residents of contaminated areas and clean-up workers. Risk 40 41 17 analyses revealed that more frequent clean-ups and greater exposure to oil are strongly 42

43 on October 1, 2021 by guest. Protected copyright. 18 associated with higher frequencies of psychological symptoms. [8-11] Rodriguez-Trigo [6] 44 45 46 19 reported that participation in a major oil spill clean-up is associated with chromosomal 47 48 20 damage, although they were unable to determine whether oil exposure itself caused the 49 50 21 abnormalities. 51 52 22 After the Hebei Spirit oil spill, more than 2,000,000 people (person-days), including over 53 54 1,200,000 person-days by volunteers, participated in clean-up activities during the first seven 55 23 56 57 3 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 months.[1] Among them were residents of Taean, a highly exposed group that experienced 5 6 2 adverse health effects, including both physical and psychological symptoms.[4, 12] Although 7 8 many studies have reported positive relationships between oil spills and acute and chronic 9 3 10 11 4 health impacts, there have been no studies quantifying the burden of disease (BOD) due to oil 12 13 5 spills, which are necessary to assess the scale of health damage at the population level as well 14 15 6 as the associatedFor compensation peer costs. review only 16 17 7 Therefore, we sought to quantify the adverse health impacts of exposure to oil spills by 18 19 20 8 applying the disability-adjusted life-year (DALY) measure developed by the World Health 21 22 9 Organization (WHO). The DALY measures health gaps as opposed to health expectancies. It 23 24 10 measures the difference between a current situation and an ideal situation where everyone 25 26 11 lives up to the age of the standard life expectancy, and in perfect health. The DALY is based 27 28 on the premise that the best approach for measuring the burden of disease is to use units of 29 12 30 31 13 time. [13] The DALY method has been used to measure the BOD attributable to specific risk 32 33 14 factors including environmental burdens of disease (EBD). This approach can be practical in 34

35 15 that the estimated adverse health burden attributable to a risk factor can be compared with http://bmjopen.bmj.com/ 36 37 16 other BODs and the analyses of disease burdens by disease and region can inform policy 38 39 40 17 prioritization to mitigate adverse health effects for specific exposed populations. 41 42 18 The purposes of this study are to assess the BOD of the residents of the contaminated

43 on October 1, 2021 by guest. Protected copyright. 44 19 Taean coastal area due to the Hebei Spirit oil spill and to analyze the BOD by disease, age, 45 46 20 sex, and sub-region. 47 48

49 21 50 51 22 METHODS 52 53 54 23 Study area and design 55 56 57 4 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 In September 2008, the Taean Environmental Health Center (TEHC) was established 5 6 2 under the support of the Korean Ministry of Environment to monitor the effects of the Hebei 7 8 Spirit oil spill on the health of the residents of the Taean coastal area. The TEHC first 9 3 10 11 4 conducted health surveys on the residents, including the referent population, one and a half 12 13 5 years after the spill, establishing a cohort. Total of 10,171 residents living in the study area 14 15 6 were surveyedFor including peer 9,246 adults (male:review 3,849; female: 5,397)only and 925 school students 16 17 7 (male: 505; female: 420 persons). The survey included a comprehensive questionnaire 18 19 20 8 gathering personal history on exposure and medical problems, and a structured questionnaire 21 22 9 assessing psychological health and asthma, and physical and laboratory examinations of 23 24 10 respiratory, cardiovascular, neurologic, and psychological systems. 25 26 11 The TEHC published its first report in 2010 after analysing the prolonged health effects 27 28 for 1.5 years, and demonstrated significant relationships between exposure concentrations 29 12 30 31 13 and adverse health outcomes.[15] To estimate the BOD caused by the Hebei Spirit oil spill, 32 33 14 we obtained the survey data from the TEHC, which included disease, residence, and 34

35 15 demographic information. Residents in Taean County and some islands in neighboring http://bmjopen.bmj.com/ 36 37 16 Boryeong County were included in the survey. The Boryeong islands were expected to be 38 39 40 17 affected by the oil spill because the islands are located in the affected area based on the wind 41 42 18 direction and tidal current at the time of the spill.[16] Kim et al.[16] reported that the VOCs

43 on October 1, 2021 by guest. Protected copyright. 44 19 moved in a southeasterly direction, following the coastline from the accident point. 45 46 20 We classified the residents of Taean County and the Boryeong islands into five regions 47 48 based on both distance from the contaminated coastline and contamination level. Area 1 is 49 21 50 51 22 less than 1.1 km away from the coastline adjacent to the accident location where the coastline 52 53 23 was contaminated at a high level. Area 2 is 1.1-4.2 km away with no coastline, but was 54 55 24 directly affected by atmospheric VOCs and PAHs due to the wind direction. Area 3 is 4.2-23 56 57 5 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 km away from the coastline with a moderately contaminated coastline, and Area 4 is more 5 6 2 than 23 km from the contaminated coastline. We classified the Boryeong Islands as a separate 7 8 area. Figure 1 shows the locations of the study areas and five classified regions. 9 3 10 11 4 Figure 1 12 13 5 14 15 6 We assessedFor health peer outcome data review regarding asthma, only allergy (rhinitis, dermatitis, 16 17 7 conjunctivitis), and mental disorders (post-traumatic stress disorder (PTSD) and depression) 18 19 20 8 that were well-documented by the TEHC.[15] We focused on subchronic (duration of a few 21 22 9 months) and chronic (over one year duration) adverse health effects of the Hebei Spirit oil 23 24 10 spill, assessing prolonged health effects over a period of 1.5 years to estimate the BOD of the 25 26 11 Hebei Spirit oil spill for the year of 2008. 27 28

29 12 30 31 13 Estimation of burden of disease attributable to oil spill 32 33 14 To estimate the BOD attributable to the oil spill, we applied the DALY method. DALY is 34

35 15 the sum of the life years lost due to disability and premature death. [13] As there was no http://bmjopen.bmj.com/ 36 37 16 premature death identified as the direct effect of the oil spill until the TEHC published the 38 39 40 17 first report in 2010, we included only years lived with disability (YLD). [13] The YLD is 41 42 18 calculated by using disease incidence and prevalence, onset age and duration of disability,

43 on October 1, 2021 by guest. Protected copyright. 44 19 and disability weight (DW). The formula we utilized is as follows; 45 46 20 47 48 49 50 21 51 52 22 [1] 53 54 55 23 where, D is disability weight; r is discount rate; a is age of onset of the disability; L is 56 57 6 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 duration of the disability; is parameter from the age weighting function; and C is constant. 5 β 6 7 8 2 We used 3% of discount rate and 0.04 and 0.0618 for β and C, respectively, as suggested by 9 10 11 3 the WHO.[13] 12 13 4 Because there was no prevalence data recorded prior to the accident, we estimated excess 14 15 5 incidences usingFor prevalence peer data from thereview TEHC under the followingonly assumptions. We first 16 17 assumed that residents living in Area 4 had no negative health consequences due to the oil 18 6 19 20 7 spill and used this as the baseline prevalence. Area 4 was assumed to be the reference area, 21 22 8 not only because the area might not be impacted by the oil spill directly as it is far from the 23 24 9 contaminated coastline, but also because residents in Area 4 rarely participate in the clean-up 25 26 10 works. Very few people participated the clean-up work except for civil servants living in the 27 28 29 11 area 4, which appeared to be an important factor to cause the difference in the prevalence of 30 31 12 disease between area 4 and others.[15] Secondly, we assumed that the prevalence of oil spill- 32 33 13 related diseases in the whole study area had been same prior to the accident. We then 34 35 14 estimated excess incidences using prevalence differences between Area 4 and the other areas http://bmjopen.bmj.com/ 36 37 stratified by area, age, sex, and disease using following formula. 38 15 39 40 16 [2] 41 42 17 here, is the excess incidence of region i and disease j; is the population of

43 on October 1, 2021 by guest. Protected copyright. 44 region i; is the prevalence of region i and disease j; is the prevalence of control 45 18 46 47 19 region and disease j. 48 49 20 The duration of disability and the average onset age were calculated using the DISMOD 50 51 21 II model [17] for PTSD, rhinitis, dermatitis, and conjunctivitis based on prevalence data from 52 53 the TEHC and the mortality rate of South Korea [18] by age, sex, and disease. For asthma, 54 22 55 56 57 7 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 the results of the Victoria, Australia study [19] were used. 5 6 2 DWs for each disease were measured following a method previously used by the Global 7 8 Burden of Disease group [13] and the Dutch Disability Weights Group.[20] A panel that 9 3 10 11 4 included 12 health and medicine specialists was convened. Eleven indicator conditions 12 13 5 including asthma, stomach cancer, diabetes, dementia, major depression, appendicitis, thyroid 14 15 6 cancer, hepatitisFor A, otitis peermedia, atopic dermatitis,review and common only cold were measured using the 16 17 7 person trade-off (PTO) method. The eleven indicator conditions were selected in accordance 18 19 20 8 with a previous study on Korean disability weight. [21] That is, we considered that they 21 22 9 should be meaningful in public health and relatively popular diseases, while distributing the 23 24 10 whole DW range (0 to 1) evenly.[21] A facilitator first explained the method and then led the 25 26 11 discussion to determine the PTO value of each participant. The discussion was repeated twice. 27 28 Then, the visual analogue scale (VAS) method was applied to obtain DWs for 118 conditions 29 12 30 31 13 including asthma, rhinitis, dermatitis, conjunctivitis, PTSD, and depression used in this study. 32 33 14 Inter-method reliability was confirmed between PTO and VAS using 11 indicator conditions, 34

35 15 which showed fairly high consistency (Spearman’s correlation coefficient 0.966, P- http://bmjopen.bmj.com/ 36 37 16 value<0.001). Reliability was confirmed by test-retest after three weeks by seven specialists 38 39 40 17 and obtained fairly high consistency (Spearman’s correlation coefficient 0.973, P- 41 42 18 value<0.001).

43 on October 1, 2021 by guest. Protected copyright. 44 19 Finally, we estimated YLD caused by the Hebei Spirit oil spill utilizing the estimated 45 46 20 excess incidence attributable to the oil spill, DW, and the onset age and duration of disability 47 48 by disease, age, sex, and area. 49 21 50 51 22 52 53 54 23 RESULTS 55 56 57 8 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 5 6 2 Summary of the characteristics of the study area 7 8 The populations of Taean County and the Boryeong islands were 63,053 individuals 9 3 10 11 4 (male: 31,636; female: 31,417) and 3,420 individuals (male: 1,805; female: 1,615) in 2008, 12 13 5 respectively. Broken down by area, 8,348 (13%), 11,272 (17%), 4,597 (7%), 38,836 (58%), 14 15 6 and 3,420 (5%)For residents peer lived in Areas review 1, 2, 3, 4, and the Boryeongonly islands, respectively. 16 17 7 Area 4, the reference area, was the most populous because it includes a city center. The 18 19 20 8 largest age group consisted of subjects in their 50s, with individuals in their 40s, 50s, and 60s 21 22 9 comprising 45% of the total population (data presented in the supplement Table S1). 23 24 10 A total of 10,171 residents living in the study area participated in the TEHC survey. 25 26 11 Among them, 4,354 (42.8%) and 5,817 (57.2%) residents were male and female, respectively. 27 28 Individuals under 15 years of age, 15-65, and over 65 years of age comprised 7.2%, 48.0%, 29 12 30 31 13 and 44.8% of the population, respectively. Residents of Area 1 participated in clean-up work 32 33 14 for an average of 122 days, and residents of Areas 2, 3, and the Boryeong islands participated 34

35 15 for averages of 50, 36, and 100 days, respectively.[14] Table 1 shows the prevalence of http://bmjopen.bmj.com/ 36 37 16 disease related to the current study by disease, sex, and area derived from a survey of 9,246 38 39 40 17 adult (over 19 years old) residents. For young people under 19 years old, we used results 41 42 18 from 925 school students (male: 505; female: 420) living in the study area.

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 9 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from Page 10 of 60 42.6 10.1 40.2 17.0 15.0 24.5 10.8 21.3 13.5 25.2 23.8 27.9 Male Female Boryeong Boryeong Islands

Area 4 2.4 9.2 4.4 10.0 4.6 5.3 3.2 11.6 8.7 7.6 20.2 15.9 Male Female

3.1 10 10 Area 3

BMJ Open 9.4 4.7 2.9 8.8 14.0 9.6 10.4 23.9 10.6 13.2 9.0 Male Female

http://bmjopen.bmj.com/

7.1 Area 2

5.1 6.1 7.0 8.8 14.8 25.2 20.2 12.6 12.6 15.4 11.7 Male Female on October 1, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

a) For peer review only Area 1 22.0 20.4 32.9 34.6 Male Male Female . Prevalence of diseases related to this study (in study %). this related to . of diseases Prevalence b) Rhinitis 41.4 34.6 Dermatitis Conjunctivitis 19.6 32.0 25.0 38.7 Depression 23.3 29.9 PTSD Table 1 Table a) Area 1: less than 1.1 km away; Area 2: 1.1-4.2 km away; Area 3: 4.2-23 km away; Area 4: more than 23 km from the contaminated contaminated the km from 23 than 4: more Area away; km 3: 4.2-23 Area away; km 2: 1.1-4.2 Area away; km 1.1 1: less than Area a) Allergy Allergy Asthma Disease coastline; b) PTSD: posttraumatic stress stress disorder. coastline; b) posttraumatic PTSD: Mental Mental disease 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 11 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 Estimated burden of disease due to oil spill 5 6 2 Excess incidence attributable to the oil spill 7 8 The excess incidence of disease caused by the oil spill varied by area, sex, and disease 9 3 10 11 4 (Table 2). There were additional incidences of asthma (2,088: 989 male; 1,099 female), 12 13 5 rhinitis (3,625: 1,855 male; 1,770 female), dermatitis (1,976: 962 male; 1,014 female), 14 15 6 conjunctivitisFor (2,992: 1,365 peer male; 1,627 review female), PTSD (2,681:1,300 only male; 1,381 female), 16 17 7 and depression (2,326: 1,370 male; 956 female) in Areas 1,2,3 and the Boryeong islands. 18 19 20 8 Area 1 had the highest additional incidences for all diseases, while Area 3, a relatively 21 22 9 less affected area than Areas 1 and 2, showed the lowest additional incidences for all diseases. 23 24 10 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 11 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 12 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 12 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

Total Total

989 1,099 962 1,014 Male Female 1,370 956 1,300 1,381 1,855 1,770 1,365 1,627

76 76 106 113 113 190 116 116 138 201 201 149 353 353 354 222 222 322 Male Female Boryeong Boryeong Islands

0 Area 3 Area 3 13 13 c

0 65 65 57 53 53 97 94 94 32 209 207 101 183 Male Male Female BMJ Open

http://bmjopen.bmj.com/

Area 2

266 169 133 222 453 481 241 231 294 214 146 158 Male Female

on October 1, 2021 by guest. Protected copyright.

a For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Area 1

839 839 729 473 473 449 838 838 581

Table 2. 2. Table a: Area 1, less than 1.1 km away; Area area. and reference 3 Area 2, between difference c: No disorder; stress coastline; b: post-traumatic 1.1-4.2 km away; Area 3, 4.2-23 km away; Area 4, more than 23 km from the contaminated Allergy Allergy Asthma Disease Depression PTSD Page 13 of 60 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from Page 14 of 60 14 14 BMJ Open http://bmjopen.bmj.com/ on October 1, 2021 by guest. Protected copyright. For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 15 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 Disability weight 5 6 2 The VAS results showed that the DW for stomach cancer (stage 4) was the highest at 7 8 0.909, and that the DW for upper respiratory infection was the lowest at 0.065, out of a total 9 3 10 11 4 of 118 conditions surveyed. Among the 118 conditions, DWs for diseases examined in this 12 13 5 study are shown in Table 3. We used the DW of mild intermittent asthma, 0.222, to represent 14 15 6 asthma, becauseFor the prevalence peer of asthma review in this study area only was estimated using survey 16 17 7 results following the modified International Study of Asthma and Allergies in Childhood 18 19 20 8 (ISSAC), and most cases of asthma were not severe.[15] 21 22 9 23 24 10 Table 3. Disability weights used in the current study 25 26 Disability 27 Disease Stage

28 weight Asthma 29 Mild intermittent 0.222 30 Mild persistent 0.291 31 Moderate persistent 0.510 32 Severe persistent 0.715 33 34 Very severe persistent 0.836

35 Allergy Rhinitis 0.205 http://bmjopen.bmj.com/ 36 Dermatitis 0.132 37 38 Conjunctivitis 0.212

39 a 40 PTSD 0.533 41 Depression 0.219 42 a: post-traumatic stress disorder 43 11 on October 1, 2021 by guest. Protected copyright. 44 45 12 46 47 13 Burden of disease due to oil spill 48 49 Figure 2 presents the YLD according to disease, sex, and age group. The YLD for 50 14 51 52 15 asthma, allergic rhinitis, dermatitis, conjunctivitis, PTSD, and depression were approximately 53 54 16 6,138 DALYs, 2,420 DALYs, 1,417 DALYs, 18 DALYs, 2,869 DALYs, and 1,862 DALYs, 55 56 57 15 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 respectively. The total YLD attributable to the Hebei Spirit oil spill for the year 2008 was 5 6 2 14,724 DALYs. The YLD for males was 7,425 DALYs and for females was 7,290 DALYs. 7 8 The YLD for males for PTSD and depression were higher than for females, and the YLD for 9 3 10 11 4 females was higher in asthma and allergies than for males (More details are shown in data 12 13 5 supplement Table S2). Among the included diseases, the YLD for asthma was the highest. 14 15 6 The YLD forFor asthma and peer allergies was thereview highest for residents only in their 40s, and the YLD for 16 17 7 mental disease was the highest for residents in their 20s. 18 19 20 8 Figure 2 21 22 9 23 24 10 DISCUSSION 25 26 27 11 28 29 12 The total YLD attributable to the Hebei Spirit oil spill was estimated at 14,724 DALYs 30 31 for 2008. Asthma was found to represent the most prominent disease burden (6,138 DALYs) 32 13 33 34 14 among the six diseases, followed by PTSD and rhinitis. The asthma burden of 6,138 DALYs

35 http://bmjopen.bmj.com/ 36 15 in the study area can be translated to 9,233 DALYs per 100,000 individuals, or 6.5 times 37 38 16 higher in contrast to the total burden of asthma for South Korea of 1,418 DALYs per 100,000 39 40 17 individuals.[22] Kim et al.[23] estimated the prevalence and economic costs of allergic 41 42 18 rhinitis in South Korea for 2007 and reported that a total of 4,068,517 people were identified

43 on October 1, 2021 by guest. Protected copyright. 44 45 19 as having allergic rhinitis. The direct cost was $223.68 million and the lost productivity was 46 47 20 estimated to be $49.25 million, for a total economic burden of $272.9 million in 2007. 48 49 21 Applying this method to the current study, with an excess incidence of 3,625 persons, the cost 50 51 of additional allergic rhinitis due to the oil spill is roughly estimated to be $243,170 for one 52 22 53 54 23 year (2008). These results indicate that oil spills near coastlines can cause considerable 55 56 57 16 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 adverse health effects and that the summation of disease burden should not be neglected, even 5 6 2 though the size of the population affected may be small. 7 8 The health impacts of oil spills on residents of contaminated areas vary by age, sex, and 9 3 10 11 4 region. Approximately one year after the accident, the burden of mental disease attributable 12 13 5 to the oil spill was higher in men than in women, whereas the burden of allergies was higher 14 15 6 in women thanFor men. The peer most heavily impactedreview area, Area 1, onlycomprised the majority of the 16 17 7 BOD attributable to the oil spill. For example, about 60% of the additional incidences of 18 19 20 8 disease occurred in Area 1, less than 1.1 km from the contaminated coast, with 22% in Area 2 21 22 9 and 6% in Area 3. Critical health effects of oil spills are more likely to affect clean-up 23 24 10 workers due to higher and repeated exposures to hazardous pollutants. Residents of Area 1 25 26 11 participated in clean-up work for an average of 122 days, and residents of Areas 2 and 3 27 28 participated for averages of 50 and 36 days. Previous studies on the acute health impact of oil 29 12 30 31 13 spills [4-6] also reported that more frequent and greater exposure to clean-up activities was 32 33 14 strongly associated with a higher occurrence of symptoms, including both acute and chronic 34

35 15 health outcomes. http://bmjopen.bmj.com/ 36 37 16 To estimate the excess incidences of illnesses caused by an oil spill, we used prevalence 38 39 40 17 data classified by distance from the contaminated coast, under the assumption that the area 41 42 18 farthest distant, Area 4, was not affected by the oil spill, and that the prevalence of diseases

43 on October 1, 2021 by guest. Protected copyright. 44 19 was similar in all study areas before the accident, which is one of limitations of this study. 45 46 20 These assumptions were used because real data describing disease prevalence before and 47 48 after the oil spill were not available. Residents in Area 4, the reference area, had very little 49 21 50 51 22 participation in the clean-up efforts, which could account for differences in disease 52 53 23 prevalence between Area 4 and other areas. In addition, seventy-two hours after the oil spill, 54 55 24 the downwind direction changed from the southeast to the west and as a result, volatile 56 57 17 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 components did not move further inland.[16] The benzene concentrations in the atmosphere 5 6 2 after the Hebei Spirit oil spill simulated in our previous study [24] also showed that 7 8 concentrations reduced rapidly with distance from the contaminated coastline. Benzene 9 3 10 11 4 concentrations over 0.1 ppm were not detected at the area more than 13 km from the spill site. 12 13 5 Therefore, the assumption that Area 4 was not affected by the oil spill is reasonable. Kim et 14 15 6 al.[25] reportedFor that the peerprevalence of rhinitisreview in Korean population only was 16.4%, 24.7%, and 16 17 7 21.7% in 20-35, 36-50, and older than 50 age group, respectively, while that of Area 4 adults 18 19 20 8 was 15.8% and 17.9% in 19-49 and over 50 age group, respectively, showing slightly lower 21 22 9 prevalence than whole Korean population. 23 24 10 It is possible that PTSD and depression may overlap. According to a previous cross- 25 26 11 sectional study of psychological disease in Korea, about 8.5% of depression is accompanied 27 28 by PTSD.[26] This indicates that the estimated disease burden may have been overestimated. 29 12 30 31 13 For this reason, we did not include anxiety in our analyses, although prevalence data for 32 33 14 anxiety were available because about 15% of anxiety is accompanied by depression in 34

35 15 Korea.[26] http://bmjopen.bmj.com/ 36 37 16 Although the PAHs and VOCs in spilled oil can cause haematologic cancer, lung cancer, 38 39 40 17 and stomach cancer,[6] we did not assess the incidences of any cancers due to the lack of data. 41 L 42 18 The TEHC analysed the urinary malondialdehyde (MDA) and 8-hydroxy-2 -deoxyguanosine

43 on October 1, 2021 by guest. Protected copyright. 44 19 (8-OHdG) levels of residents in the contaminated area, as oxidative stress indices and the 45 46 20 TEHC found that the urinary MDA and 8-OHdG concentrations of the residents living in 47 48 Area 1 were much higher than in other areas. The average urinary MDA concentration of 49 21 50 51 22 residents in Area 1 was 5.42 (SD=3.14) µmol/g Cr, whereas that of other areas was 3.83 52 53 23 (SD=2.58) µmol/g Cr.[15] The average 8-OHdG concentration for Area 1 was 6.84 54 55 24 (SD=3.84) µg/g Cr, while that of other areas was 4.67 (SD=3.00) µg/g Cr. Previous studies 56 57 18 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 reported that oxidative DNA damage is associated with increased risk of cancer, including 5 6 2 haepatocellular carcinoma, lung cancer, and breast cancer, and thus the MDA and haepatic 8- 7 8 OHdG levels are useful as markers to identify subjects at extremely high-risk.[27-29] The 9 3 10 11 4 increased urinary MDA and 8-OHdG concentrations in the residents in this study area may 12 13 5 cause cancers in the future. By monitoring more long-term effects, the BOD imposed by 14 15 6 these diseasesFor may be quantified peer and added review to the results of this only study. 16 17 7 This is the first study to quantify the BOD of a single environmental disaster. The EBD 18 19 20 8 of the current study was estimated for a highly exposed population after a single event. This 21 22 9 study demonstrates that BOD is an objective and comprehensive metric for estimating and 23 24 10 comparing the health effects of environmental hazards and disasters across different regions 25 26 11 and time periods. One of strengths of this study is that we measured BOD attributable to oil 27 28 spill by sub-region as well as age and sex including both physical and mental health 29 12 30 31 13 outcomes, which can contribute to prioritize the mitigation policy to reduce adverse health 32 33 14 effects. 34

35 15 http://bmjopen.bmj.com/ 36 37 38 16 Conclusions 39 40 17 Oil spills near coastlines can cause considerable adverse health effects. The BOD for one 41 42 18 year for the residents living near contaminated coastal areas is significant and is related to

43 on October 1, 2021 by guest. Protected copyright. 44 45 19 proximity to the spill as well as participation in clean-up efforts. Asthma and PTSD were 46 47 20 found to comprise the most prominent disease burden in the contaminated areas. The health 48 49 21 impacts of oil spills on residents vary with age, sex, and region implying that rehabilitation 50 51 policy should be community-specific. 52 22 53 54 55 56 57 19 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 As the health effects of hazardous pollutants from oil spills are long-lasting, close 5 6 2 follow-up studies are required to estimate the cumulative disease burden of oil spills based on 7 8 the identification of long-term health effects. 9 3 10 11 4 12 13 14 5 ARTICLE SUMMARY 15 For peer review only 16 17 6 Article focus 18 7 ▶ This study aimed at assessing the burden of disease (BOD) of the residents 19 8 living in the contaminated area due to an environmental disaster, oil spill, and 20 21 9 analyzed the BOD attributable to the oil spill by disease, age group, and sub- 22 10 region, which can inform policy prioritization to mitigate adverse health effects 23 11 for specific exposed populations. 24 25 12 26 13 Key messages 27 14 ▶ This study provides an opportunity to assess the disease burden of the 28 29 15 residents due to an environmental disaster, integrating mental and physical 30 16 symptoms. 31 32 17 ▶ Among diseases, years lived with disabilities (YLDs) for mental disease such as 33 post-traumatic stress disorder (PTSD) and depression attributable to the oil spill 34 18

35 19 are higher in males than females and those for asthma and allergies are higher http://bmjopen.bmj.com/ 36 20 in females than males. 37 38 21 ▶ Oil spills near coastlines can cause considerable adverse health effects and that 39 22 the summation of disease burden is not negligible, even though the size of the 40 23 population affected may be small. 41 24 42 25 Strength and limitations

43 on October 1, 2021 by guest. Protected copyright. 44 26 ▶ This is the first study to quantify the BOD of a single environmental disaster. 45 46 27 ▶ This study demonstrates that BOD is an objective and comprehensive metric for 47 28 estimating and comparing the health effects of environmental hazards and 48 29 disasters across different regions and time periods. 49 50 30 ▶To estimate excess incidence, we used prevalence data classified by distance 51 31 from the contaminated coast, under the assumption that the area farthest 52 32 distant, Area 4, was not affected by the oil spill, and that the prevalence of 53 33 diseases was similar in all study areas before the accident because real data 54 34 describing disease prevalence before and after the oil spill were not available. 55 56 57 20 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 We did not consider long-term health outcomes such as cancers and chronic 5 ▶ 6 2 degenerative diseases mainly because of lack of data. 7 3 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 21 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 5 6 2 Acknowledgements 7 3 This work was supported by the Taean Environmental Health Center. The 8 9 10 4 authors would like to thank Woo-Cheol Jung and Jongil Hur for initiating the 11 12 5 study and providing the funding and data. 13 14 15 6 For peer review only 16 17 7 Competing interests 18 19 20 8 The authors have no competing interests. 21 22 23 9 24 25 10 Funding 26 27 28 11 This work was supported through the Taean Environmental Health Center by 29 30 12 the Ministry of Environment, Republic of Korea. 31 32 33 13 34

35 http://bmjopen.bmj.com/ 36 14 Contributorship statement 37 38 15 Young-Min Kim: Prepared a manuscript of this paper and conducted statistical 39 40 41 16 analysis. 42

43 on October 1, 2021 by guest. Protected copyright. 17 Jae-Hyun Park: Supervised the technical aspect of estimation of burden of 44 45 46 18 disease. Conducted a survey on the disability weight. 47 48 49 19 Kyusik Choi: Conducted statistical analysis and participated in the discussion 50 51 20 section. 52 53 54 21 Su-Ryeon Noh: Provided data and participated in the discussion of the outcome. 55 56 57 22 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 Young-Hyun Choi: Provided a framework of interpretation of findings and 5 6 7 2 participated in the discussion of the outcome. 8 9 3 Hae-Kwan Cheong: Designed the study. Supervised the estimation of exposure 10 11 12 4 and health outcome. Edited and finalized the manuscript. 13 14 15 5 For peer review only 16 17 6 Data sharing 18 19 20 7 This study was conducted as a part of a project on longterm health effects of oil 21 22 8 spill on human health granted by Ministry of Environment, Republic of Korea, 23 24 25 9 under the custody of Taean Environmental Health Center. Data of the health 26 27 28 10 outcome may be shared for the researchers if their proposal meets the aim of the 29 30 11 project. 31 32 33 12 34

35 http://bmjopen.bmj.com/ 36 13 References 37 14 38 15 1. County chief of Taean. Taean Tonggyeyeongam, No.18. Taean: Planning and Inspection 39 40 16 Office of Taean County 2008. http://law.taean.go.kr/asp/tong/tong_file/2007/02.pdf 41 42 17 (date accessed 5 Oct 2010).

43 on October 1, 2021 by guest. Protected copyright. 44 2. Ji K, Seo J, Liu X, et al. Genotoxicity and endocrine-disruption potentials of sediment 45 18 46 47 19 near an oil spill site: two years after the Hebei Spirit oil spill. Environ Sci Technol 48 49 20 2011;45:7481-7488. 50 51 21 3. World Health Organization, International Agency for Research on Cancer. IARC 52 53 22 Monographs on the Evaluation of Carcinogenic Risks to Humans overall Evaluations of 54 55 56 57 23 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 Carcinogenicity: An Updating of IARC Monographs Volumes 1 to 42 (Supplement 7). 5 6 2 Lyon: International Agency for Research on Cancer 1998. 7 8 4. Sim MS, Jo IJ, Song HG. Acute health problems related to the operation mounted to clean 9 3 10 11 4 the Hebei Spirit oil spill in Taean, Korea. Mar Pollut Bull 2010;60:51-57. 12 13 14 5 5. Zock JP, Rodrı´guez-Trigo G, Pozo-Rodrı´guez F, et al. SEPAR-Prestige Study Group: 15 For peer review only 16 6 prolonged respiratory symptoms in clean-up workers of the Prestige oil spill. Am J Resp 17 18 7 Crit Care Med 2007;176:610-616. 19 20 21 8 6. Rodrı´guez-Trigo G, Zock JP, Pozo-Rodrı´guez F, et al. SEPAR-Prestige Study Group: 22 23 9 Health changes in fishermen 2 years after clean-up of the Prestige oil spill. Ann Intern 24 25 10 Med 2010;153:489-498. 26 27 7. Lyons RA, Temple JM, Evans D, et al. Acute health effects of the Sea Empress oil spill. J 28 11 29 30 12 Epidemiol Community Health 1999;53:306-331. 31 32 13 8. Song M, Hong YC, Cheong HK, et al. Psychological health in residents participating in 33 34 14 clean-up works of Hebei Spirit oil spill. J Prev Med Public Health 2009;42:82-88.

35 http://bmjopen.bmj.com/ 36 15 [Korean] 37 38 39 16 9. Palinkas LA, Russell J, Downs MA, et al. Ethnic differences in stress, coping, and 40 41 17 depressive symptoms after the Exxon Valdez oil spill. J Nerv Ment Dis 1992;180:287- 42

43 18 295. on October 1, 2021 by guest. Protected copyright. 44 45 19 10. Palinkas LA, Petterson JS, Russell J, et al. Community patterns of psychiatric disorders 46 47 after the Exxon Valdez oil spill. Am J Psychiatr 1993;150:1517-1523. 48 20 49 50 21 11. Palinkas LA, Petterson JS, Russell JC, et al. Ethnic differences in symptoms of 51 52 22 posttraumatic stress after the Exxon Valdez oil spill. Prehosp Disaster Med 2004;19:102- 53 54 23 112. 55 56 57 24 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 12. Lee CH, Kang YA, Chang KJ, et al. Acute health effects of the Hebei oil spill on the 5 6 2 residents of Taean, Korea. J Prev Med Public Health 2010;43:166-73. [Korean] 7 8 13. Murray CJ, Lopez AD. The Global Burden of Disease: A Comprehensive Assessment of 9 3 10 11 4 Mortality and Disability from Disease, Injuries and Risk Factors in 1990 and Projected to 12 13 5 2020. Cambridge, MA: Harvard University Press1996. 14 15 6 14. Ezzati M,For Lopez AD,peer Rodger HSV, review et al. Selected major only risk factors and global and 16 17 18 7 regional burden of disease. Lancet 2002;360:1347–60. 19 20 21 8 15. Taean Environmental Health Center, Ministry of Environment, Republic of Korea. Mid 22 23 9 and Long Term Health Effect of Hebei Spirit Oil Spill: First Year Survey Report. Taean: 24 25 10 Taean Environmental Health Center 2011.[Korean] 26 27 11 16. Kim JH, Kwak BK, Ha M, et al. Modeling human exposure levels to airborne volatile 28 29 30 12 organic compounds the by Hebei Spirit oil spill. Environ Health Toxicol 31 32 13 2012;27:e2011006. 33 34 14 17. Barendregt JJ, Oortmarssen GJ, Vos T, et al. A generic model for the assessment of

35 http://bmjopen.bmj.com/ 36 15 disease epidemiology: the computational basis of DisMod II. Popul Health Metr 37 38 39 16 2003;1:4-10. 40 41 17 18. Statistics Korea. Annual Report on the Cause of Death Statistics (2008). Seoul: Statistics 42

43 18 Korea 2009. on October 1, 2021 by guest. Protected copyright. 44 45 19 19. Public Health Division, Department of Human Services, Australia. Victorian Burden of 46 47 Disease Study: Morbidity. Melbourne: Public Health Division 1999. 48 20 49 50 21 20. Stoudhard MEA. Disability weights for diseases: a modified protocol and results for a 51 52 22 Western European region. Eur J Public Health 2000;10:24-30. 53 54 23 21. Lee JK, Yoon SJ, Do YK, Kwon YH, et al. Disability weights for diseases in Korea. 55 56 57 25 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 26 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 Korean J Prev Med 2003;36:1-26 (in Korean). 5 6 2 22. Yoon SJ, Bae SC, Lee SI, et al. Measuring the burden of disease in Korea. J Korean Med 7 8 Sci 2007;22:518-523. 9 3 10 11 4 23. Kim SY, Yoon SJ, Jo MW, et al. Economic burden of allergic rhinitis in Korea, Am J 12 13 5 Rhinol Allergy 2010;24:e110-e113. 14 15 6 24. Kim YM,For Cheong HK,peer Kim J, et al. reviewScientific basis of environmental only health contingency 16 17 7 planning for a coastal oil spill. J Prev Med Public Health 2009;42:73-81. [Korean] 18 19 20 8 25. Kim TB, Kim YK, Chang YS, et al. Association between sensitization to outdoor spider 21 22 9 mites and clinical manifestations of asthma and rhinitis in the general population of 23 24 10 adults. J Korean Med Sci 2006;21:247-52. 25 26 11 26. Ministry of Health and Welfare, College of Medicine, Seoul National University. The 27 28 Epidemiological Survey of Psychiatric Illnesses in Korea. Seoul: Ministry of Health and 29 12 30 31 13 Welfare 2006. [Korean] 32 33 14 27. Gonenc A, Oezkan Y, Torun M, et al. Plasma malondialdehyde (MDA) levels in breast 34

35 15 and lung cancer patients. J Clin Pharm Ther 2001;26:141-144. http://bmjopen.bmj.com/ 36 37 16 28. Suzana S, Normah H, Fatimah A, et al. Antioxidant intake and status, and oxidative 38 39 40 17 stress in relation to breast cancer risk: a case-control study. Asian Pacific J Cancer Prev 41 42 18 2008;9:343-350.

43 on October 1, 2021 by guest. Protected copyright. 44 19 29. Tanaka H, Fujita N, Sugimoto R, et al. Hepatic oxidative DNA damage is associated with 45 46 20 increased risk for hepatocellular carcinoma in chronic hepatitis C. Br J Cancer 47 48 2008;98:580-586. 49 21 50 51 22 52 53 54 55 56 57 26 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 27 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 5 6 7 2 Figures 8 9 10 3 Figure 1. Study area and classification of affected regions. 11 12 13 4 ★: accident site, about 8 km away from the Taean coastline. 14 15 5 For peer review only 16 17 18 6 Figure 2. Burdens of disease attributable to the Hebei Spirit oil spill by disease, age and sex. 19 20 21 7 YLD: years lived with disability; PTSD: post-traumatic stress disorder. 22 23 24 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 27 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 28 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 137x105mm (300 x 300 DPI) 34 35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 29 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 248x185mm (300 x 300 DPI) 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 30 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 Data supplement 5 6 7 Table S1. Demographic characteristics by study area (Unit: number of persons) 8 9 Age Area 1 Area 2 Area 3 Area 4 Boryeong islands 10 group 11 Male Female Male Female Male Female Male Female Male Female 12 0 - 9 220 189 361 321 96 90 1,751 1,641 139 115 13 10 - 19 334 321 416 417 192 151 2,425 2,177 192 158 14 15 20 - 29 509For 339 peer 706 439review 279 228 only 2,531 1,963 230 153 16 30 - 39 536 296 697 387 248 162 2,697 2,410 238 167 17 18 40 - 49 667 547 758 606 329 311 3,268 2,960 287 227 19 50 - 59 767 848 876 940 425 436 2,856 3,039 281 271 20 21 60 - 69 660 733 1,001 1,084 397 432 2,266 2,550 247 247 22 70 - 79 445 578 707 1,001 251 366 1,311 1,906 155 198 23 24 80 + 102 257 164 391 65 139 323 762 37 79 25 26 Sum 4,240 4,108 5,686 5,586 2,282 2,315 19,428 19,408 1,805 1,615 27 Total : 66,473 28 29 Area 1: less than 1.1 km away; Area 2: 1.1-4.2 km away; Area 3: 4.2-23 km away; Area 4: 30 31 more than 23 km from the contaminated coastline. 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 Table S2. Prevalence of disease for adults by age, sex, and area (%). 5 Boryeong 6 Age Area 1 Area 2 Area 3 Area 4 Disease islands 7 group 8 Male Female Male Female Male Female Male Female Male Female Asthma 9 19-44 18.2 8.2 10.9 11.1 9.4 10.0 16.1 27.5 6.7 3.4 10 45-49 16.7 18.8 8.8 11.6 0.0 18.2 20.0 36.8 8.3 5.7 11 50-54 20.9 18.3 10.7 7.1 0.0 0.0 33.3 30.0 12.0 4.9 12 55-59 20.2 18.3 14.4 7.8 13.3 0.0 12.1 28.6 2.0 4.1 13 60-64 20.0 23.7 11.3 9.9 10.0 9.5 25.5 23.2 6.7 8.2 14 65-69 20.6 24.5 14.4 9.8 12.5 8.3 22.4 29.3 9.0 8.3 15 70-74For 26.9 peer 23.8 19.5 review 14.9 10.0 17.9 only 25.0 25.4 13.1 11.0 16 75+ 27.8 16.9 22.2 18.2 11.1 13.3 39.1 27.3 10.7 10.5 17 Rhinitis 19-44 31.8 41.2 25.2 20.2 26.7 24.4 41.9 54.9 17.4 19.3 18 45-49 45.2 43.6 34.8 51.1 24.5 30.0 33.3 42.1 13.5 13.0 19 50-54 40.7 38.2 20.6 30.4 50.0 27.3 51.9 56.7 20.0 20.3 20 55-59 45.0 33.5 21.3 18.4 22.2 23.1 51.5 34.8 19.4 16.1 21 60-64 43.0 37.2 25.6 20.9 20.0 38.5 41.2 35.4 20.6 11.2 22 65-69 33.9 29.9 28.7 23.2 30.0 19.0 36.7 37.9 18.5 16.3 23 70-74 41.4 34.1 24.0 17.3 37.5 20.8 41.7 30.6 26.4 15.5 24 75+ 49.5 25.4 24.5 20.0 25.0 25.0 43.5 30.3 17.4 17.0 25 Dermatiti 19-44 27.3 15.3 6.5 11.1 9.4 0.0 9.7 11.8 4.4 6.9 26 s 45-49 7.1 27.7 2.9 10.1 0.0 0.0 6.7 16.2 0.0 5.7 27 50-54 16.5 23.0 5.3 2.4 0.0 0.0 22.2 26.7 2.7 4.9 28 55-59 21.7 23.9 4.4 6.5 6.7 0.0 6.1 12.9 4.0 7.6 29 60-64 14.5 30.2 4.0 6.2 5.0 9.5 11.8 12.2 0.0 4.7 30 65-69 15.2 26.4 4.3 5.1 12.5 0.0 10.2 8.6 1.9 2.2 31 70-74 30.3 27.4 4.5 7.3 5.0 7.1 6.3 19.0 2.8 2.9 32 75+ 21.5 18.5 9.4 4.7 0.0 0.0 8.7 12.1 4.8 3.5 33 Conjuncti 19-44 29.5 36.5 21.7 20.0 13.2 10.0 6.5 23.5 4.4 8.6 34 vitis 45-49 31.0 40.6 11.8 10.1 0.0 27.3 20.0 15.8 10.8 8.7 35 50-54 37.4 44.5 6.6 10.5 0.0 0.0 18.5 33.3 9.1 10.6 http://bmjopen.bmj.com/ 36 55-59 28.7 37.2 10.8 12.7 33.3 0.0 25.0 31.4 11.4 9.0 37 60-64 33.9 38.6 15.2 13.3 5.0 4.8 21.6 17.3 7.3 16.0 38 65-69 27.3 41.0 12.6 14.2 12.5 37.5 10.2 19.0 9.4 8.4 39 70-74 35.2 38.1 11.5 12.9 10.0 25.0 22.9 30.6 13.2 9.8 40 75+ 32.7 29.2 13.2 9.0 0.0 6.7 8.7 21.2 4.7 7.1 41 PTSD 19-44 40.0 25.6 3.9 12.0 0.0 0.0 9.5 10.1 2.1 7.6 42 45-49 42.9 44.1 12.5 16.2 0.0 25.0 12.5 14.0 17.5 12.3

43 on October 1, 2021 by guest. Protected copyright. 50-54 31.5 38.2 16.5 10.5 0.0 0.0 15.0 18.1 12.2 5.3 44 55-59 37.1 35.3 7.1 10.4 12.5 0.0 4.8 8.9 4.5 6.3 45 60-64 36.7 38.4 10.3 7.7 4.5 0.0 7.5 12.2 5.7 0.6 46 47 65-69 35.9 35.5 7.1 7.7 0.0 0.0 8.5 14.9 4.1 2.5 48 70-74 30.4 36.6 4.7 5.6 8.7 6.3 17.5 15.1 1.9 1.8 49 75+ 18.4 20.5 3.3 2.3 0.0 0.0 12.5 15.6 0.0 0.0 Depressio 19-44 29.5 27.9 6.5 15.6 7.4 20.0 12.9 28.6 6.7 25.9 50 n 51 45-49 42.9 33.0 20.0 21.7 0.0 25.0 21.4 21.1 8.1 20.0 52 50-54 23.3 29.3 14.7 15.0 0.0 7.7 23.1 27.1 7.8 14.6 53 55-59 24.8 26.7 8.8 10.4 20.0 7.7 15.6 14.3 7.8 11.2 54 60-64 24.8 29.8 9.3 11.8 10.0 9.5 18.0 22.5 4.9 8.2 55 65-69 19.4 30.3 5.6 18.6 12.5 20.8 26.5 25.9 3.8 8.7 56 70-74 22.8 29.6 6.3 14.4 10.0 14.3 27.7 30.6 4.8 10.0 57 75+ 15.5 35.2 12.4 13.7 0.0 6.7 21.7 39.4 2.4 9.6 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 32 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 For young people under 19 years old, we used results from 925 school students (male: 505; 5 6 female: 420) living in the study area. Prevalence of asthma for elementary school student was 7 8 18.2% and 7.6% for highly contaminated area and less contaminated area, respectively. In 9 10 11 particular, the Taean Environmental Health Center (TEHC) surveyed the incidence of asthma 12 13 newly occurred after the oil spill. That is, the asthma incidence for 2008 was 6.7% and 1.0% 14 15 for highly contaminatedFor areapeer and less contaminated review area, respectively. only Therefore we used the 16 17 incidence (6.7%) in highly contaminated area for Area 1 and Boryeong island and that (1.0%) 18 19 20 in less contaminated area for Area 2 and Area 3. For 13~18 years old, new incidence of 21 22 asthma for 2008 was 5.1%. We used the asthma incidence for elementary school student for 23 24 0-9 age group and the incidence for 13-18 years old for 10-19 age group. 25 26 Prevalence of rhinitis was 38.3% and 32.4% for elementary school student in highly 27 28 contaminated area and less contaminated area, respectively, and 34.7% for the 13-18 years 29 30 31 old. There was no difference in prevalence between highly contaminated area and less 32 33 contaminated area. We, therefore, used only the prevalence of elementary school student for 34

35 0-9 age group. http://bmjopen.bmj.com/ 36 37 Prevalence of dermatitis was 19.5% and 16.4% for elementary school student in highly 38 39 40 contaminated area and less contaminated area, respectively, and 8.4% and 5.8% for the 13-18 41 42 years old in highly contaminated area and less contaminated area, respectively. We used the

43 on October 1, 2021 by guest. Protected copyright. 44 prevalence difference in dermatitis for elementary school student for 0-9 age group and that 45 46 for 13-18 years old for 10-19 age group to calculate the incidence in 2008. 47 48 Prevalence of conjunctivitis was 12.4% and 14.7% for elementary school student in highly 49 50 51 contaminated area and less contaminated area, respectively, and 18.7% and 15.2% for the 13- 52 53 18 years old in highly contaminated area and less contaminated area, respectively. 54 55 Prevalence of post-traumatic stress disorder (PTSD) was 5.1% and 1.8% for elementary 56 57 school student in highly contaminated area and less contaminated area, respectively, and 4.7% 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 33 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 and 2.1% for the 13-18 years old in highly contaminated area and less contaminated area, 5 6 respectively. Considering that children under 10 years old did not participate in clean-up 7 8 work which is crucial factor to cause the PTSD according to the TEHC report, we assumed 9 10 11 no new incidence of PTSD in 0-9 age group and used the averaged prevalence of elementary 12 13 school student and 13-18 years old as the incidence of 10-19 age group. 14 15 Prevalence ofFor mild depression peer was 8.3% review and 2.9% for the only elementary school student in 16 17 highly contaminated area and less contaminated area, respectively, and 5.9% for the 13-18 18 19 20 years old in the whole area. We used the difference in prevalence of the elementary school 21 22 student as the incidence of 10-19 age group. 23 24 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 34 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 Table S3. Excess incidence of disease due to Hebei Spirit oil spill by age, sex, and area (%). 5 Disease Age Boryeong 6 Area 1 Area 2 Area 3 7 group islands 8 Male Female Male Female Male Female Male Female 9 Asthma 0~9 6.7 6.7 1.0 1.0 1.0 1.0 6.7 6.7 10 10~19 5.1 5.1 1.0 1.0 1.0 1.0 5.1 5.1 11 20~49 9.9 8.9 2.3 6.8 1.4 9.5 10.6 27.6 12 50~59 13.5 13.8 6.2 3.0 5.7 0.0 15.7 24.8 13 60~69 12.4 15.9 5.0 1.6 3.4 0.7 16.1 18.0 14 70+ 15.4 9.6 9.0 5.8 0.2 4.9 20.2 15.6 15 Rhinitis For0~9 peer5.9 5.9review 0.0 0.0 only0.0 0.0 5.9 5.9 16 10~19 ------17 20~49 23.1 26.2 14.5 19.5 10.2 11.1 22.2 32.3 18 50~59 23.1 17.7 1.3 6.2 16.4 7.0 32.0 27.5 19 60~69 18.9 19.8 7.6 8.3 5.8 15.0 19.4 22.9 20 70+ 23.6 13.5 3.6 2.4 9.4 6.7 20.7 14.3 21 Dermatitis 0~9 3.1 3.1 - - - - 3.1 3.1 22 10~19 2.6 2.6 - - - - 2.6 2.6 23 20~49 15.0 15.2 3.6 2.7 5.9 0.0 5.9 7.7 24 50~59 15.8 17.2 0.8 2.6 0.0 0.0 10.8 13.5 25 60~69 13.9 24.9 3.3 2.9 4.9 3.7 10.0 7.0 26 70+ 22.2 19.7 0.8 3.0 5.0 1.8 3.7 12.4 27 Conjunctivitis 0~9 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 28 10~19 3.5 3.5 0.0 0.0 0.0 0.0 3.5 3.5 29 30 20~49 22.6 29.9 9.5 7.7 4.8 4.5 5.6 11.0 31 50~59 22.8 31.0 1.3 0.7 0.0 8.3 11.5 22.5 32 60~69 22.3 27.6 4.6 2.5 13 0.0 7.5 5.9 33 70+ 25.0 25.2 3.4 5.1 2.7 22.8 6.9 17.5 34 PTSDa 0~9 ------

35 10~19 4.9 2.0 4.9 2.0 0.0 0.0 4.9 2.0 http://bmjopen.bmj.com/ 36 20~49 31.6 24.9 0.9 4.2 0.0 0.0 3.7 2.1 37 50~59 26.0 31.0 3.4 4.7 0.0 0.0 1.5 7.7 38 60~69 31.4 35.4 3.7 6.2 0.0 0.0 3.0 12.0 39 70+ 23.4 27.6 3.0 3.1 0.0 0.0 14.0 14.5 40 Depression 0~9 ------41 10~19 5.4 5.4 0.0 0.0 0.0 0.0 5.4 5.4 42 20~49 28.8 7.5 5.9 0.9 0.4 2.5 9.8 1.9

43 50~59 16.3 15.1 3.9 0.2 6.1 0.0 11.5 7.8 on October 1, 2021 by guest. Protected copyright. 44 60~69 17.8 21.6 3.1 6.8 6.9 6.7 17.9 15.7 45 70+ 15.5 22.6 5.7 4.2 2.6 2.2 21.1 25.2 46 a: post-traumatic stress disorder 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

- -

60.6 40.2 58.6 59.6 120.6 107.7 129.1 576.4 Female Female

- 94.9 33.3 Male 126.9 339.1 269.8 266.9 154.5 1,861.9 1285.4

- Mental Mental disease Depression a

b - 65.0 27.4 65.9 101.5 Male Female 328.8 316.8 197.4 321.5 160.0 238.3 234.1 197.7 335.4 279.4 1534.0 1335.2

0.1 1.4 1.2 1.8 2.2 1.1 1.6 9.4 0.0 Female

BMJ Open

http://bmjopen.bmj.com/

attributable to Hebei Spirit oil spill by disease, sex, age DALY) and (Unit: spill sex, disease, oil by Spirit to Hebei attributable

0.0 0.0 23.9 22.5 0.1 on October 1, 2021 by guest. Protected copyright. 72.5 110.6 60.8 109.5 0.7 16.2 18.5 19.9 17.5 0.0 Male Female Male Female Male 213.9 210.5 231.3 225.9 196.7 201.7 62.5 273.4 124.0 164.1 198.2 36.3 176.2 286.6 63.1 133.8 127.2 103.2 1.8 178.1 1.8 158.5 1.8 1.3 1.2 1104.8 1104.8 1315.3 704.3 712.7 8.7 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

Allergy Rhinitis Dermatitis Conjunctivitis PTSD

For peer review only 88.2 137.3 724.6 599.9 738.1 589.4 323.3 149.2 3350.0 Female

Asthma Male Years lived with disability (YLDs) with disability lived Years

0~9 0~9 101.3 70+ 70+ 117.0 Age Sum 2787.7 a: post-traumatic stress disorder; b: Data available not b: Data stress a: disorder; post-traumatic Total Total 6,137.7 2,420.0 1,416.9 18.2 2,869.2 10~19 144.8 30~39 40~49 505.6 50~59 469.4 602.9 20~29 540.8 60~69 305.9 Table S4. Table Page 35 of 60 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open Page 36 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 Burden of disease attributable to the Hebei Spirit oil spill in FormattedFormattedFormatted 9 10 2 Taean, Korea 11

12 1 2 2 3 3 13 3 Young-Min Kim , Jae-Hyun Park , Kyusik Choi , Su Ryeon Noh , Young-Hyun Choi , Hae-

14 2 15 4 For peerKwan Cheong review only 16 1 17 5 Department of Environmental Health, Rollins School of Public Health, Emory University, 18 19 6 1518 Clifton Road NE, Atlanta, GA, USA

20 2 21 7 Department of Social and Preventive Medicine, Sungkyunkwan University School of 22 Medicine, 2066 Seobu-ro, Jangan-gu, Suwon, 440-746, Republic of Korea 23 8 24 3Taean Environmental Health Center, 1952-16 Seohae-ro, Taean, Chungnam, 357-902, Re 25 9 26 10 public of Korea 27 28 11 29 30 12 * Corresponding author: Hae-Kwan Cheong 31 32 13 2066 Seobu-ro, Jangan-gu, Suwon, Gyeonggi-do, 440-746, Korea 33 34 14 Tel: +82-31-299-6300, Fax: +82-031-299-6299

35 http://bmjopen.bmj.com/ 36 15 E-mail: [email protected] 37 38 16 39 40 17 * Keywords: oil spill, years lived with disability (YLD), disability-adjusted life-year (DALY), 41 42 18 polycyclic aromatic hydrocarbons (PAHs), post-traumatic stress disorder (PTSD), disability

43 on October 1, 2021 by guest. Protected copyright. 44 19 weight 45 46 20 47 48 21 Word count: Formatted:Formatted:Formatted: Font: Bold 49 22 Abstract: 29757 words 50 23 Main text: 33522915 words 51 24 No. of tables: 3 52 25 No. of Figures: 2 53 1 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 37 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 No. of references: 279 9 2 ABSTRACT 10 11 3 Objectives: We aimed at assessing the burden of disease (BOD) of the residents of theliving 12 in contaminated coastal area with oil spill and also analyzed the BOD and at analyzing the 13 4 14 BOD attributable to thethe oil spill by disease, age, sexgroup, and sub-region. 15 5 For peer review only 16 6 Design: HHealth impact assessment study by measuring years lived with disability (YLD) 17 18 7 due to an oil spillof a specific environmental exposure from oil spill. 19 20 21 8 22 23 9 Setting: Use of secondary health outcome data. AA whole population of a community 24 25 10 affected by an anthropogenic environmental disaster and secondary health outcome data. 26 27 11 Participants: For the health outcome survey, 10,171 individuals (male: 4,354; female: 5,817) Formatted:Formatted:Formatted: Font: Not Bold 28 Formatted:Formatted:Formatted: Font: Not Bold 12 participated, covering 66,473 populations (male: 33,441; female: 33,032) living in the 29 Formatted:Formatted:Formatted: Font: Not Bold 30 Formatted:Formatted:Formatted: Font: Not Bold 13 contaminated coastal area.63,053 individuals (male: 31,636; female: 31,417) and 3,420 31 Formatted:Formatted:Formatted: Font: Not Bold 32 individuals (male: 1,805; female: 1,615) from two counties, respectively Formatted:Formatted:Formatted: Font: Not Bold 33 14 Formatted:Formatted:Formatted: Font: Not Bold 34 15 Interventions: None. Observational study on the effect of a specific environmental health Formatted:Formatted:Formatted: Font: Not Bold 35 http://bmjopen.bmj.com/ 36 16 hazard.. 37 38 17 Primary and secondary outcome measures: Using YLD method, BOD including both 39 40 18 physical and mental diseases was measured using disability-adjusted life-year (DALY) 41 42 19 method. For the YLD measurement, excess based on the incidences of illnesses related to oil

43 on October 1, 2021 by guest. Protected copyright. 44 20 spill were estimated from and the comparison of prevalence of the health outcomes between 45 46 21 contaminated areas and reference area without contamination. related to the oil spill. Formatted:Formatted:Formatted: Font: Bold 47 48 22 Results: Years lost due to disabilities (YLD) attributable to the oil spill were estimated to be 49 50 23 14,724 DALYs (male: 7,425 DALYs; female: 7,299 DALYs) for the year 2008. The YLD of 51 52 24 mental diseases including post-traumatic stress disorder (PTSD) and depression for males for 53 2 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 38 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 mental diseases were higher than that for females., Tand the YLD for females was higher in 9 asthma and allergies (rhinitis, dermatitis, conjunctivitis) than that for males. The effects of 10 2 11 3 asthma and allergies were the greatest for people in their 40s, with the burden of mental 12 13 4 illness being the greatest for those in their 20s. Proximity to the spill site was associated with 14 15 5 increased burden of disease.For peer review only 16 17 6 Conclusions: An oil spill near a coastline can cause substantial adverse health effects. As the 18 19 7 health effects of hazardous pollutants from oil spills are long-lasting, close follow-up studies 20 21 8 are required to identify chronic health effects. 22 23 9 Trial registration: Not applicable FormattedFormattedFormatted 24 25 26 10 27 28 11 INTRODUCTION 29 30 12 On December 7, 2007, the Hebei Spirit oil tanker spilled 12,547 kl (10,900 tons) of crude 31 32 13 oil into the Yellow Sea, approximately 8 km away from the coast of Taean, Korea. This was 33 34 14 the largest oil spill in the history of Korea. Spilled oil contaminated over 1,000 km of the

35 http://bmjopen.bmj.com/ 36 15 western coast of Korea including the Taean Coast National Park.[1] Crude oil contains 37 38 16 various volatile organic compounds (VOCs) such as benzene, toluene, ethylbenzene, xylenes, 39 40 17 and polycyclic aromatic hydrocarbons (PAHs),[2] all of which are potential air-borne toxins. 41 42 18 Certain VOCs, benzene in particular, are carcinogenic to humans and are associated with

43 on October 1, 2021 by guest. Protected copyright. 44 19 haematologic cancer.[3] Previous studies [4-11] have reported that oil spills have caused 45 adverse health effects for residents of contaminated areas and clean-up workers. Risk 46 20 47 analyses revealed that more frequent clean-ups and greater exposure to oil are strongly 48 21 49 22 associated with higher frequencies of psychological symptoms. [8-11] Rodriguez-Trigo [6] 50 51 23 reported that participation in a major oil spill clean-up is associated with chromosomal 52 53 3 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 39 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 damage, although they were unable to determine whether oil exposure itself caused the 9 abnormalities. 10 2 11 3 After the Hebei Spirit oil spill, more than 2,000,000 people (person-days), including over 12 13 4 1,200,000 person-days by volunteers, participated in clean-up activities during the first seven 14 15 5 months.[1] Among themFor were residents peer of Taean, a highlyreview exposed group that onlyexperienced 16 17 6 adverse health effects, including both physical and psychological symptoms.[4, 12] Although 18 19 7 many studies have reported positive relationships between oil spills and acute and chronic 20 21 8 health impacts, there have been no studies quantifying the burden of disease (BOD) due to oil 22 23 9 spills, which are necessary to assess the scale of health damage at the population level as well 24 25 10 as the associated compensation costs. 26 27 11 Therefore, we sought to quantify the adverse health impacts of exposure to oil spills by 28 29 12 applying the disability-adjusted life-year (DALY) measure developed by the World Health 30 31 13 Organization (WHO). The DALY measures health gaps as opposed to health expectancies. It 32 33 14 measures the difference between a current situation and an ideal situation where everyone 34 15 lives up to the age of the standard life expectancy, and in perfect health. The DALY is based 35 http://bmjopen.bmj.com/ 36 on the premise that the best approach for measuring the burden of disease is to use units of 37 16 38 17 time. [13] The DALY method has been used to measure the BOD attributable to specific risk 39 40 18 factors including environmental burdens of disease (EBD). This approach can beis practical 41 42 19 in thatbecause the estimated adverse health burden attributable to a risk factorn oil spill can

43 on October 1, 2021 by guest. Protected copyright. 44 20 be compared with other BODs, including environmental burdens of disease (EBD). and Thus, 45 46 21 the analyses of disease burdens by disease and region can inform policy prioritization to 47 48 22 mitigate adverse health effects for specific exposed populations. 49 50 51 52 53 4 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 40 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 The purposes of this study are toWe aimed at assessing the BOD of the residents of the 9 contaminated Taean coastal area due to the Hebei Spirit oil spill and toat analyzeing the BOD 10 2 11 3 attributable to the oil spill by disease, age, sexgroup, and sub-region. 12 13 4 14 15 For peer review only 16 5 METHODS 17 18 6 Study area and design 19 20 7 In September 2008, the Taean Environmental Health Center (TEHC) was established 21 22 8 under the support of the Korean Ministry of Environment to monitor the effects of the Hebei 23 24 9 Spirit oil spill on the health of the residents of the Taean coastal area. The TEHC first 25 conducted health surveys on the residents, including the referent population, one and a half 26 10 27 years after the spill, establishing a cohort. Follow-up surveys were conducted every two years 28 11 29 12 on the 1,900 individuals highly exposed to oil and its remnants, including residents in the 30 Formatted:Formatted:Formatted: Highlight 31 13 contaminated coastal area, public services persons who worked on the clean-up, and 32 33 14 susceptible populations, including school children and pregnant women. Total of 10,171 34

35 15 residents living in the study area were surveyed including 9,246 adults (male: 3,849; female: http://bmjopen.bmj.com/ 36 37 16 5,397) and 925 school students (male: 505; female: 420 persons). The survey included a 38 39 17 comprehensive questionnaire gathering personal history on exposure and medical problems, 40 41 18 and a structured questionnaire assessing psychological health and asthma, and physical and 42

43 19 laboratory examinations of respiratory, cardiovascular, neurologic, and psychological systems. on October 1, 2021 by guest. Protected copyright. 44 45 20 The TEHC published its first report in 2010 after analysing the prolonged health effects 46 47 21 for 1.5 years, and demonstrated significant relationships between exposure concentrations 48 49 22 and adverse health outcomes.[154] To estimate the BOD caused by the Hebei Spirit oil spill, 50 51 23 we obtained the survey data from the TEHC, which included disease, residence, and 52 53 5 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 41 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 demographic information. Residents in Taean County and some islands in neighboring 9 Boryeong County were included in the survey. The Boryeong islands were expected to be 10 2 11 3 affected by the oil spill because the islands are located in the affected area based on the wind 12 13 4 direction and tidal current at the time of the spill.[165] Kim et al.[165] reported that the 14 15 5 VOCs moved in a southeasterlyFor direction, peer following the review coastline from the accident only point. 16 17 6 We classified the residents of Taean County and the Boryeong islands into five regions 18 19 7 based on both distance from the contaminated coastline and contamination level. Area 1 is 20 21 8 less than 1.1 km away from the coastline adjacent to the accident location where the coastline 22 23 9 was contaminated at a high level. Area 2 is 1.1-4.2 km away with no coastline, but was 24 25 10 directly affected by atmospheric VOCs and PAHs due to the wind direction. Area 3 is 4.2-23 26 27 11 km away from the coastline with a moderately contaminated coastline, and Area 4 is more 28 29 12 than 23 km from the contaminated coastline. Area 4 was assumed to be the reference area, 30 31 13 not only because the area might not be impacted by the oil spill directly as it is far from the 32 33 14 contaminated coastline, but also because residents in Area 4 rarely participate in the clean-up 34 15 efforts. We classified the Boryeong Islands as a separate area. Figure 1 shows the locations of 35 http://bmjopen.bmj.com/ 36 the study areas and five classified regions. 37 16 38 17 Figure 1 39 40 18 41 42 19 We assessed health outcome data regarding asthma, allergy (rhinitis, dermatitis, Formatted:Formatted:Formatted: Not Highlight

43 on October 1, 2021 by guest. Protected copyright. 44 20 conjunctivitis), and mental disorders (post-traumatic stress disorder (PTSD) and depression), 45 46 21 conditions that were well-documented by the TEHC.[15] We focused on subchronic (duration Formatted:Formatted:Formatted: Not Highlight 47 48 22 of a few months) and chronic (over one year duration) adverse health effects of the Hebei 49 50 23 Spirit oil spill, assessing prolonged health effects over a period of 1.5 years to estimate the 51 52 24 BOD of the Hebei Spirit oil spill for the year of 2008. 53 6 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 42 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 9 Estimation of burden of disease attributable to oil spill 10 2 11 3 To estimate the BOD attributable to the oil spill, we applied the DALY method. DALY is 12 13 4 the sum of the life years lost due to disability and premature death. [13] As there was no 14 15 5 premature death identifiedFor as the direct peer effect of the oilreview spill until the TEHC publishedonly the 16 17 6 first report in 2010, we calculatincluded only years lived withlost due to disability (YLD). [13] 18 19 7 The YLD is calculated by using disease incidence and prevalence, onset age and duration of 20 21 8 disability, and disability weight (DW). The formula we utilized is as follows; 22 23 9 24 25 Formatted:Formatted:Formatted: Centered 1 1 26 Formatted:Formatted:Formatted: Font: Times New Roman 27 28 10 1 [1] 29 30 11 where, D is disability weight; r is discount rate; a is age of onset of the disability; L is 31 32 Formatted:Formatted:Formatted: Font: Italic 33 12 duration of the disability; β is parameter from the age weighting function; and C is constant. 34

35 http://bmjopen.bmj.com/ We used 3% of discount rate and 0.04 and 0.0618 for and C, respectively, as suggested by 36 13 β 37 38 14 the WHO.[13] 39 40 15 Because there was no prevalence data recorded prior to the accident, we estimated excess 41 42 16 incidences using prevalence data from the TEHC under the following assumptions. We first

43 on October 1, 2021 by guest. Protected copyright. 44 17 assumed that residents living in Area 4 had no negative health consequences due to the oil 45 46 18 spill and used this as the baseline prevalence. Area 4 was assumed to be the reference area, 47 48 19 not only because the area might not be impacted by the oil spill directly as it is far from the 49 50 20 contaminated coastline, but also because residents in Area 4 rarely participate in the clean-up 51 52 21 works. Very few people participated the clean-up work except for civil servants living in the 53 7 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 43 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 area 4, which appeared to be an important factor to cause the difference in the prevalence of 9 disease between area 4 and others.[15] Secondly, we assumed that the prevalence of oil spill- 10 2 11 3 related diseases in the whole study area had been same prior to the accident. We then 12 13 4 estimated excess incidences using prevalence differences between Area 4 and the other areas 14 15 5 stratified by area, age, diseaseFor, and sexpeer, and disease using review following formula. only 16 17 6 ∑ [2] 18 19 7 here, is the excess incidence of region i and disease j; is the population of 20 21 8 region i; is the prevalence of region i and disease j; is the prevalence of control 22 23 9 region and disease j. 24 25 10 The duration of disability and the average onset age were calculated using the DISMOD 26 27 11 II model [176] for PTSD, rhinitis, dermatitis, and conjunctivitis based on prevalence data 28 29 12 from the TEHC and the mortality rate of South Korea [187] by age, sex, and disease. For 30 31 13 asthma, the results of the Victoria, Australia study [198] were used. 32 33 14 DWs for each disease were measured following a method previously used by the Global 34

35 15 Burden of Disease group [13] and the Dutch Disability Weights Group.[2019] A panel that http://bmjopen.bmj.com/ 36 37 16 included 12 health and medicine specialists was convened. Eleven indicator conditions 38 39 17 including asthma, stomach cancer, diabetes, dementia, major depression, appendicitis, thyroid 40 41 18 cancer, hepatitis A, otitis media, atopic dermatitis, and common cold were measured using the 42 19 person trade-off (PTO) method. The eleven indicator conditions were selected in accordance

43 on October 1, 2021 by guest. Protected copyright. 44 45 20 with a previous study on Korean disability weight. [21] That is, we considered that they 46 should be meaningful in public health and relatively popular diseases, while distributing the 47 21 48 whole DW range (0 to 1) evenly.[21] A facilitator first explained the method and then led the 49 22 50 23 discussion to determine the PTO value of each participant. The discussion was repeated twice. 51 52 53 8 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 44 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 Then, the visual analogue scale (VAS) method was applied to obtain DWs for 118 conditions 9 including asthma, rhinitis, dermatitis, conjunctivitis, PTSD, and depression used the target 10 2 11 3 conditions ofin this study. Inter-method reliability was confirmed between PTO and VAS 12 13 4 using 11 indicator conditions, which showed fairly high consistency (Spearman’s correlation 14 15 5 coefficient 0.966, P-value<0.001).For Reliabilitypeer was confirmed review by test-retest after onlythree weeks 16 17 6 by seven specialists and obtained fairly high consistency (Spearman’s correlation coefficient 18 19 7 0.973, P-value<0.001). 20 21 8 Finally, we estimated YLD caused by the Hebei Spirit oil spill utilizing the estimated 22 23 9 excess incidence attributable to the oil spill, DW, and the onset age and duration of disability 24 25 10 by disease, age, sex, and area. 26 27 11 28 29 12 RESULTS 30 31 13 32 33 14 Summary of the characteristics of the study area 34

35 15 The populations of Taean County and the Boryeong islands were 63,053 individuals http://bmjopen.bmj.com/ 36 37 16 (male: 31,636; female: 31,417) and 3,420 individuals (male: 1,805; female: 1,615) in 2008, 38 39 17 respectively. Broken down by area, 8,348 (13%), 11,272 (17%), 4,597 (7%), 38,836 (58%), 40 41 18 and 3,420 (5%) residents lived in Areas 1, 2, 3, 4, and the Boryeong islands, respectively. 42

43 19 Area 4, the reference area, was the most populous because it includes a city center. The on October 1, 2021 by guest. Protected copyright. 44 45 20 largest age group consisted of subjects in their 50s, with individuals in their 40s, 50s, and 60s 46 47 21 comprising 45% of the total population (data presented in the supplement Table S1). 48 49 22 A total of 10,171 residents living in the study area participated in the TEHC survey. Formatted:Formatted:Formatted: Not Highlight 50 51 23 Among them, 4,354 (42.8%) and 5,817 (57.2%) residents were male and female, respectively. 52 53 9 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 45 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 Individuals under 15 years of age, 15-65, and over 65 years of age comprised 7.2%, 48.0%, 9 and 44.8% of the population, respectively. Residents of Area 1 participated in clean-up work 10 2 11 3 for an average of 122 days, and residents of Areas 2, 3, and the Boryeong islands participated 12 13 4 for averages of 50, 36, and 100 days, respectively.[14] Table 1 shows the prevalence of 14 15 5 disease related to the currentFor study bypeer disease, sex, and review area derived from a survey only of 9,246 16 17 6 adult (over 19 years old) residents. For young people under 19 years old, we used results 18 19 7 from 925 school students (male: 505; female: 420) living in the study area. 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 10 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from Page 46 of 60 42.6 10.1 40.2 17.0 15.0 24.5 10.8 21.3 13.5 25.2 23.8 27.9 Male Female Boryeong Islands

Area 4 Area 2.4 2.4 9.2 4.4 10.0 4.6 5.3 3.2 11.6 8.7 8.7 7.6 20.2 15.9 Male Female

BMJ Open

3.1 http://bmjopen.bmj.com/ 11 11 Area Area 3

9.4 4.7 2.9 8.8 14.0 9.6 10.4 23.9 10.6 13.2 9.0 Male Female

7.1 on October 1, 2021 by guest. Protected copyright. Area 2 Area

5.1 6.1 7.0 8.8 14.8 25.2 20.2 12.6 12.6 15.4 11.7 Male Female For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

For peer review only a) Area Area 1 22.0 20.4 32.9 34.6 Male Male Female . Prevalence of diseases related to this study (in %). thisstudy related to diseases Prevalence of . b) Dermatitis Conjunctivitis 19.6 32.0 25.0 38.7 Rhinitis 41.4 34.6 Depression 23.3 29.9 PTSD Mental diseaseMental Table 1 Table a) Area 1: less than 1.1 km away; Area 2: 1.1-4.2 km away; Area 3: 4.2-23 km away; Area 4: more than 23 km from the from contaminated km 23Area than 4: more away; km Area 3: 4.2-23 away; km Area2: 1.1-4.2 away; 1.1 km Areathan 1: less a) disorder. stress PTSD: posttraumatic b)coastline; Allergy Allergy Asthma Disease 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 47 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 Estimated burden of disease due to oil spill FormattedFormattedFormatted 9 Excess incidence attributable to the oil spill 10 2 11 3 The excess incidence of disease caused by the oil spill varied by area, sex, and disease 12 13 4 (Table 2). There were additional incidences of asthma (2,088: 989 male; 1,099 female), 14 15 5 rhinitis (3,625: 1,855 For male; 1,770 peer female), dermatitis review (1,976: 962 male; 1,014 only female), 16 17 6 conjunctivitis (2,992: 1,365 male; 1,627 female), PTSD (2,681:1,300 male; 1,381 female), 18 19 7 and depression (2,326: 1,370 male; 956 female) in Areas 1,2,3 and the Boryeong islands. 20 21 8 Area 1 had the highest additional incidences for all diseases, while Area 3, a relatively 22 23 9 less affected area than Areas 1 and 2, showed the lowest additional incidences for all diseases. 24 25 10 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 12 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from Page 48 of 60

Total Total

989 989 1,099 962 962 1,014 Male Male Female 1,300 1,381 1,370 956 1,855 1,770 1,365 1,627

76 76 106 201 201 149 222 222 322 353 353 354 113 113 190 116 116 138 Male Female Boryeong Boryeong Islands

BMJ Open

0 0 http://bmjopen.bmj.com/ Area 3 Area 13 c c

0 65 65 57 53 53 97 94 94 32 209 209 207 101 101 183 Male Female

Area Area 2

on October 1, 2021 by guest. Protected copyright. 266 266 169 133 133 222 241 241 231 453 453 481 294 294 214 146 146 158 Male Female

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

a For peer review only Area 1 Area

839 839 729 473 473 449 838 838 581 Male Female 1,092 1,054 Excess incidence caused by the oil spill (Unit: number of persons) persons) theof (Unit: oil spill number by caused Excess incidence Rhinitis Conjunctivitis 857 1,040 Dermatitis 606 687 b b

Depression Table 2. Table a: Area 1, less than 1.1 km away; Area 2, 1.1-4.2 km away; Area 3, 4.2-23 km away; Area 4, more than 23 km from the contaminated coastline; coastline; fromcontaminated the 23km Area than more away; 4, km Areaaway; 3, 4.2-23 Areakm away; 2, 1.1-4.2 less km 1.1 than Area 1, a: and area. reference Area 3 between difference No c: stress disorder; b: post-traumatic Allergy Allergy Asthma Disease PTSD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 49 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 Disability weight FormattedFormattedFormatted 9 The VAS results showed that the DW for stomach cancer (stage 4) was the highest at 10 2 11 3 0.909, and that the DW for upper respiratory infection was the lowest at 0.065, out of a total 12 13 4 of 118 conditions surveyed. Among the 118 conditions, DWs for diseases examined in this 14 15 5 study are shown in TableFor 3. We used peerthe DW of mild intermittentreview asthma, 0.222, onlyto represent 16 17 6 asthma, because the prevalence of asthma in this study area was estimated using survey 18 19 7 results following the modified International Study of Asthma and Allergies in Childhood 20 21 8 (ISSAC), and most cases of asthma were not severe.[154] 22 23 9 24 25 10 Table 3. Disability weights used in the current study 26 27 Disability Disease Stage 28 weight 29 Asthma Mild intermittent 0.222 30 Mild persistent 0.291 31 Moderate persistent 0.510 32 Severe persistent 0.715 33 Very severe persistent 0.836 34 Allergy Rhinitis 0.205 35 http://bmjopen.bmj.com/ 36 Dermatitis 0.132 37 Conjunctivitis 0.212 38 PTSDa 0.533 39 40 Depression 0.219 41 11 a: post-traumatic stress disorder 42

43 12 on October 1, 2021 by guest. Protected copyright. 44 45 13 Burden of disease due to oil spill 46 47 14 Figure 2 presents the YLD according to disease, sex, and age group. The YLD for 48 49 15 asthma, allergic rhinitis, dermatitis, conjunctivitis, PTSD, and depression were approximately 50 51 16 6,138 DALYs, 2,420 DALYs, 1,417 DALYs, 18 DALYs, 2,869 DALYs, and 1,862 DALYs, 52 53 14 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 50 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 respectively. The total YLD attributable to the Hebei Spirit oil spill for the year 2008 was 9 14,724 DALYs. The YLD for males was 7,425 DALYs and for females was 7,290 DALYs. 10 2 11 3 The YLD for males for PTSD and depression were higher than for females, and the YLD for 12 13 4 females was higher in asthma and allergies than for males (More details are shown in data 14 15 5 supplement Table S2). ForAmong the includedpeer diseases, thereview YLD for asthma was only the highest. 16 17 6 The YLD for asthma and allergies was the highest for residents in their 40s, and the YLD for 18 19 7 mental disease was the highest for residents in their 20s. 20 21 8 Figure 2 22 23 9 24 25 10 DISCUSSION 26 27

28 11 29 12 The total YLD attributable to the Hebei Spirit oil spill was estimated at 14,724 DALYs 30 31 13 for 2008. Asthma was found to represent the most prominent disease burden (6,138 DALYs) 32 33 14 among the sixeven diseases, followed by PTSD and rhinitis. The asthma burden of 6,138 34

35 15 DALYs in the study area can be translated to 9,233 DALYs per 100,000 individuals, or 6.5 http://bmjopen.bmj.com/ 36 37 16 times higher in contrast to the total burden of asthma for South Korea of 1,418 DALYs per 38 39 17 100,000 individuals.[220] 40 41 18 Kim et al.[231] estimated the prevalence and economic costs of allergic rhinitis in South Formatted:Formatted:Formatted: Indent: First line: 1.5 ch 42

43 19 Korea for 2007 and reported that a total of 4,068,517 people were identified as having on October 1, 2021 by guest. Protected copyright. 44 45 20 allergic rhinitis. The direct cost was $223.68 million and the lost productivity was estimated 46 47 21 to be $49.25 million, for a total economic burden of $272.9 million in 2007. Applying this 48 49 22 method to the current study, with an excess incidence of 3,625 persons, the cost of additional 50 51 23 allergic rhinitis due to the oil spill is roughly estimated to be $243,170 for one year (2008). 52 53 15 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 51 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 These results indicate that oil spills near coastlines can cause considerable adverse health 9 effects and that the summation of disease burden should not be neglected, even though the 10 2 11 3 size of the population affected may be small. 12 13 4 The health impacts of oil spills on residents of contaminated areas vary by age, sex, and 14 15 5 region. Approximately Forone year after peer the accident, the reviewburden of mental disease only attributable 16 17 6 to the oil spill was higher in men than in women, whereas the burden of allergies was higher 18 19 7 in women than men. The most heavily impacted area, Area 1, comprised the majority of the 20 21 8 BOD attributable to the oil spill. For example, about 60% of the additional incidences of 22 23 9 disease occurred in Area 1, less than 1.1 km from the contaminated coast, with 22% in Area 2 24 25 10 and 6% in Area 3. Critical health effects of oil spills are more likely to affect clean-up 26 27 11 workers due to higher and repeated exposures to hazardous pollutants. Residents of Area 1 28 29 12 participated in clean-up work for an average of 122 days, and residents of Areas 2 and 3 30 31 13 participated for averages of 50 and 36 days. Previous studies on the acute health impact of oil 32 33 14 spills [4-6] also reported that more frequent and greater exposure to clean-up activities was 34 15 strongly associated with a higher occurrence of symptoms, including both acute and chronic 35 http://bmjopen.bmj.com/ 36 health outcomes. 37 16 38 17 ToWe estimated the excess incidences of illnesses caused by an oil spill, we useding 39 Formatted:Formatted:Formatted: Indent: First line: 1 ch 40 18 prevalence data classified by distance from the contaminated coast, under the assumption that 41 42 19 the area farthest distant, Area 4, was not affected by the oil spill, and that the prevalence of

43 on October 1, 2021 by guest. Protected copyright. 44 20 diseases was similar in all study areas before the accident, which is one of limitations of this 45 46 21 study. These assumptions were used because real data describing disease prevalence before 47 48 22 and after the oil spill were not available. Residents in Area 4, the reference area, had very 49 50 23 little participation in the clean-up efforts, which could account for differences in disease 51 52 24 prevalence between Area 4 and other areas. In addition, seventy-two hours after the oil spill, 53 16 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 52 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 the downwind direction changed from the southeast to the west and as a result, volatile 9 components did not move further inland.[165] The benzene concentrations in the atmosphere 10 2 11 3 after the Hebei Spirit oil spill simulated in our previous study [242] also showed that 12 13 4 concentrations reduced rapidly with distance from the contaminated coastline. Benzene 14 15 5 concentrations over 0.1For ppm were not peer detected at the areareview more than 13 km from onlythe spill site. 16 17 6 Therefore, the assumption that Area 4 was not affected by the oil spill is reasonable. Kim et 18 19 7 al.[253] reported that the prevalence of rhinitis in Korean population was 16.4%, 24.7%, 20 21 8 and 21.7% in 20-35, 36-50, and older than 50 age group, respectively, while that of Area 4 22 23 9 adults was 15.8% and 17.9% in 19-49 and over 50 age group, respectively, showing slightly 24 25 10 lower prevalence than whole Korean population. 26 27 11 It is possible that PTSD and depression may overlap. According to a previous cross- 28 29 12 sectional study of psychological disease in Korea, about 8.5% of depression is accompanied 30 31 13 by PTSD.[264] This indicates that the estimated disease burden may have been overestimated. 32 33 14 For this reason, Wwe did not include anxiety in our analyses, although prevalence data for 34 15 anxiety were available because about 15% of anxiety is accompanied by depression in 35 http://bmjopen.bmj.com/ 36 Korea.[26] 37 16 38 17 Although the PAHs and VOCs in spilled oil can cause haematologic cancer, lung cancer, 39 40 18 and stomach cancer,[6] we did not assess the incidences of any cancers due to the lack of data. 41 L 42 19 The TEHC analysed the urinary malondialdehyde (MDA) and 8-hydroxy-2 -deoxyguanosine

43 on October 1, 2021 by guest. Protected copyright. 44 20 (8-OHdG) levels of residents in the contaminated area, as oxidative stress indices and the 45 46 21 TEHC found that the urinary MDA and 8-OHdG concentrations of the residents living in 47 48 22 Area 1 were much higher than in other areas. The average urinary MDA concentration of 49 50 23 residents in Area 1 was 5.42 (SD=3.14) µmol/g Cr, whereas that of other areas was 3.83 51 52 24 (SD=2.58) µmol/g Cr.[154] The average 8-OHdG concentration for Area 1 was 6.84 53 17 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 53 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 (SD=3.84) µg/g Cr, while that of other areas was 4.67 (SD=3.00) µg/g Cr. Previous studies 9 reported that oxidative DNA damage is associated with increased risk of cancer, including 10 2 11 3 haepatocellular carcinoma, lung cancer, and breast cancer, and thus the MDA and haepatic 8- 12 13 4 OHdG levels are useful as markers to identify subjects at extremely high-risk.[275-297] The 14 15 5 increased urinary MDAFor and 8-OHdG peer concentrations inreview the residents in this study only area may 16 17 6 cause cancers in the future. By monitoring more long-term effects, the BOD imposed by 18 19 7 these diseases may be quantified and added to the results of this study. 20 21 8 This is the first study to quantify the BOD of a single environmental disaster. The EBD 22 23 9 of the current study was estimated for a highly exposed population after a single event. This 24 25 10 study demonstrates that BOD is an objective and comprehensive metric for estimating and 26 27 11 comparing the health effects of environmental hazards and disasters across different regions 28 29 12 and time periods. One of strengths of this study is that we measured BOD attributable to oil 30 31 13 spill by sub-region as well as age and sex including both physical and mental health 32 33 14 outcomes, which can contribute to prioritize the mitigation policy to reduce adverse health 34 15 effects. 35 http://bmjopen.bmj.com/ 36

37 16 38 39 17 Conclusions Formatted:Formatted:Formatted: Font: 16 pt 40 Formatted:Formatted:Formatted: Indent: First line: 0" 41 18 Oil spills near coastlines can cause considerable adverse health effects. The BOD for one Formatted:Formatted:Formatted: Font: 16 pt, Bold 42

43 19 year for the residents living near contaminated coastal areas is significant and is related to on October 1, 2021 by guest. Protected copyright. 44 45 20 proximity to the spill as well as participation in clean-up efforts. Asthma and PTSD were 46 47 21 found to comprise the most prominent disease burden in the contaminated areas. The health 48 49 22 impacts of oil spills on residents vary with age, sex, and region implying that rehabilitation 50 51 52 53 18 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 54 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 policy should be community-specific. The closer the population is to the accident site, the 9 greater the BOD is for that population and accident. 10 2 11 3 As the health effects of hazardous pollutants from oil spills are long-lasting, close 12 Formatted:Formatted:Formatted: Justified, Indent: First line: 0.3", Line spacing: Double, Don't 13 4 follow-up studies are required to estimate the cumulative disease burden of oil spills based on hyphenate, Don't adjust space between 14 Asian text and numbers, Font Alignment: 15 5 the identification of long-termFor health peereffects. review only 16 17 6 18 19 7 Formatted:Formatted:Formatted: Justified, Indent: First line: 20 0.3", Line spacing: Double, Don't 21 hyphenate, Don't adjust space between 22 Asian text and numbers, Font Alignment: 23 24 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 19 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 55 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 ARTICLE SUMMARY 9 10 2 Article focus 11 3 ▶ This study aimed at assessing the burden of disease (BOD) of the residents 12 living in the contaminated area due to an environmental disaster, oil spill,of the 13 4 14 5 contaminated coastal area 15 6 ▶ and at analyzedingFor the BOD peer attributable to reviewthe oil spill by disease, onlyage group, 16 7 and sub-region, which can inform policy prioritization to mitigate adverse health 17 8 effects for specific exposed populations. 18 9 19 10 Key messages 20 21 11 ▶ This study providesd an opportunity to assess the disease burden of the 22 12 residents due to an environmental disaster, integrating mental and physical 23 13 symptoms. 24 25 14 ▶ Among diseases, years lived withlost due to disabilities (YLDs) for mental 26 15 disease such as post-traumatic stress disorder (PTSD) and depression 27 16 attributable to the oil spill are higher in males than females and those for 28 17 asthma and allergies are higher in females than males. 29 30 18 ▶ Oil spills near coastlines can cause considerable adverse health effects and that 31 19 the summation of disease burden is not negligible, even though the size of the 32 20 population affected may be small. 33 21 34 22 Strength and limitations 35 23 ▶ This is the first study to quantify the BOD of a single environmental disaster. http://bmjopen.bmj.com/ 36 37 24 ▶ This study demonstrates that BOD is an objective and comprehensive metric for 38 25 estimating and comparing the health effects of environmental hazards and 39 26 disasters across different regions and time periods. 40 27 ▶To estimate excess incidence, we used prevalence data classified by distance 41 28 from the contaminated coast, under the assumption that the area farthest 42 29 distant, Area 4, was not affected by the oil spill, and that the prevalence of

43 30 diseases was similar in all study areas before the accident because real data on October 1, 2021 by guest. Protected copyright. 44 31 describing disease prevalence before and after the oil spill were not available. 45 32 ▶ We did not consider Llong-term health outcomes such as cancers and chronic 46 47 33 degenerative diseases mainly because of lack of datacould not be estimated. 48 34 49 50 51 52 53 20 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 56 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 Acknowledgements Formatted:Formatted:Formatted: Indent: Left 0 ch, First line: 0 ch 9 2 This work was supported by the Taean Environmental Health Center. The authors would like 10 11 3 to thank Woo-Cheol Jung and Jongil Hur for initiating the study and providing the funding 12 13 4 and data. 14 15 5 For peer review only 16 17 6 Competing interests 18 19 7 The authors have no competing interests. 20 21 8 22 23 9 Funding 24 25 10 This work was supported through the Taean Environmental Health Center by the Ministry of 26 27 11 Environment, Republic of Korea. 28 29 12 30 31 13 References 32 14 33 15 1. County chief of Taean. Taean Tonggyeyeongam, No.18. Taean: Planning and Inspection 34

35 16 Office of Taean County 2008. http://law.taean.go.kr/asp/tong/tong_file/2007/02.pdf http://bmjopen.bmj.com/ 36 37 17 (date accessed 5 Oct 2010). 38 39 18 2. Ji K, Seo J, Liu X, et al. Genotoxicity and endocrine-disruption potentials of sediment 40 41 19 near an oil spill site: two years after the Hebei Spirit oil spill. Environ Sci Technol 42 20 2011;45:7481-7488.

43 on October 1, 2021 by guest. Protected copyright. 44 3. World Health Organization, International Agency for Research on Cancer. IARC 45 21 46 Monographs on the Evaluation of Carcinogenic Risks to Humans overall Evaluations of 47 22 48 23 Carcinogenicity: An Updating of IARC Monographs Volumes 1 to 42 (Supplement 7). 49 50 24 Lyon: International Agency for Research on Cancer 1998. 51 52 53 21 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 57 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 4. Sim MS, Jo IJ, Song HG. Acute health problems related to the operation mounted to clean 9 the Hebei Spirit oil spill in Taean, Korea. Mar Pollut Bull 2010;60:51-57. 10 2 11 12 3 5. Zock JP, Rodrı´guez-Trigo G, Pozo-Rodrı´guez F, et al. SEPAR-Prestige Study Group: 13 14 4 prolonged respiratory symptoms in clean-up workers of the Prestige oil spill. Am J Resp 15 For peer review only 16 5 Crit Care Med 2007;176:610-616. 17 18 6 6. Rodrı´guez-Trigo G, Zock JP, Pozo-Rodrı´guez F, et al. SEPAR-Prestige Study Group: 19 20 7 Health changes in fishermen 2 years after clean-up of the Prestige oil spill. Ann Intern 21 22 8 Med 2010;153:489-498. 23 24 9 7. Lyons RA, Temple JM, Evans D, et al. Acute health effects of the Sea Empress oil spill. J 25 26 10 Epidemiol Community Health 1999;53:306-331. 27 28 11 8. Song M, Hong YC, Cheong HK, et al. Psychological health in residents participating in 29 30 12 clean-up works of Hebei Spirit oil spill. J Prev Med Public Health 2009;42:82-88. 31 32 13 [Korean] 33 34 14 9. Palinkas LA, Russell J, Downs MA, et al. Ethnic differences in stress, coping, and

35 http://bmjopen.bmj.com/ 36 15 depressive symptoms after the Exxon Valdez oil spill. J Nerv Ment Dis 1992;180:287- 37 295. 38 16 39 10. Palinkas LA, Petterson JS, Russell J, et al. Community patterns of psychiatric disorders 40 17 41 18 after the Exxon Valdez oil spill. Am J Psychiatr 1993;150:1517-1523. 42

43 on October 1, 2021 by guest. Protected copyright. 19 11. Palinkas LA, Petterson JS, Russell JC, et al. Ethnic differences in symptoms of 44 45 20 posttraumatic stress after the Exxon Valdez oil spill. Prehosp Disaster Med 2004;19:102- 46 47 21 112. 48 49 22 12. Lee CH, Kang YA, Chang KJ, et al. Acute health effects of the Hebei oil spill on the 50 51 23 residents of Taean, Korea. J Prev Med Public Health 2010;43:166-73. [Korean] 52 53 22 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 58 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 13. Murray CJ, Lopez AD. The Global Burden of Disease: A Comprehensive Assessment of Formatted:Formatted:Formatted: Indent: Left: 0", Hanging: 9 1.77 ch, First line: -1.77 ch, Outline 2 Mortality and Disability from Disease, Injuries and Risk Factors in 1990 and Projected to numbered + Level: 1 + Numbering Style: 10 1, 2, 3, … + Start at: 1 + Alignment: Left + 11 Aligned at: 0" + Indent at: 0" 3 2020. Cambridge, MA: Harvard University Press1996. 12 13 4 13.14. Ezzati M, Lopez AD, Rodger HSV, et al. Selected major risk factors and global and 14 15 For peer review only Formatted:Formatted:Formatted: Font: Italic 16 5 regional burden of disease. Lancet 2002;360:1347–60. 17 Formatted:Formatted:Formatted: Font: Bold 18 6 14.15. Taean Environmental Health Center, Ministry of Environment, Republic of Korea. 19 20 7 Mid and Long Term Health Effect of Hebei Spirit Oil Spill: First Year Survey Report. 21 22 8 Taean: Taean Environmental Health Center 2011.[Korean] 23 24 9 15.16. Kim JH, Kwak BK, Ha M, et al. Modeling human exposure levels to airborne 25 26 10 volatile organic compounds the by Hebei Spirit oil spill. Environ Health Toxicol 27 28 11 2012;27:e2011006. 29 30 12 16.17. Barendregt JJ, Oortmarssen GJ, Vos T, et al. A generic model for the assessment of 31 32 13 disease epidemiology: the computational basis of DisMod II. Popul Health Metr Formatted:Formatted:Formatted: Normal, Indent: Left: 0", Hanging: 1.77 ch, First line: -1.77 ch, 33 Line spacing: Double, Don't hyphenate, 14 2003;1:4-10. 34 Font Alignment: Baseline

35 Formatted:Formatted:Formatted: Font: (Default) Times New http://bmjopen.bmj.com/ 15 17.18. Statistics Korea. Annual Report on the Cause of Death Statistics (2008). Seoul: 36 Roman, 12 pt, Kern at 1.5 pt 37 Formatted:Formatted:Formatted: Font: (Default) Times New Statistics Korea 2009. 38 16 Roman, 12 pt, Kern at 1.5 pt 39 FormattedFormattedFormatted 17 18.19. Public Health Division, Department of Human Services, Australia. Victorian Burden 40 Formatted:Formatted:Formatted: Font: (Default) Times New 41 Roman, 12 pt 18 of Disease Study: Morbidity. Melbourne: Public Health Division 1999. Formatted:Formatted:Formatted: Indent: Left: 0", Hanging: 42 1.77 ch

43 on October 1, 2021 by guest. Protected copyright. 19 20. Stoudhard MEA. Disability weights for diseases: a modified protocol and results for a Formatted:Formatted:Formatted: Font: (Default) Times New 44 Roman, 12 pt Formatted:Formatted:Formatted: Font: (Default) Times New 45 20 Western European region. Eur J Public Health 2000;10:24-30. 46 Roman, 12 pt 47 Formatted:Formatted:Formatted: Font: (Default) Times New 21 19.21. Lee JK, Yoon SJ, Do YK, Kwon YH, et al. Disability weights for diseases in Korea. Roman, 12 pt, Italic 48 Formatted:Formatted:Formatted: Font: (Default) Times New 49 22 Korean J Prev Med 2003;36:1-26 (in Korean). Roman, 12 pt 50 Formatted:Formatted:Formatted: Font: (Default) Times New 51 23 20.22. Yoon SJ, Bae SC, Lee SI, et al. Measuring the burden of disease in Korea. J Korean Roman, 12 pt, Bold 52 Formatted:Formatted:Formatted: Font: (Default) Times New Roman, 12 pt 53 23 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 59 of 60 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 Med Sci 2007;22:518-523. 9 21.23. Kim SY, Yoon SJ, Jo MW, et al. Economic burden of allergic rhinitis in Korea, Am J 10 2 11 3 Rhinol Allergy 2010;24:e110-e113. 12 13 4 22.24. Kim YM, Cheong HK, Kim J, et al. Scientific basis of environmental health 14 15 5 contingency planningFor for a coastal peer oil spill. J Prev review Med Public Health 2009; only42:73-81. 16 17 6 [Korean] 18 19 7 23.25. Kim TB, Kim YK, Chang YS, et al. Association between sensitization to outdoor 20 21 8 spider mites and clinical manifestations of asthma and rhinitis in the general population 22 23 9 of adults. J Korean Med Sci 2006;21:247-52. 24 25 10 24.26. Ministry of Health and Welfare, College of Medicine, Seoul National University. The 26 27 11 Epidemiological Survey of Psychiatric Illnesses in Korea. Seoul: Ministry of Health and 28 29 12 Welfare 2006. [Korean] 30 31 13 25.27. Gonenc A, Oezkan Y, Torun M, et al. Plasma malondialdehyde (MDA) levels in 32 33 14 breast and lung cancer patients. J Clin Pharm Ther 2001;26:141-144. 34 15 26.28. Suzana S, Normah H, Fatimah A, et al. Antioxidant intake and status, and oxidative 35 http://bmjopen.bmj.com/ 36 stress in relation to breast cancer risk: a case-control study. Asian Pacific J Cancer Prev 37 16 38 17 2008;9:343-350. 39 40 18 27.29. Tanaka H, Fujita N, Sugimoto R, et al. Hepatic oxidative DNA damage is associated 41 42 19 with increased risk for hepatocellular carcinoma in chronic hepatitis C. Br J Cancer

43 on October 1, 2021 by guest. Protected copyright. 44 20 2008;98:580-586. 45 46 21 47 48 49 50 51 52 53 24 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 60 of 60 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 1 Figures 9 10 11 2 Figure 1. Study area and classification of affected regions. 12 13 3 ★: accident site, about 8 km away from the Taean coastline. 14 15 4 For peer review only 16 17 18 5 Figure 2. Burdens of disease attributabledue to the Hebei Spirit oil spill by disease, age and 19 20 6 sex. in Taean County and the Boryeong Islands, Korea. 21 22 7 YLD: years lived with disability; PTSD: Mental disease includes post-traumatic stress 23 24 8 disorder (PTSD) and depression; Allergy includes rhinitis, dermatitis, and conjunctivitis. 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 25 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

Burden of disease attributable to the Hebei Spirit oil spill in Taean, Korea

ForJournal: peerBMJ Open review only Manuscript ID: bmjopen-2013-003334.R2

Article Type: Research

Date Submitted by the Author: 08-Aug-2013

Primary Subject Public health Heading:

Secondary Subject Heading: Epidemiology, Health economics, Occupational and environmental medicine

EPIDEMIOLOGY, HEALTH ECONOMICS, Epidemiology < THORACIC http://bmjopen.bmj.com/ Keywords: MEDICINE, MENTAL HEALTH, TOXICOLOGY

on October 1, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

43 on October 1, 2021 by guest. Protected copyright. 44 18 polycyclic aromatic hydrocarbons (PAHs), post-traumatic stress disorder (PTSD), disability 45 46 19 weight 47 48 20 Word count: 49 21 Abstract: 299 words 50 51 22 Main text: 3356 words 52 23 No. of tables: 3 53 24 No. of Figures: 2 54 25 No. of references: 29 55 56 57 1 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 2 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 ABSTRACT 5 6 2 Objectives: We aimed at assessing the burden of disease (BOD) of the residents living in 7 8 contaminated coastal area with oil spill and also analyzed the BOD attributable to the oil spill 9 3 10 11 4 by disease, age, sex, and sub-region. 12 13 5 Design: Health impact assessment by measuring years lived with disability (YLD) due to an 14 15 6 oil spill. For peer review only 16 17 18 7 Setting: A whole population of a community affected by an anthropogenic environmental 19 20 8 disaster and secondary health outcome data. 21 22 23 9 Participants: Based on the health outcome survey upon 10,171 individuals (male: 4,354; 24 25 10 female: 5,817), BOD of 66,473 populations (male: 33,441; female: 33,032) was measured. 26 27 11 Interventions: None. Observational study on the effect of a specific environmental health 28 29 12 hazard. 30 31 32 13 Primary and secondary outcome measures: Using disability adjusted life year (DALY) 33 34 14 method, BOD including both physical and mental diseases was measured. For the BOD

35 http://bmjopen.bmj.com/ 36 15 measurement, excess incidences of illnesses related to oil spill were estimated from the 37 38 16 comparison of prevalence of the health outcomes between contaminated areas and reference 39 40 17 area without contamination. 41 42 18 Results: YLD attributable to the oil spill were estimated to be 14,724 DALYs (male: 7,425

43 on October 1, 2021 by guest. Protected copyright. 44 45 19 DALYs; female: 7,299 DALYs) for the year 2008. The YLD of mental diseases including 46 47 20 post-traumatic stress disorder (PTSD) and depression for males were higher than that for 48 49 21 females. The YLD for females was higher in asthma and allergies (rhinitis, dermatitis, 50 51 52 22 conjunctivitis) than that for males. The effects of asthma and allergies were the greatest for 53 54 23 people in their 40s, with the burden of mental illness being the greatest for those in their 20s. 55 56 57 2 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 3 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

43 on October 1, 2021 by guest. Protected copyright. 18 associated with higher frequencies of psychological symptoms. [8-11] Rodriguez-Trigo [6] 44 45 46 19 reported that participation in a major oil spill clean-up is associated with chromosomal 47 48 20 damage, although they were unable to determine whether oil exposure itself caused the 49 50 21 abnormalities. 51 52 22 After the Hebei Spirit oil spill, more than 2,000,000 people (person-days), including over 53 54 1,200,000 person-days by volunteers, participated in clean-up activities during the first seven 55 23 56 57 3 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 4 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

43 on October 1, 2021 by guest. Protected copyright. 44 19 Taean coastal area due to the Hebei Spirit oil spill and to analyze the BOD by disease, age, 45 46 20 sex, and sub-region. 47 48

49 21 50 51 22 METHODS 52 53 54 23 Study area and design 55 56 57 4 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 5 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

43 on October 1, 2021 by guest. Protected copyright. 44 19 moved in a southeasterly direction, following the coastline from the accident point. 45 46 20 We classified the residents of Taean County and the Boryeong islands into five regions 47 48 based on both distance from the contaminated coastline and contamination level. Area 1 is 49 21 50 51 22 less than 1.1 km away from the coastline adjacent to the accident location where the coastline 52 53 23 was contaminated at a high level. Area 2 is 1.1-4.2 km away with no coastline, but was 54 55 24 directly affected by atmospheric VOCs and PAHs due to the wind direction. Area 3 is 4.2-23 56 57 5 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 6 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

29 12 30 31 13 Estimation of burden of disease attributable to oil spill 32 33 14 To estimate the BOD attributable to the oil spill, we applied the DALY method. DALY is 34

43 on October 1, 2021 by guest. Protected copyright. 44 19 and disability weight (DW). The formula we utilized is as follows; 45 46 20 47 48 49 1 1 50 51 52 21 [1] 53 1 54 55 22 where, D is disability weight; r is discount rate; a is age of onset of the disability; L is 56 57 6 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 7 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 duration of the disability; is parameter from the age weighting function; and C is constant. 5 β 6 7 8 2 We used 3% of discount rate and 0.04 and 0.0618 for β and C, respectively, as suggested by 9 10 11 3 the WHO.[13] 12 13 4 Because there was no prevalence data recorded prior to the accident, we estimated excess 14 15 5 incidences usingFor prevalence peer data from thereview TEHC under the followingonly assumptions. We first 16 17 assumed that residents living in Area 4 had no negative health consequences due to the oil 18 6 19 20 7 spill and used this as the baseline prevalence. Area 4 was assumed to be the reference area, 21 22 8 not only because the area might not be impacted by the oil spill directly as it is far from the 23 24 9 contaminated coastline, but also because residents in Area 4 rarely participate in the clean-up 25 26 10 works. Secondly, we assumed that the prevalence of oil spill-related diseases in the whole 27 28 29 11 study area had been same prior to the accident. We then estimated excess incidences using 30 31 12 prevalence differences between Area 4 and the other areas stratified by area, age, sex, and 32 33 13 disease using following formula. 34 35 14 [2] http://bmjopen.bmj.com/ 36 ∑ 37 here, is the excess incidence of region i and disease j; is the population of 38 15 39 40 region i; is the prevalence of region i and disease j; is the prevalence of control 16 41 42 17 region and disease j.

43 on October 1, 2021 by guest. Protected copyright. 44 45 18 The duration of disability and the average onset age were calculated using the DISMOD 46 47 19 II model [17] for PTSD, rhinitis, dermatitis, and conjunctivitis based on prevalence data from 48 49 20 the TEHC and the mortality rate of South Korea [18] by age, sex, and disease. For asthma, 50 51 21 the results of the Victoria, Australia study [19] were used. 52 53 DWs for each disease were measured following a method previously used by the Global 54 22 55 56 57 7 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 8 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 Burden of Disease group [13] and the Dutch Disability Weights Group.[20] A panel that 5 6 2 included 12 health and medicine specialists was convened. Eleven indicator conditions 7 8 including asthma, stomach cancer, diabetes, dementia, major depression, appendicitis, thyroid 9 3 10 11 4 cancer, hepatitis A, otitis media, atopic dermatitis, and common cold were measured using the 12 13 5 person trade-off (PTO) method. The eleven indicator conditions were selected in accordance 14 15 6 with a previousFor study onpeer Korean disability review weight. [21] That only is, we considered that they 16 17 7 should be meaningful in public health and relatively popular diseases, while distributing the 18 19 20 8 whole DW range (0 to 1) evenly.[21] A facilitator first explained the method and then led the 21 22 9 discussion to determine the PTO value of each participant. The discussion was repeated twice. 23 24 10 Then, the visual analogue scale (VAS) method was applied to obtain DWs for 118 conditions 25 26 11 including asthma, rhinitis, dermatitis, conjunctivitis, PTSD, and depression used in this study. 27 28 Inter-method reliability was confirmed between PTO and VAS using 11 indicator conditions, 29 12 30 31 13 which showed fairly high consistency (Spearman’s correlation coefficient 0.966, P- 32 33 14 value<0.001). Reliability was confirmed by test-retest after three weeks by seven specialists 34

35 15 and obtained fairly high consistency (Spearman’s correlation coefficient 0.973, P- http://bmjopen.bmj.com/ 36 37 16 value<0.001). 38 39 40 17 Finally, we estimated YLD caused by the Hebei Spirit oil spill utilizing the estimated 41 42 18 excess incidence attributable to the oil spill, DW, and the onset age and duration of disability

43 on October 1, 2021 by guest. Protected copyright. 44 19 by disease, age, sex, and area. 45 46 47 20 48 49 21 RESULTS 50 51

52 22 53 54 23 Summary of the characteristics of the study area 55 56 57 8 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 9 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 The populations of Taean County and the Boryeong islands were 63,053 individuals 5 6 2 (male: 31,636; female: 31,417) and 3,420 individuals (male: 1,805; female: 1,615) in 2008, 7 8 respectively. Broken down by area, 8,348 (13%), 11,272 (17%), 4,597 (7%), 38,836 (58%), 9 3 10 11 4 and 3,420 (5%) residents lived in Areas 1, 2, 3, 4, and the Boryeong islands, respectively. 12 13 5 Area 4, the reference area, was the most populous because it includes a city center. The 14 15 6 largest age groupFor consisted peer of subjects inreview their 50s, with individuals only in their 40s, 50s, and 60s 16 17 7 comprising 45% of the total population (Data are presented in Data Supplement, Table S1). 18 19 20 8 A total of 10,171 residents living in the study area participated in the TEHC survey. 21 22 9 Among them, 4,354 (42.8%) and 5,817 (57.2%) residents were male and female, respectively. 23 24 10 Individuals under 15 years of age, 15-65, and over 65 years of age comprised 7.2%, 48.0%, 25 26 11 and 44.8% of the population, respectively. Residents of Area 1 participated in clean-up work 27 28 for an average of 122 days, and residents of Areas 2, 3, and the Boryeong islands participated 29 12 30 31 13 for averages of 50, 36, and 100 days, respectively.[14] Table 1 shows the prevalence of 32 33 14 disease related to the current study by disease, sex, and area derived from a survey of 9,246 34

35 15 adult (over 19 years old) residents. For young people under 19 years old, we used results http://bmjopen.bmj.com/ 36 37 16 from 925 school students (male: 505; female: 420) living in the study area. More details 38 39 40 17 including prevalence by age are shown in Data Supplement, Table S2. 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 9 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from Page 10 of 56 42.6 10.1 40.2 17.0 15.0 24.5 10.8 21.3 13.5 25.2 23.8 27.9 Male Female Boryeong Boryeong Islands

Area 4 2.4 9.2 4.4 10.0 4.6 5.3 3.2 11.6 8.7 7.6 20.2 15.9 Male Female

3.1 10 10 Area 3

BMJ Open 9.4 4.7 2.9 8.8 14.0 9.6 10.4 23.9 10.6 13.2 9.0 Male Female

http://bmjopen.bmj.com/

7.1 Area 2

5.1 6.1 7.0 8.8 14.8 25.2 20.2 12.6 12.6 15.4 11.7 Male Female on October 1, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

a) For peer review only Area 1 22.0 20.4 32.9 34.6 Male Male Female . Prevalence of diseases related to this study (in study %). this related to . of diseases Prevalence b) Rhinitis 41.4 34.6 Dermatitis Conjunctivitis 19.6 32.0 25.0 38.7 Depression 23.3 29.9 PTSD Table 1 Table a) Area 1: less than 1.1 km away; Area 2: 1.1-4.2 km away; Area 3: 4.2-23 km away; Area 4: more than 23 km from the contaminated contaminated the km from 23 than 4: more Area away; km 3: 4.2-23 Area away; km 2: 1.1-4.2 Area away; km 1.1 1: less than Area a) Allergy Allergy Asthma Disease coastline; b) PTSD: posttraumatic stress stress disorder. coastline; b) posttraumatic PTSD: Mental Mental disease 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 11 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 Estimated burden of disease due to oil spill 5 6 2 Excess incidence attributable to the oil spill 7 8 The excess incidence of disease, estimated from the available prevalence data, caused by 9 3 10 11 4 the oil spill varied by area, sex, and disease (Table 2). Excess incidence by age is shown in 12 13 5 Data Supplement, Table S3.There were additional incidences of asthma (2,088: 989 male; 14 15 6 1,099 female),For rhinitis (3,625:peer 1,855 male; review 1,770 female), dermatitis only (1,976: 962 male; 1,014 16 17 7 female), conjunctivitis (2,992: 1,365 male; 1,627 female), PTSD (2,681:1,300 male; 1,381 18 19 20 8 female), and depression (2,326: 1,370 male; 956 female) in Areas 1,2,3 and the Boryeong 21 22 9 islands. 23 24 10 Area 1 had the highest additional incidences for all diseases, while Area 3, a relatively 25 26 11 less affected area than Areas 1 and 2, showed the lowest additional incidences for all diseases. 27 28

29 12 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

Total Total

989 1,099 962 1,014 Male Female 1,370 956 1,300 1,381 1,855 1,770 1,365 1,627

76 76 106 113 113 190 116 116 138 201 201 149 353 353 354 222 222 322 Male Female Boryeong Boryeong Islands

0 Area 3 Area 3 12 12 c

0 65 65 57 53 53 97 94 94 32 209 207 101 183 Male Male Female BMJ Open

http://bmjopen.bmj.com/

Area 2

266 169 133 222 453 481 241 231 294 214 146 158 Male Female

a For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Area 1

839 839 729 473 473 449 838 838 581

Table 2. 2. Table a: Area 1, less than 1.1 km away; Area 2, 1.1-4.2 km away; Area 3, 4.2-23 km away; Area 4, more than 23 km from the contaminated coastline; coastline; contaminated the km from 23 than more 4, away; Area km 4.2-23 3, away; Area km 1.1-4.2 2, Area away; than km 1.1 less 1, a: Area area. and reference 3 Area between c: difference stress disorder; b:No post-traumatic Allergy Allergy Asthma Disease Depression PTSD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 13 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

28 weight Asthma 29 Mild intermittent 0.222 30 Mild persistent 0.291 31 Moderate persistent 0.510 32 Severe persistent 0.715 33 34 Very severe persistent 0.836

35 Allergy Rhinitis 0.205 http://bmjopen.bmj.com/ 36 Dermatitis 0.132 37 38 Conjunctivitis 0.212

39 a 40 PTSD 0.533 41 Depression 0.219 42 a: post-traumatic stress disorder 43 11 on October 1, 2021 by guest. Protected copyright. 44 45 12 46 47 13 Burden of disease due to oil spill 48 49 Figure 2 presents the YLD according to disease, sex, and age group. The YLD for 50 14 51 52 15 asthma, allergic rhinitis, dermatitis, conjunctivitis, PTSD, and depression were approximately 53 54 16 6,138 DALYs, 2,420 DALYs, 1,417 DALYs, 18 DALYs, 2,869 DALYs, and 1,862 DALYs, 55 56 57 13 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 14 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 respectively. The total YLD attributable to the Hebei Spirit oil spill for the year 2008 was 5 6 2 14,724 DALYs. The YLD for males was 7,425 DALYs and for females was 7,290 DALYs. 7 8 The YLD for males for PTSD and depression were higher than for females, and the YLD for 9 3 10 11 4 females was higher in asthma and allergies than for males (More details are shown in Data 12 13 5 Supplement, Table S4). Among the included diseases, the YLD for asthma was the highest. 14 15 6 The YLD forFor asthma and peer allergies was thereview highest for residents only in their 40s, and the YLD for 16 17 7 mental disease was the highest for residents in their 20s. 18 19 20 8 Figure 2 21 22 9 23 24 10 DISCUSSION 25 26 27 11 28 29 12 The total YLD attributable to the Hebei Spirit oil spill was estimated at 14,724 DALYs 30 31 for 2008. Asthma was found to represent the most prominent disease burden (6,138 DALYs) 32 13 33 34 14 among the six diseases, followed by PTSD and rhinitis. The asthma burden of 6,138 DALYs

43 on October 1, 2021 by guest. Protected copyright. 44 45 19 as having allergic rhinitis. The direct cost was $223.7 million and the lost productivity was 46 47 20 estimated to be $49.3 million, for a total economic burden of $272.9 million in 2007. 48 49 21 Applying this method to the current study, with an excess incidence of 3,625 persons, the cost 50 51 of additional allergic rhinitis due to the oil spill is roughly estimated to be $243,170 for one 52 22 53 54 23 year (2008). These results indicate that oil spills near coastlines can cause considerable 55 56 57 14 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 15 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

35 15 health outcomes. http://bmjopen.bmj.com/ 36 37 16 To estimate the excess incidences of illnesses caused by an oil spill, we used prevalence 38 39 40 17 data classified by distance from the contaminated coast, under the assumption that the farthest 41 42 18 region, Area 4, was not affected by the oil spill, and that the prevalence of diseases had been

43 on October 1, 2021 by guest. Protected copyright. 44 19 same in all study areas before the accident, which is one of limitations of this study. These 45 46 20 assumptions were used because real data describing disease prevalence before and after the 47 48 oil spill were not available. Residents in Area 4, the reference area, had very little 49 21 50 51 22 participation in the clean-up efforts, which could account for differences in disease 52 53 23 prevalence between Area 4 and other areas. In addition, seventy-two hours after the oil spill, 54 55 24 the downwind direction changed from the southeast to the west and as a result, volatile 56 57 15 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 16 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

43 on October 1, 2021 by guest. Protected copyright. 44 19 (8-OHdG) levels of residents in the contaminated area, as oxidative stress indices and the 45 46 20 TEHC found that the urinary MDA and 8-OHdG concentrations of the residents living in 47 48 Area 1 were much higher than in other areas. The average urinary MDA concentration of 49 21 50 51 22 residents in Area 1 was 5.42 (SD=3.14) µmol/g Cr, whereas that of other areas was 3.83 52 53 23 (SD=2.58) µmol/g Cr.[15] The average 8-OHdG concentration for Area 1 was 6.84 (SD=3.84) 54 55 24 µg/g Cr, while that of other areas was 4.67 (SD=3.00) µg/g Cr. Previous studies reported that 56 57 16 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 17 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 oxidative DNA damage is associated with increased risk of cancer, including haepatocellular 5 6 2 carcinoma, lung cancer, and breast cancer, and thus the MDA and haepatic 8-OHdG levels are 7 8 useful as markers to identify subjects at extremely high-risk.[27-29] The increased urinary 9 3 10 11 4 MDA and 8-OHdG concentrations in the residents in this study area may be useful to track 12 13 5 the risks for excess cancers in the future. By monitoring more long-term effects, the BOD 14 15 6 imposed by theseFor diseases peer may be quantified review and added to the resultsonly of this study. 16 17 7 This is the first study to quantify the BOD of a single environmental disaster. The EBD 18 19 20 8 of the current study was estimated for a highly exposed population after a single event. This 21 22 9 study demonstrates that BOD is an objective and comprehensive metric for estimating and 23 24 10 comparing the health effects of environmental hazards and disasters across different regions 25 26 11 and time periods. One of strengths of this study is that we measured BOD attributable to oil 27 28 spill by sub-region as well as age and sex including both physical and mental health 29 12 30 31 13 outcomes, which can help to prioritize the health policies and mitigation efforts to reduce 32 33 14 adverse health effects. 34

43 on October 1, 2021 by guest. Protected copyright. 44 45 19 proximity to the spill as well as participation in clean-up efforts. Asthma and PTSD were 46 47 20 found to comprise the most prominent disease burden in the contaminated areas. The health 48 49 21 impacts of oil spills on residents vary with age, sex, and region implying that any policies to 50 51 reduce the excess BOD should be community-specific. 52 22 53 54 55 56 57 17 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 18 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 As the health effects of hazardous pollutants from oil spills are long-lasting, close 5 6 2 follow-up studies are required to estimate the cumulative disease burden of oil spills based on 7 8 the identification of long-term health effects. 9 3 10 11 4 12 13 14 5 ARTICLE SUMMARY 15 For peer review only 16 17 6 Article focus 18 7 ▶ This study aimed at assessing the burden of disease (BOD) of the residents 19 8 living in the contaminated area due to an environmental disaster, oil spill, and 20 21 9 analyzed the BOD attributable to the oil spill by disease, age group, and sub- 22 10 region, which can inform policy prioritization to mitigate adverse health effects. 23 11 24 25 12 Key messages 26 13 ▶ This study provides an opportunity to assess the disease burden of the 27 28 14 residents due to an environmental disaster, integrating mental and physical 29 15 symptoms. 30 31 16 ▶ Among diseases, years lived with disabilities (YLDs) for mental disease such as 32 17 post-traumatic stress disorder (PTSD) and depression attributable to the oil spill 33 are higher in males than females and those for asthma and allergies are higher 34 18

35 19 in females than males. http://bmjopen.bmj.com/ 36 20 ▶ Oil spills near coastlines can cause considerable adverse health effects and that 37 38 21 the summation of disease burden is not negligible, even though the size of the 39 22 population affected may be small. 40 23 41 24 Strength and limitations 42

43 25 ▶ This is the first study to quantify the BOD of a single environmental disaster. on October 1, 2021 by guest. Protected copyright. 44 45 26 ▶ This study demonstrates that BOD is an objective and comprehensive metric for 46 27 estimating and comparing the health effects of environmental hazards and 47 28 disasters across different regions and time periods. 48 To estimate excess incidence, we used prevalence data classified by distance 49 29 ▶ 50 30 from the contaminated coast, under the assumption that the farthest region, 51 31 Area 4, has not been affected by the oil spill, and that the prevalence of 52 32 diseases had been same across the whole study area before the accident 53 33 because prevalence data recorded prior to the oil spill were not available. 54 55 56 57 18 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 19 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 19 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 20 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 Acknowledgements 5 6 2 This work was supported by the Taean Environmental Health Center. The authors would like 7 8 3 to thank Woo-Cheol Jung and Jongil Hur for initiating the study and providing the funding 9 10 4 and data. 11 12 5 13 14 15 6 CompetingFor interests peer review only 16 17 7 The authors have no competing interests. 18 19 8 20 21 22 9 Funding 23 24 10 This work was supported through the Taean Environmental Health Center by the Ministry of 25 26 11 Environment, Republic of Korea. 27 28

29 12 30 31 13 References 32 14 33 15 1. County chief of Taean. Taean Tonggyeyeongam, No.18. Taean: Planning and Inspection 34

35 http://bmjopen.bmj.com/ 36 16 Office of Taean County 2008. http://law.taean.go.kr/asp/tong/tong_file/2007/02.pdf 37 38 17 (date accessed 5 Oct 2010). 39 40 18 2. Ji K, Seo J, Liu X, et al. Genotoxicity and endocrine-disruption potentials of sediment 41 42 19 near an oil spill site: two years after the Hebei Spirit oil spill. Environ Sci Technol

43 on October 1, 2021 by guest. Protected copyright. 44 2011;45:7481-7488. 45 20 46 47 21 3. World Health Organization, International Agency for Research on Cancer. IARC 48 49 22 Monographs on the Evaluation of Carcinogenic Risks to Humans overall Evaluations of 50 51 23 Carcinogenicity: An Updating of IARC Monographs Volumes 1 to 42 (Supplement 7). 52 53 24 Lyon: International Agency for Research on Cancer 1998. 54 55 56 57 20 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 21 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 4. Sim MS, Jo IJ, Song HG. Acute health problems related to the operation mounted to clean 5 6 2 the Hebei Spirit oil spill in Taean, Korea. Mar Pollut Bull 2010;60:51-57. 7 8 9 3 5. Zock JP, Rodrı´guez-Trigo G, Pozo-Rodrı´guez F, et al. SEPAR-Prestige Study Group: 10 11 4 prolonged respiratory symptoms in clean-up workers of the Prestige oil spill. Am J Resp 12 13 14 5 Crit Care Med 2007;176:610-616. 15 For peer review only 16 6 6. Rodrı´guez-Trigo G, Zock JP, Pozo-Rodrı´guez F, et al. SEPAR-Prestige Study Group: 17 18 19 7 Health changes in fishermen 2 years after clean-up of the Prestige oil spill. Ann Intern 20 21 8 Med 2010;153:489-498. 22 23 9 7. Lyons RA, Temple JM, Evans D, et al. Acute health effects of the Sea Empress oil spill. J 24 25 10 Epidemiol Community Health 1999;53:306-331. 26 27 8. Song M, Hong YC, Cheong HK, et al. Psychological health in residents participating in 28 11 29 30 12 clean-up works of Hebei Spirit oil spill. J Prev Med Public Health 2009;42:82-88. 31 32 13 [Korean] 33 34 14 9. Palinkas LA, Russell J, Downs MA, et al. Ethnic differences in stress, coping, and

35 http://bmjopen.bmj.com/ 36 15 depressive symptoms after the Exxon Valdez oil spill. J Nerv Ment Dis 1992;180:287- 37 38 39 16 295. 40 41 17 10. Palinkas LA, Petterson JS, Russell J, et al. Community patterns of psychiatric disorders 42

43 18 after the Exxon Valdez oil spill. Am J Psychiatr 1993;150:1517-1523. on October 1, 2021 by guest. Protected copyright. 44 45 19 11. Palinkas LA, Petterson JS, Russell JC, et al. Ethnic differences in symptoms of 46 47 posttraumatic stress after the Exxon Valdez oil spill. Prehosp Disaster Med 2004;19:102- 48 20 49 50 21 112. 51 52 22 12. Lee CH, Kang YA, Chang KJ, et al. Acute health effects of the Hebei oil spill on the 53 54 23 residents of Taean, Korea. J Prev Med Public Health 2010;43:166-73. [Korean] 55 56 57 21 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 22 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 13. Murray CJ, Lopez AD. The Global Burden of Disease: A Comprehensive Assessment of 5 6 2 Mortality and Disability from Disease, Injuries and Risk Factors in 1990 and Projected to 7 8 2020. Cambridge, MA: Harvard University Press1996. 9 3 10 11 4 14. Ezzati M, Lopez AD, Rodger HSV, et al. Selected major risk factors and global and 12 13 5 regional burden of disease. Lancet 2002;360:1347–60. 14 15 For peer review only 16 6 15. Taean Environmental Health Center, Ministry of Environment, Republic of Korea. Mid 17 18 19 7 and Long Term Health Effect of Hebei Spirit Oil Spill: First Year Survey Report. Taean: 20 21 8 Taean Environmental Health Center 2011.[Korean] 22 23 9 16. Kim JH, Kwak BK, Ha M, et al. Modeling human exposure levels to airborne volatile 24 25 10 organic compounds the by Hebei Spirit oil spill. Environ Health Toxicol 26 27 11 2012;27:e2011006. 28 29 30 12 17. Barendregt JJ, Oortmarssen GJ, Vos T, et al. A generic model for the assessment of 31 32 13 disease epidemiology: the computational basis of DisMod II. Popul Health Metr 33 34 14 2003;1:4-10.

35 http://bmjopen.bmj.com/ 36 15 18. Statistics Korea. Annual Report on the Cause of Death Statistics (2008). Seoul: Statistics 37 38 39 16 Korea 2009. 40 41 17 19. Public Health Division, Department of Human Services, Australia. Victorian Burden of 42

43 18 Disease Study: Morbidity. Melbourne: Public Health Division 1999. on October 1, 2021 by guest. Protected copyright. 44 45 19 20. Stoudhard MEA. Disability weights for diseases: a modified protocol and results for a 46 47 48 20 Western European region. Eur J Public Health 2000;10:24-30. 49 50 21 21. Lee JK, Yoon SJ, Do YK, Kwon YH, et al. Disability weights for diseases in Korea. 51 52 22 Korean J Prev Med 2003;36:1-26 (in Korean). 53 54 23 22. Yoon SJ, Bae SC, Lee SI, et al. Measuring the burden of disease in Korea. J Korean Med 55 56 57 22 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 23 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 Sci 2007;22:518-523. 5 6 2 23. Kim SY, Yoon SJ, Jo MW, et al. Economic burden of allergic rhinitis in Korea, Am J 7 8 Rhinol Allergy 2010;24:e110-e113. 9 3 10 11 4 24. Kim YM, Cheong HK, Kim J, et al. Scientific basis of environmental health contingency 12 13 5 planning for a coastal oil spill. J Prev Med Public Health 2009;42:73-81. [Korean] 14 15 6 25. Kim TB,For Kim YK, Changpeer YS, et al. review Association between onlysensitization to outdoor spider 16 17 7 mites and clinical manifestations of asthma and rhinitis in the general population of 18 19 20 8 adults. J Korean Med Sci 2006;21:247-52. 21 22 9 26. Ministry of Health and Welfare, College of Medicine, Seoul National University. The 23 24 10 Epidemiological Survey of Psychiatric Illnesses in Korea. Seoul: Ministry of Health and 25 26 11 Welfare 2006. [Korean] 27 28 27. Gonenc A, Oezkan Y, Torun M, et al. Plasma malondialdehyde (MDA) levels in breast 29 12 30 31 13 and lung cancer patients. J Clin Pharm Ther 2001;26:141-144. 32 33 14 28. Suzana S, Normah H, Fatimah A, et al. Antioxidant intake and status, and oxidative 34

35 15 stress in relation to breast cancer risk: a case-control study. Asian Pacific J Cancer Prev http://bmjopen.bmj.com/ 36 37 16 2008;9:343-350. 38 39 40 17 29. Tanaka H, Fujita N, Sugimoto R, et al. Hepatic oxidative DNA damage is associated with 41 42 18 increased risk for hepatocellular carcinoma in chronic hepatitis C. Br J Cancer

43 on October 1, 2021 by guest. Protected copyright. 44 19 2008;98:580-586. 45 46 20 47 48 49 50 51 52 53 54 55 56 57 23 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 24 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 Figures 5 6 7 8 2 Figure 1. Study area and classification of affected regions. 9 10 ★ 11 3 : accident site, about 8 km away from the Taean coastline. 12 13 4 14 15 For peer review only 16 5 Figure 2. Burdens of disease attributable to the Hebei Spirit oil spill by disease, age and sex. 17 18 19 6 YLD: years lived with disability; PTSD: post-traumatic stress disorder. 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 24 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 25 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 137x105mm (300 x 300 DPI) 34 35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 26 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 248x185mm (300 x 300 DPI) 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 27 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 28 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

43 on October 1, 2021 by guest. Protected copyright. 50-54 31.5 38.2 16.5 10.5 0.0 0.0 15.0 18.1 12.2 5.3 44 55-59 37.1 35.3 7.1 10.4 12.5 0.0 4.8 8.9 4.5 6.3 45 60-64 36.7 38.4 10.3 7.7 4.5 0.0 7.5 12.2 5.7 0.6 46 47 65-69 35.9 35.5 7.1 7.7 0.0 0.0 8.5 14.9 4.1 2.5 48 70-74 30.4 36.6 4.7 5.6 8.7 6.3 17.5 15.1 1.9 1.8 49 75+ 18.4 20.5 3.3 2.3 0.0 0.0 12.5 15.6 0.0 0.0 Depressio 19-44 29.5 27.9 6.5 15.6 7.4 20.0 12.9 28.6 6.7 25.9 50 n 51 45-49 42.9 33.0 20.0 21.7 0.0 25.0 21.4 21.1 8.1 20.0 52 50-54 23.3 29.3 14.7 15.0 0.0 7.7 23.1 27.1 7.8 14.6 53 55-59 24.8 26.7 8.8 10.4 20.0 7.7 15.6 14.3 7.8 11.2 54 60-64 24.8 29.8 9.3 11.8 10.0 9.5 18.0 22.5 4.9 8.2 55 65-69 19.4 30.3 5.6 18.6 12.5 20.8 26.5 25.9 3.8 8.7 56 70-74 22.8 29.6 6.3 14.4 10.0 14.3 27.7 30.6 4.8 10.0 57 75+ 15.5 35.2 12.4 13.7 0.0 6.7 21.7 39.4 2.4 9.6 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 29 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

43 on October 1, 2021 by guest. Protected copyright. 44 prevalence difference in dermatitis for elementary school student for 0-9 age group and that 45 46 for 13-18 years old for 10-19 age group to calculate the incidence in 2008. 47 48 Prevalence of conjunctivitis was 12.4% and 14.7% for elementary school student in highly 49 50 51 contaminated area and less contaminated area, respectively, and 18.7% and 15.2% for the 13- 52 53 18 years old in highly contaminated area and less contaminated area, respectively. 54 55 Prevalence of post-traumatic stress disorder (PTSD) was 5.1% and 1.8% for elementary 56 57 school student in highly contaminated area and less contaminated area, respectively, and 4.7% 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 30 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 31 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

- -

60.6 40.2 58.6 59.6 120.6 107.7 129.1 576.4 Female Female

- 94.9 33.3 Male 126.9 339.1 269.8 266.9 154.5 1,861.9 1285.4

- Mental Mental disease Depression a

b - 65.0 27.4 65.9 101.5 Male Female 328.8 316.8 197.4 321.5 160.0 238.3 234.1 197.7 335.4 279.4 1534.0 1335.2

0.1 1.4 1.2 1.8 2.2 1.1 1.6 9.4 0.0 Female

BMJ Open

http://bmjopen.bmj.com/

attributable to Hebei Spirit oil spill by disease, sex, age DALY) and (Unit: spill sex, disease, oil by Spirit to Hebei attributable

Allergy Rhinitis Dermatitis Conjunctivitis PTSD

For peer review only 88.2 137.3 724.6 599.9 738.1 589.4 323.3 149.2 3350.0 Female

Asthma Male Years lived with disability (YLDs) with disability lived Years

0~9 0~9 101.3 70+ 70+ 117.0 Age Sum 2787.7 a: post-traumatic stress disorder; b: Data available not b: Data stress a: disorder; post-traumatic Total Total 6,137.7 2,420.0 1,416.9 18.2 2,869.2 10~19 144.8 30~39 40~49 505.6 50~59 469.4 602.9 20~29 540.8 60~69 305.9 Table S4. Table 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 33 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

43 on October 1, 2021 by guest. Protected copyright. 44 18 polycyclic aromatic hydrocarbons (PAHs), post-traumatic stress disorder (PTSD), disability 45 46 19 weight 47 48 20 Word count: 49 21 Abstract: 2997 words 50 51 22 Main text: 33562 words 52 23 No. of tables: 3 53 24 No. of Figures: 2 54 25 No. of references: 29 55 56 57 1 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 34 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 ABSTRACT 5 6 2 Objectives: We aimed at assessing the burden of disease (BOD) of the residents living in 7 8 contaminated coastal area with oil spill and also analyzed the BOD attributable to the oil spill 9 3 10 11 4 by disease, age, sex, and sub-region. 12 13 5 Design: Health impact assessment by measuring years lived with disability (YLD) due to an 14 15 6 oil spill. For peer review only 16 17 18 7 Setting: A whole population of a community affected by an anthropogenic environmental 19 20 8 disaster and secondary health outcome data. 21 22 23 9 Participants: Based onFor the health outcome survey, including upon 10,171 individuals 24 25 10 (male: 4,354; female: 5,817) participated, BOD of covering 66,473 populations (male: 33,441; 26 27 11 female: 33,032) was measured living in the contaminated coastal area. 28 29 12 Interventions: None. Observational study on the effect of a specific environmental health 30 31 32 13 hazard. 33 34 14 Primary and secondary outcome measures: Using disability adjusted life year

35 http://bmjopen.bmj.com/ 36 15 (DALY)YLD method, BOD including both physical and mental diseases was measured. For 37 38 16 the BOYLD measurement, excess incidences of illnesses related to oil spill were estimated 39 40 17 from the comparison of prevalence of the health outcomes between contaminated areas and 41 42 18 reference area without contamination.

43 on October 1, 2021 by guest. Protected copyright. 44 45 19 Results: YLD attributable to the oil spill were estimated to be 14,724 DALYs (male: 7,425 46 47 20 DALYs; female: 7,299 DALYs) for the year 2008. The YLD of mental diseases including 48 49 21 post-traumatic stress disorder (PTSD) and depression for males were higher than that for 50 51 52 22 females. The YLD for females was higher in asthma and allergies (rhinitis, dermatitis, 53 54 23 conjunctivitis) than that for males. The effects of asthma and allergies were the greatest for 55 56 57 2 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 35 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 people in their 40s, with the burden of mental illness being the greatest for those in their 20s. 5 6 2 Proximity to the spill site was associated with increased burden of disease. 7 8 Conclusions: An oil spill near a coastline can cause substantial adverse health effects. As the 9 3 10 11 4 health effects of hazardous pollutants from oil spills are long-lasting, close follow-up studies 12 13 5 are required to identify chronic health effects. 14 15 6 Trial registration:For Not applicablepeer review only 16 17 18 7 19 20 21 8 INTRODUCTION 22 23 9 On December 7, 2007, the Hebei Spirit oil tanker spilled 12,547 kl (10,900 tons) of crude 24 25 26 10 oil into the Yellow Sea, approximately 8 km away from the coast of Taean, Korea. This was 27 28 11 the largest oil spill in the history of Korea. Spilled oil contaminated over 1,000 km of the 29 30 12 western coast of Korea including the Taean Coast National Park.[1] Crude oil contains 31 32 13 various volatile organic compounds (VOCs) such as benzene, toluene, ethylbenzene, xylenes, 33 34 14 and polycyclic aromatic hydrocarbons (PAHs),[2] all of which are potential air-borne toxins. 35 http://bmjopen.bmj.com/ 36 37 15 Certain VOCs, benzene in particular, are carcinogenic to humans and are associated with 38 39 16 haematologic cancer.[3] Previous studies [4-11] have reported that oil spills have caused 40 41 17 adverse health effects for residents of contaminated areas and clean-up workers. Risk 42

43 on October 1, 2021 by guest. Protected copyright. 18 analyses revealed that more frequent clean-ups and greater exposure to oil are strongly 44 45 46 19 associated with higher frequencies of psychological symptoms. [8-11] Rodriguez-Trigo [6] 47 48 20 reported that participation in a major oil spill clean-up is associated with chromosomal 49 50 21 damage, although they were unable to determine whether oil exposure itself caused the 51 52 22 abnormalities. 53 54 55 56 57 3 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 36 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 After the Hebei Spirit oil spill, more than 2,000,000 people (person-days), including over 5 6 2 1,200,000 person-days by volunteers, participated in clean-up activities during the first seven 7 8 months.[1] Among them were residents of Taean, a highly exposed group that experienced 9 3 10 11 4 adverse health effects, including both physical and psychological symptoms.[4, 12] Although 12 13 5 many studies have reported positive relationships between oil spills and acute and chronic 14 15 6 health impacts,For there have peer been no studies review quantifying the burden only of disease (BOD) due to oil 16 17 7 spills, which are necessary to assess the scale of health damage at the population level as well 18 19 20 8 as the associated compensation costs. 21 22 9 Therefore, we sought to quantify the adverse health impacts of exposure to oil spills by 23 24 10 applying the disability-adjusted life-year (DALY) measure developed by the World Health 25 26 11 Organization (WHO). The DALY measures health gaps as opposed to health expectancies. It 27 28 measures the difference between a current situation and an ideal situation where everyone 29 12 30 31 13 lives up to the age of the standard life expectancy, and in perfect health. The DALY is based 32 33 14 on the premise that the best approach for measuring the burden of disease is to use units of 34

35 15 time. [13] The DALY method has been used to measure the BOD attributable to specific risk http://bmjopen.bmj.com/ 36 37 16 factors including environmental burdens of disease (EBD). This approach can be practical in 38 39 40 17 that the estimated adverse health burden attributable to a risk factor can be compared with 41 42 18 other BODs and the analyses of disease burdens by disease and region can inform policy

43 on October 1, 2021 by guest. Protected copyright. 44 19 prioritization to mitigate adverse health effects for specific exposed populations. 45 46 20 The purposes of this study are to assess the BOD of the residents of the contaminated 47 48 Taean coastal area due to the Hebei Spirit oil spill and to analyze the BOD by disease, age, 49 21 50 51 22 sex, and sub-region. 52 53 23 54 55 56 57 4 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 37 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 METHODS 5 6 7 2 Study area and design 8 9 3 In September 2008, the Taean Environmental Health Center (TEHC) was established 10 11 under the support of the Korean Ministry of Environment to monitor the effects of the Hebei 12 4 13 14 5 Spirit oil spill on the health of the residents of the Taean coastal area. The TEHC first 15 For peer review only 16 6 conducted health surveys on the residents, including the referent population, one and a half 17 18 7 years after the spill, establishing a cohort. Total of 10,171 residents living in the study area 19 20 8 were surveyed including 9,246 adults (male: 3,849; female: 5,397) and 925 school students 21 22 23 9 (male: 505; female: 420 persons). The survey included a comprehensive questionnaire 24 25 10 gathering personal history on exposure and medical problems, and a structured questionnaire 26 27 11 assessing psychological health and asthma, and physical and laboratory examinations of 28 29 12 respiratory, cardiovascular, neurologic, and psychological systems. 30 31 The TEHC published its first report in 2010 after analysing the prolonged health effects 32 13 33 34 14 for 1.5 years, and demonstrated significant relationships between exposure concentrations

35 http://bmjopen.bmj.com/ 36 15 and adverse health outcomes.[15] To estimate the BOD caused by the Hebei Spirit oil spill, 37 38 16 we obtained the survey data from the TEHC, which included disease, residence, and 39 40 17 demographic information. Residents in Taean County and some islands in neighboring 41 42 18 Boryeong County were included in the survey. The Boryeong islands were expected to be

43 on October 1, 2021 by guest. Protected copyright. 44 45 19 affected by the oil spill because the islands are located in the affected area based on the wind 46 47 20 direction and tidal current at the time of the spill.[16] Kim et al.[16] reported that the VOCs 48 49 21 moved in a southeasterly direction, following the coastline from the accident point. 50 51 We classified the residents of Taean County and the Boryeong islands into five regions 52 22 53 54 23 based on both distance from the contaminated coastline and contamination level. Area 1 is 55 56 57 5 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 38 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 less than 1.1 km away from the coastline adjacent to the accident location where the coastline 5 6 2 was contaminated at a high level. Area 2 is 1.1-4.2 km away with no coastline, but was 7 8 directly affected by atmospheric VOCs and PAHs due to the wind direction. Area 3 is 4.2-23 9 3 10 11 4 km away from the coastline with a moderately contaminated coastline, and Area 4 is more 12 13 5 than 23 km from the contaminated coastline. We classified the Boryeong Islands as a separate 14 15 6 area. Figure For1 shows the peerlocations of the reviewstudy areas and five classifiedonly regions. 16 17 7 Figure 1 18 19 20 8 21 22 9 We assessed health outcome data regarding asthma, allergy (rhinitis, dermatitis, 23 24 10 conjunctivitis), and mental disorders (post-traumatic stress disorder (PTSD) and depression) 25 26 11 that were well-documented by the TEHC.[15] We focused on subchronic (duration of a few 27 28 months) and chronic (over one year duration) adverse health effects of the Hebei Spirit oil 29 12 30 31 13 spill, assessing prolonged health effects over a period of 1.5 years to estimate the BOD of the 32 33 14 Hebei Spirit oil spill for the year of 2008. 34

35 15 http://bmjopen.bmj.com/ 36 37 16 Estimation of burden of disease attributable to oil spill 38 39 40 17 To estimate the BOD attributable to the oil spill, we applied the DALY method. DALY is 41 42 18 the sum of the life years lost due to disability and premature death. [13] As there was no

43 on October 1, 2021 by guest. Protected copyright. 44 19 premature death identified as the direct effect of the oil spill until the TEHC published the 45 46 20 first report in 2010, we included only years lived with disability (YLD). [13] The YLD is 47 48 calculated by using disease incidence and prevalence, onset age and duration of disability, 49 21 50 51 22 and disability weight (DW). The formula we utilized is as follows; 52 53 23 54 55 56 57 6 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 39 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 5 1 1 6 7 8 1 1 [1] 9 10 2 where, D is disability weight; r is discount rate; a is age of onset of the disability; L is 11 12 13 3 duration of the disability; β is parameter from the age weighting function; and C is constant. 14 15 For peer review only 16 4 We used 3% of discount rate and 0.04 and 0.0618 for β and C, respectively, as suggested by 17 18 19 5 the WHO.[13] 20 21 22 6 Because there was no prevalence data recorded prior to the accident, we estimated excess 23 24 7 incidences using prevalence data from the TEHC under the following assumptions. We first 25 26 8 assumed that residents living in Area 4 had no negative health consequences due to the oil 27 28 9 spill and used this as the baseline prevalence. Area 4 was assumed to be the reference area, 29 30 not only because the area might not be impacted by the oil spill directly as it is far from the 31 10 32 33 11 contaminated coastline, but also because residents in Area 4 rarely participate in the clean-up 34

35 12 works. Very few people participated the clean-up work except for civil servants living in the http://bmjopen.bmj.com/ 36 37 13 area 4, which appeared to be an important factor to cause the difference in the prevalence of 38 39 14 disease between area 4 and others.[15] Secondly, we assumed that the prevalence of oil spill- 40 41 42 15 related diseases in the whole study area had been same prior to the accident. We then

43 on October 1, 2021 by guest. Protected copyright. 44 16 estimated excess incidences using prevalence differences between Area 4 and the other areas 45 46 17 stratified by area, age, sex, and disease using following formula. 47 48 18 [2] 49 ∑ 50 here, is the excess incidence of region i and disease j; is the population of 51 19 52 53 region i; is the prevalence of region i and disease j; is the prevalence of control 20 54 55 21 region and disease j. 56 57 7 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 40 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 The duration of disability and the average onset age were calculated using the DISMOD 5 6 2 II model [17] for PTSD, rhinitis, dermatitis, and conjunctivitis based on prevalence data from 7 8 the TEHC and the mortality rate of South Korea [18] by age, sex, and disease. For asthma, 9 3 10 11 4 the results of the Victoria, Australia study [19] were used. 12 13 5 DWs for each disease were measured following a method previously used by the Global 14 15 6 Burden of DiseaseFor group peer [13] and the review Dutch Disability Weights only Group.[20] A panel that 16 17 7 included 12 health and medicine specialists was convened. Eleven indicator conditions 18 19 20 8 including asthma, stomach cancer, diabetes, dementia, major depression, appendicitis, thyroid 21 22 9 cancer, hepatitis A, otitis media, atopic dermatitis, and common cold were measured using the 23 24 10 person trade-off (PTO) method. The eleven indicator conditions were selected in accordance 25 26 11 with a previous study on Korean disability weight. [21] That is, we considered that they 27 28 should be meaningful in public health and relatively popular diseases, while distributing the 29 12 30 31 13 whole DW range (0 to 1) evenly.[21] A facilitator first explained the method and then led the 32 33 14 discussion to determine the PTO value of each participant. The discussion was repeated twice. 34

35 15 Then, the visual analogue scale (VAS) method was applied to obtain DWs for 118 conditions http://bmjopen.bmj.com/ 36 37 16 including asthma, rhinitis, dermatitis, conjunctivitis, PTSD, and depression used in this study. 38 39 40 17 Inter-method reliability was confirmed between PTO and VAS using 11 indicator conditions, 41 42 18 which showed fairly high consistency (Spearman’s correlation coefficient 0.966, P-

43 on October 1, 2021 by guest. Protected copyright. 44 19 value<0.001). Reliability was confirmed by test-retest after three weeks by seven specialists 45 46 20 and obtained fairly high consistency (Spearman’s correlation coefficient 0.973, P- 47 48 value<0.001). 49 21 50 51 22 Finally, we estimated YLD caused by the Hebei Spirit oil spill utilizing the estimated 52 53 23 excess incidence attributable to the oil spill, DW, and the onset age and duration of disability 54 55 24 by disease, age, sex, and area. 56 57 8 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 41 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 5 6 7 2 RESULTS 8 9 3 10 11 Summary of the characteristics of the study area 12 4 13 14 5 The populations of Taean County and the Boryeong islands were 63,053 individuals 15 For peer review only 16 6 (male: 31,636; female: 31,417) and 3,420 individuals (male: 1,805; female: 1,615) in 2008, 17 18 7 respectively. Broken down by area, 8,348 (13%), 11,272 (17%), 4,597 (7%), 38,836 (58%), 19 20 8 and 3,420 (5%) residents lived in Areas 1, 2, 3, 4, and the Boryeong islands, respectively. 21 22 23 9 Area 4, the reference area, was the most populous because it includes a city center. The 24 25 10 largest age group consisted of subjects in their 50s, with individuals in their 40s, 50s, and 60s 26 27 11 comprising 45% of the total population (Ddata are presented in Datathe Ssupplement, Table 28 29 12 S1). 30 31 A total of 10,171 residents living in the study area participated in the TEHC survey. 32 13 33 34 14 Among them, 4,354 (42.8%) and 5,817 (57.2%) residents were male and female, respectively.

35 http://bmjopen.bmj.com/ 36 15 Individuals under 15 years of age, 15-65, and over 65 years of age comprised 7.2%, 48.0%, 37 38 16 and 44.8% of the population, respectively. Residents of Area 1 participated in clean-up work 39 40 17 for an average of 122 days, and residents of Areas 2, 3, and the Boryeong islands participated 41 42 18 for averages of 50, 36, and 100 days, respectively.[14] Table 1 shows the prevalence of

43 on October 1, 2021 by guest. Protected copyright. 44 45 19 disease related to the current study by disease, sex, and area derived from a survey of 9,246 46 47 20 adult (over 19 years old) residents. For young people under 19 years old, we used results 48 49 21 from 925 school students (male: 505; female: 420) living in the study area. More details 50 51 including prevalence by age are shown in Data Supplement, Table S2. 52 22 53 54 55 56 57 9 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from Page 42 of 56 42.6 10.1 40.2 17.0 15.0 24.5 10.8 21.3 13.5 25.2 23.8 27.9 Male Female Boryeong Boryeong Islands

Area 4 2.4 9.2 4.4 10.0 4.6 5.3 3.2 11.6 8.7 7.6 20.2 15.9 Male Female

3.1 10 10 Area 3

BMJ Open 9.4 4.7 2.9 8.8 14.0 9.6 10.4 23.9 10.6 13.2 9.0 Male Female

http://bmjopen.bmj.com/

7.1 Area 2

5.1 6.1 7.0 8.8 14.8 25.2 20.2 12.6 12.6 15.4 11.7 Male Female on October 1, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml

a) For peer review only Area 1 22.0 20.4 32.9 34.6 Male Male Female . Prevalence of diseases related to this study (in study %). this related to . of diseases Prevalence b) Rhinitis 41.4 34.6 Dermatitis Conjunctivitis 19.6 32.0 25.0 38.7 Depression 23.3 29.9 PTSD Table 1 Table a) Area 1: less than 1.1 km away; Area 2: 1.1-4.2 km away; Area 3: 4.2-23 km away; Area 4: more than 23 km from the contaminated contaminated the km from 23 than 4: more Area away; km 3: 4.2-23 Area away; km 2: 1.1-4.2 Area away; km 1.1 1: less than Area a) Allergy Allergy Asthma Disease coastline; b) PTSD: posttraumatic stress stress disorder. coastline; b) posttraumatic PTSD: Mental Mental disease 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 43 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

29 12 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

Total Total

989 1,099 962 1,014 Male Female 1,370 956 1,300 1,381 1,855 1,770 1,365 1,627

76 76 106 113 113 190 116 116 138 201 201 149 353 353 354 222 222 322 Male Female Boryeong Boryeong Islands

0 Area 3 Area 3 12 12 c

0 65 65 57 53 53 97 94 94 32 209 207 101 183 Male Male Female BMJ Open

http://bmjopen.bmj.com/

Area 2

266 169 133 222 453 481 241 231 294 214 146 158 Male Female

a For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Area 1

839 839 729 473 473 449 838 838 581

Table 2. 2. Table a: Area 1, less than 1.1 km away; Area 2, 1.1-4.2 km away; Area 3, 4.2-23 km away; Area 4, more than 23 km from the contaminated coastline; coastline; contaminated the km from 23 than more 4, away; Area km 4.2-23 3, away; Area km 1.1-4.2 2, Area away; than km 1.1 less 1, a: Area area. and reference 3 Area between c: difference stress disorder; b:No post-traumatic Allergy Allergy Asthma Disease Depression PTSD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 45 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

28 weight Asthma 29 Mild intermittent 0.222 30 Mild persistent 0.291 31 Moderate persistent 0.510 32 Severe persistent 0.715 33 34 Very severe persistent 0.836

35 Allergy Rhinitis 0.205 http://bmjopen.bmj.com/ 36 Dermatitis 0.132 37 38 Conjunctivitis 0.212

39 a 40 PTSD 0.533 41 Depression 0.219 42 a: post-traumatic stress disorder 43 11 on October 1, 2021 by guest. Protected copyright. 44 45 12 46 47 13 Burden of disease due to oil spill 48 49 Figure 2 presents the YLD according to disease, sex, and age group. The YLD for 50 14 51 52 15 asthma, allergic rhinitis, dermatitis, conjunctivitis, PTSD, and depression were approximately 53 54 16 6,138 DALYs, 2,420 DALYs, 1,417 DALYs, 18 DALYs, 2,869 DALYs, and 1,862 DALYs, 55 56 57 13 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 46 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 respectively. The total YLD attributable to the Hebei Spirit oil spill for the year 2008 was 5 6 2 14,724 DALYs. The YLD for males was 7,425 DALYs and for females was 7,290 DALYs. 7 8 The YLD for males for PTSD and depression were higher than for females, and the YLD for 9 3 10 11 4 females was higher in asthma and allergies than for males (More details are shown in Ddata 12 13 5 DSsupplement, Table S42). Among the included diseases, the YLD for asthma was the 14 15 6 highest. TheFor YLD for asthma peer and allergies review was the highest for only residents in their 40s, and the 16 17 7 YLD for mental disease was the highest for residents in their 20s. 18 19 20 8 Figure 2 21 22 9 23 24 10 DISCUSSION 25 26 27 11 28 29 12 The total YLD attributable to the Hebei Spirit oil spill was estimated at 14,724 DALYs 30 31 for 2008. Asthma was found to represent the most prominent disease burden (6,138 DALYs) 32 13 33 34 14 among the six diseases, followed by PTSD and rhinitis. The asthma burden of 6,138 DALYs

43 on October 1, 2021 by guest. Protected copyright. 44 45 19 as having allergic rhinitis. The direct cost was $223.768 million and the lost productivity was 46 47 20 estimated to be $49.325 million, for a total economic burden of $272.9 million in 2007. 48 49 21 Applying this method to the current study, with an excess incidence of 3,625 persons, the cost 50 51 of additional allergic rhinitis due to the oil spill is roughly estimated to be $243,170 for one 52 22 53 54 23 year (2008). These results indicate that oil spills near coastlines can cause considerable 55 56 57 14 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 47 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

35 15 health outcomes. http://bmjopen.bmj.com/ 36 37 16 To estimate the excess incidences of illnesses caused by an oil spill, we used prevalence 38 39 40 17 data classified by distance from the contaminated coast, under the assumption that the area 41 42 18 farthest regiondistant, Area 4, was not affected by the oil spill, and that the prevalence of

43 on October 1, 2021 by guest. Protected copyright. 44 19 diseases had been samewas similar in all study areas before the accident, which is one of 45 46 20 limitations of this study. These assumptions were used because real data describing disease 47 48 prevalence before and after the oil spill were not available. Residents in Area 4, the reference 49 21 50 51 22 area, had very little participation in the clean-up efforts, which could account for differences 52 53 23 in disease prevalence between Area 4 and other areas. In addition, seventy-two hours after the 54 55 24 oil spill, the downwind direction changed from the southeast to the west and as a result, 56 57 15 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 48 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 volatile components did not move further inland.[16] The benzene concentrations in the 5 6 2 atmosphere after the Hebei Spirit oil spill simulated in our previous study [24] also showed 7 8 that concentrations reduced rapidly with distance from the contaminated coastline. Benzene 9 3 10 11 4 concentrations over 0.1 ppm were not detected at the area more than 13 km from the spill site. 12 13 5 Therefore, the assumption that Area 4 was not affected by the oil spill is reasonable. Kim et 14 15 6 al.[25] reportedFor that the peerprevalence of rhinitisreview in Korean population only was 16.4%, 24.7%, and 16 17 7 21.7% in 20-35, 36-50, and older than 50 age group, respectively, while that of Area 4 adults 18 19 20 8 was 15.8% and 17.9% in 19-49 and over 50 age group, respectively, showing slightly lower 21 22 9 prevalence than whole Korean population. 23 24 10 It is possible that PTSD and depression may overlap. According to a previous cross- 25 26 11 sectional study of psychological disease in Korea, about 8.5% of depression is accompanied 27 28 by PTSD.[26] This indicates that the estimated disease burden may have been overestimated. 29 12 30 31 13 For this reason, we did not include anxiety in our analyses, although prevalence data for 32 33 14 anxiety were available because about 15% of anxiety is accompanied by depression in 34

43 on October 1, 2021 by guest. Protected copyright. 44 19 (8-OHdG) levels of residents in the contaminated area, as oxidative stress indices and the 45 46 20 TEHC found that the urinary MDA and 8-OHdG concentrations of the residents living in 47 48 Area 1 were much higher than in other areas. The average urinary MDA concentration of 49 21 50 51 22 residents in Area 1 was 5.42 (SD=3.14) µmol/g Cr, whereas that of other areas was 3.83 52 53 23 (SD=2.58) µmol/g Cr.[15] The average 8-OHdG concentration for Area 1 was 6.84 (SD=3.84) 54 55 24 µg/g Cr, while that of other areas was 4.67 (SD=3.00) µg/g Cr. Previous studies reported that 56 57 16 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 49 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 oxidative DNA damage is associated with increased risk of cancer, including haepatocellular 5 6 2 carcinoma, lung cancer, and breast cancer, and thus the MDA and haepatic 8-OHdG levels are 7 8 useful as markers to identify subjects at extremely high-risk.[27-29] The increased urinary 9 3 10 11 4 MDA and 8-OHdG concentrations in the residents in this study area may be useful to track 12 13 5 the risks for excess cause cancers in the future. By monitoring more long-term effects, the 14 15 6 BOD imposedFor by these diseasespeer may be quantifiedreview and added toonly the results of this study. 16 17 7 This is the first study to quantify the BOD of a single environmental disaster. The EBD 18 19 20 8 of the current study was estimated for a highly exposed population after a single event. This 21 22 9 study demonstrates that BOD is an objective and comprehensive metric for estimating and 23 24 10 comparing the health effects of environmental hazards and disasters across different regions 25 26 11 and time periods. One of strengths of this study is that we measured BOD attributable to oil 27 28 spill by sub-region as well as age and sex including both physical and mental health 29 12 30 31 13 outcomes, which can helpcontribute to prioritize the health policies and mitigation 32 33 14 effortspolicy to reduce adverse health effects. 34

43 on October 1, 2021 by guest. Protected copyright. 44 45 19 proximity to the spill as well as participation in clean-up efforts. Asthma and PTSD were 46 47 20 found to comprise the most prominent disease burden in the contaminated areas. The health 48 49 21 impacts of oil spills on residents vary with age, sex, and region implying that any policies to 50 51 reduce the excess BOD should be community-specificrehabilitation policy should be 52 22 53 54 23 community-specific. 55 56 57 17 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 50 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

43 on October 1, 2021 by guest. Protected copyright. 44 26 ▶ This is the first study to quantify the BOD of a single environmental disaster. 45 46 27 ▶ This study demonstrates that BOD is an objective and comprehensive metric for 47 28 estimating and comparing the health effects of environmental hazards and 48 29 disasters across different regions and time periods. 49 50 30 ▶To estimate excess incidence, we used prevalence data classified by distance 51 31 from the contaminated coast, under the assumption that the area farthest 52 32 region distant, Area 4, haswas not been affected by the oil spill, and that the 53 33 prevalence of diseases had beenwas sameimilar across thein wholeall study 54 55 56 57 18 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 51 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

1 2 3 4 1 areas before the accident because real prevalence data recorded prior to 5 2 describing disease prevalence before and after the oil spill were not available. 6 7 3 ▶ We did not consider long-term health outcomes such as cancers and chronic 8 4 degenerative diseases mainly because of lack of data. 9 5 10 11 12 13 14 15 For peer review only 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 19 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 52 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

29 12 30 31 13 References 32 14 33 15 1. County chief of Taean. Taean Tonggyeyeongam, No.18. Taean: Planning and Inspection 34

43 on October 1, 2021 by guest. Protected copyright. 44 2011;45:7481-7488. 45 20 46 47 21 3. World Health Organization, International Agency for Research on Cancer. IARC 48 49 22 Monographs on the Evaluation of Carcinogenic Risks to Humans overall Evaluations of 50 51 23 Carcinogenicity: An Updating of IARC Monographs Volumes 1 to 42 (Supplement 7). 52 53 24 Lyon: International Agency for Research on Cancer 1998. 54 55 56 57 20 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 53 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

43 18 after the Exxon Valdez oil spill. Am J Psychiatr 1993;150:1517-1523. on October 1, 2021 by guest. Protected copyright. 44 45 19 11. Palinkas LA, Petterson JS, Russell JC, et al. Ethnic differences in symptoms of 46 47 posttraumatic stress after the Exxon Valdez oil spill. Prehosp Disaster Med 2004;19:102- 48 20 49 50 21 112. 51 52 22 12. Lee CH, Kang YA, Chang KJ, et al. Acute health effects of the Hebei oil spill on the 53 54 23 residents of Taean, Korea. J Prev Med Public Health 2010;43:166-73. [Korean] 55 56 57 21 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 54 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

43 18 Disease Study: Morbidity. Melbourne: Public Health Division 1999. on October 1, 2021 by guest. Protected copyright. 44 45 19 20. Stoudhard MEA. Disability weights for diseases: a modified protocol and results for a 46 47 48 20 Western European region. Eur J Public Health 2000;10:24-30. 49 50 21 21. Lee JK, Yoon SJ, Do YK, Kwon YH, et al. Disability weights for diseases in Korea. 51 52 22 Korean J Prev Med 2003;36:1-26 (in Korean). 53 54 23 22. Yoon SJ, Bae SC, Lee SI, et al. Measuring the burden of disease in Korea. J Korean Med 55 56 57 22 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 55 of 56 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

43 on October 1, 2021 by guest. Protected copyright. 44 19 2008;98:580-586. 45 46 20 47 48 49 50 51 52 53 54 55 56 57 23 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml BMJ Open Page 56 of 56 BMJ Open: first published as 10.1136/bmjopen-2013-003334 on 20 September 2013. Downloaded from

35 http://bmjopen.bmj.com/ 36 37 38 39 40 41 42

43 on October 1, 2021 by guest. Protected copyright. 44 45 46 47 48 49 50 51 52 53 54 55 56 57 24 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml