In April 2010, 11 people were killed in an explosion on the Deepwater Horizon drilling rig in the Gulf of Mexico. The explosion also caused millions of barrels of oil to flow into the Gulf – the largest accidental marine oil spill in history. In addition to its environmental and economic impacts,Reference Kostka, Prakash and Overholt 1 , Reference Sumaila, Cisneros-Montemayor and Dyck 2 some studies conducted soon after the Deepwater Horizon oil spill (DHOS) suggest that the spill had psychological effects in the regional population.Reference Grattan, Roberts and Mahan 3 , Reference Gill, Picou and Ritchie 4 Consistent with previous technological and natural disasters, increased exposure to the DHOS appears to be related to increased depression and anxiety among Gulf residents.Reference Grattan, Roberts and Mahan 3 – Reference Rung, Oral and Fontham 10 The association between DHOS exposure and mental health within 1 year of the spill has been established for populations both directly and indirectly exposed.Reference Grattan, Roberts and Mahan 3 , Reference Gill, Picou and Ritchie 4 , Reference Fan, Prescott and Zhao 9 Of note, a study that followed Gulf Coast residents from June 2010 (when oil was still flowing) through April 2011 found that psychological symptoms decreased over the study period.Reference Cope, Slack, Blanchard and Lee 11 This is consistent with other literature suggesting that most disaster- and other trauma-exposed individuals recover from immediate psychological effects, though some people may be more susceptible to sustained or late-onset mental health problems.Reference Ayer, Setodji and Schultz 12 – Reference North and Pfefferbaum 15 In the study with the longest follow-up period following DHOS, to our knowledge, DHOS exposure was associated with depression, domestic conflict, and general mental distress in a sample of over 2,000 women living in coastal Louisiana 2, 4, and 6 years after the spill.Reference Rung, Gaston and Oral 8 , Reference Rung, Oral and Fontham 10 These more vulnerable individuals with chronic symptoms may need additional support and services in the years following a disaster. Furthermore, the types of symptoms that persist or develop in the years following a disaster and the risk factors for chronic or late-onset symptoms may be different from those identified in studies focused on more immediate outcomes.
In addition to a need for more data on longer-term mental health symptoms following the DHOS, there are some key mental/behavioral health aspects that deserve more attention. One important mental health dimension not yet explored in relation to the DHOS is illness anxiety. Illness anxiety is characterized by excessive concern or worry about having or getting a serious illness. Technologic disasters and environmental crises have commonly triggered health scares,Reference Bromet and Havenaar 16 , Reference Princeembury and Rooney 17 heightened illness concern and related anxiety,Reference Burger and Gochfeld 18 and led to “outbreaks” of sociogenic physical symptoms and syndromes.Reference Bartholomew and Wessely 19 , Reference Ismail, Everitt and Blatchley 20 For some individuals, such worries may become significant enough to warrant clinical intervention or meet diagnostic criteria for an illness anxiety disorder.Reference Association 21 In a study conducted 6 years after the DHOS, exposed Gulf Coast residents reported persistent worry about the impacts (including health impacts) of the spill.Reference Parker, Finucane and Ayer 22 However, to our knowledge, there has not yet been an investigation into illness anxiety after DHOS.
Furthermore, the association between DHOS exposure and alcohol use has received little attention. There is a large body of literature showing that trauma exposure is a significant risk factor for alcohol abuse and dependence,Reference Choenni, Hammink and van de Mheen 23 – Reference Tomaka, Magoc, Morales-Monks and Reyes 26 in part because individuals may habitually use alcohol to cope with trauma-related stress and other negative emotions, and their use of alcohol can become problematic over time.Reference Lowe, Sampson, Young and Galea 25 , Reference Langdon, Fox and King 27 We are aware of only one study that has examined alcohol use patterns in Gulf Coast residents following the DHOS. Gould and colleaguesReference Gould, Teich and Pemberton 28 examined data from the National Survey on Drug Use and Health (NSDUH) conducted before the DHOS (2007-2009) and after the DHOS (2011) and found that increases of alcohol use in the previous month after the DHOS were greater among Gulf residents compared with the rest of the surveyed population. However, the authors noted that this finding could be attributed to many factors, because exposure to the DHOS was not measured in the NSDUH. To our knowledge, there are no published studies that have tested the association between DHOS exposure and alcohol use among Gulf Coast residents.
In addition to degree of exposure to the disaster itself, previous research shows that demographics (eg, low income, female gender) and the experience of other potentially traumatic eventsReference Goldmann and Galea 29 , Reference Neria, Nandi and Galea 30 increase risk for behavioral health problems after disasters like the DHOS. In general, individuals who have experienced a greater number of potentially traumatic events are at higher risk for depression, anxiety, alcohol use, and illness anxiety.Reference Weck, Neng, Goller and Muller-Marbach 31 – Reference Schumm, Briggs-Phillips and Hobfoll 33 Some DHOS studies have accounted for exposure to previous disasters (eg, Hurricane Katrina) in their examination of mental health impact.Reference Osofsky, Osofsky and Hansel 7 , Reference Blackmon, Lee and Cochran 34 Only 1 study has accounted for other types of trauma (eg, interpersonal trauma like physical or sexual assault), which are strongly related to subsequent mental health problems,Reference Cisler, Begle and Amstadter 35 , Reference Resnick, Acierno and Kilpatrick 36 in examining mental health following the DHOS. In that study, Cherry and colleagues (2017) tested whether lifetime trauma exposure was associated with coping style and psychological distress (posttraumatic stress disorder [PTSD] and depression) following DHOS exposure in 64 commercial fishers in southern Louisiana.Reference Cherry, Lyon and Sampson 37 They found that lifetime trauma was a significant predictor of some types of coping, but did not predict PTSD or depression above and beyond covariates in the models. However, this was a relatively small study of a unique population. The role of lifetime trauma in behavioral health among a larger, more representative population of residents exposed to the DHOS has not yet been examined, to our knowledge.
In the current study, we hypothesized that DHOS exposure would be related to depression, anxiety, alcohol use, and illness anxiety 5 years after the spill. We expected that associations between exposure and behavioral health would generally be weaker than those identified in studies examining more immediate impacts. We then tested whether individuals with a history of other potentially traumatic experiences and who reported higher levels of DHOS exposure were most likely to report mental health problems 5 years after the DHOS. We hypothesized that those with greater DHOS exposure and who had experienced a greater number of traumatic events would report more depression, anxiety, illness anxiety, and alcohol use problems compared with those with less DHOS and/or other trauma exposure.
METHODS
Sampling
The sample is a randomly selected, representative group of 2,520 adult residents of 56 counties located in the coastal areas along the Gulf of Mexico. This total includes both traditional landline telephone users (1,617 respondents) and cell phone users (903 respondents). The design of the landline sample ensured representation of both listed and unlisted numbers by the use of random digit dialing. The overall landline sample was drawn at the county-level and grouped into 11 regions across the 5 states. For the landline sample, an adult respondent was randomly selected from the total number of adults living in the household. The cell phone sample was randomly drawn from known, available phone number banks dedicated to wireless service in the region.
Procedure
All procedures were reviewed and approved by the RAND Corporation's institutional review board. Data were collected through telephone interviews conducted from April 22 to August 6, 2016. We made up to 8 attempts per landline phone number, and up to 4 attempts on each cell phone number. The calls were staggered over daytime versus evening, and weekday versus weekend (including at least 1 daytime call), to maximize the chances of making contact with a potential respondent. Interviewing was also spread as evenly as possible across the field period. An effort was made to re-contact most interview breakoffs and refusals to attempt to convert them to completed interviews, with up to 3 attempts per respondent. The response rate was 4.4% for the landline sample, 4.5% for the cell phone sample, and 4.5% overall. This response rate is the percentage of residential households or personal cell phones for which an interview was completed. The rates are calculated using the American Association for Public Opinion Research’s method for Response Rate 3 as published in their Standard Definitions. 38 , Reference de Heer 39 Response rates have declined steadily for all surveys over the past several decades,Reference de Heer 39 – Reference Tortora 41 and our response rate is slightly lower but generally consistent with a study of response rates recently conducted by the Pew Research Center.Reference Center 42
Measures
Demographic Characteristics
These included gender (male or female), age (years), race (white, black or African American, American Indian or Alaska Native, Asian, Native Hawaiian or other Pacific Islander), ethnicity (Hispanic/Latino or non-Hispanic/Latino), and highest education level (no high school diploma, high school diploma/GED, some college, associate’s degree, vocational/technical diploma, bachelor’s degree, graduate degree). The race and ethnicity variables were recoded due to small cell sizes into white non-Hispanic, black non-Hispanic, Hispanic, and other race/ethnicity.
Trauma History
Trauma history was assessed with twelve items where respondents reported on adulthood trauma, “since the age of 18,” adapted from the Trauma History Screen (THS).Reference Carlson, Smith and Palmieri 43 For example, “Since you were 18, were you in a really bad car, boat, train, or airplane accident?” The count of endorsed past traumatic experiences was used for these analyses (range=0-12).
Exposure to the DHOS and Its Effects
Twelve items were used to measure respondents’ recalled exposure to the oil spill and its effects, including occupational, economic, and lifestyle effects. Respondents indicated whether they or a family member worked in the oil and gas, fishing/seafood, or tourism-related industries at the time of the spill, as well as whether they worked on any shoreline or water cleanup activities. Respondents also indicated whether the spill had caused property loss or damage, financial loss, or loss of a job or hours at a job, and whether they had filed a claim as a result of the spill. Respondents were asked whether they or their family fish commercially and, if so, did the oil spill damage areas where they fish. Finally, respondents indicated whether the oil spill affected hunting, fishing, or gathering activities; exercise or recreational patterns; and dietary or eating patterns. The count of endorsed exposures and effects (out of 12) was used as a score (range=0-12). Scores of participants who reported that they were not residents of the Gulf Coast at the time of the DHOS were recoded to equal zero.
Alcohol Abuse
This was measured with the Alcohol Use Disorders Identification Test (AUDIT-C),Reference Bush, Kivlahan and McDonell 44 a 3-item self-report screening measure for hazardous drinking and alcohol use disorders. The AUDIT-C is an adapted version of the 10-item AUDIT and assesses drinking frequency and quantity. Items were scored on a 0-4 scale and summed for a total score (range=0-12).
Depression
Depression was measured using the Patient Health Questionnaire (PHQ-2),Reference Kroenke, Spitzer and Williams 45 a depression screener assessing the frequency with which respondents experienced a lack of interest/pleasure (ie, anhedonia) and depressed mood. Items were scored on a 0-3 scale and summed for a total score (range=0-6).
Anxiety
Anxiety was assessed with the Generalized Anxiety Disorder (GAD-2) screener.Reference Kroenke, Spitzer and Williams 46 The GAD-2 measures how often respondents experience nervousness, anxiety, and worry. Items were rated on a 0-3 scale and summed for a total score (range=0-6).
Illness anxiety
Respondents rated 6 items that asked about the extent to which they experienced illness worry or fear (eg, “Do you often worry about the possibility that you have a serious illness?”), using a shortened version of the Whiteley Index with more favorable psychometric properties.Reference Pilowsky 47 , Reference Welch, Carleton and Asmundson 48 Items were rated on a 0-4 scale and summed for a total score ranging from 0 to 24 (α=0.88).Reference Welch, Carleton and Asmundson 48
Weighting
Weights were created to make respondents more representative of the underlying population.Reference Heeringa, West and Berglund 49 – Reference Schafer and Graham 51 A base weight was computed to account for probability of selection into the study based on population density of the county and the number of adults in the household (for landline respondents). In the second step, the base weight was adjusted using raking,Reference Battaglia, Izrael, Hoglin and Frankel 52 an iterative procedure that matches ethnicity, race, education, household income, gender, and age to the known population profile for the specified region from the 2012-2016 American Community Survey 5-year estimates.Reference Heeringa, West and Berglund 49 , Reference Little and Rubin 50 , Reference Bureau 53 Although weighting cannot eliminate every source of nonresponse bias, our approach to draw a random sample and our rigorous efforts to contact sampled persons, combined with accepted weighting techniques, have a strong record of yielding unbiased results.
Analyses
First, we ran descriptive analyses to examine the means, standard deviations, and frequencies of key variables. Next, we ran a series of linear regressions to test whether trauma history and DHOS exposure are independently associated with the 4 mental health outcomes of interest (ie, alcohol abuse, depression, anxiety, and illness anxiety) after accounting for relevant demographic covariates. We then mean-centered the continuous trauma and DHOS exposure variables and computed an interaction term. This interaction term, along with trauma and DHOS exposure/main effects and demographics, was entered into 4 additional linear regressions to test whether trauma history exacerbated the association between DHOS exposure and each mental health variable.
RESULTS
Descriptives
Table 1 shows the demographic characteristics of the sample as well as mean levels of trauma, exposure, and mental health. Trauma history and DHOS exposure were significantly correlated (r=0.34, P<0.0001). The sample included similar proportions of males and females (unweighted: females=60%, males=40%), with just over half (58.5%) of respondents identifying as white (unweighted=66.1%). The mean age of the sample was about 48 years (unweighted=58 years), and the majority of respondents had at least a high school diploma. Respondents endorsed an average of almost 2 DHOS exposures or effects, and almost 4 potentially traumatic events in adulthood. Depression, anxiety, and alcohol abuse average scores were in the mild range, on average. The average illness anxiety score was relatively mild, about 5 on a 0-24 scale.
Sample Characteristics (N=2,520)
SE=standard error.
Main Effects of Trauma and DHOS Exposure on Mental Health
Table 2 displays the main effects of trauma and DHOS exposure on the 4 mental health outcomes, both separately (Models 1 and 2) and together (Model 3), after accounting for demographics in each case. In the first set of models, trauma and demographic covariates were independent variables predicting depression (F[12,2513]=8,019; P<0.0001; R2=0.11), alcohol abuse (F[12,2490]=9.23, P<0.0001, R2=0.11), anxiety (F[12,2506]=11.75, P<0.0001, R2=0.19), and illness anxiety (F[12,2518]=6.54, P<0.0001, R2=0.12). In the second set of models, DHOS exposure and demographic covariates were independent variables predicting depression (F[12,2513]= 8.42, P<0.0001, R2=0.08), alcohol abuse (F[12,2490]=7.80, P<0.0001, R2=0.10), anxiety (F[12,2506]=7.95, P<0.0001, R2=0.12), and illness anxiety (F[12,2518]=6.02, P<0.0001, R2=0.07). The third set of models includes both the main effects of DHOS exposure and trauma, along with demographic covariates, predicting depression (F[13,2513]=7.64, P<0.0001, R2=0.11), alcohol abuse (F[13,2490]=8,058; P<<0.0001; R2=0.11), anxiety (F[13,2506]=10.62, P<0.0001, R2=0.19), and illness anxiety (F[13,2518]=7.39, P<0.0001, R2=0.12). In general, individuals reporting a higher number of traumatic events also reported significantly more symptoms of depression, alcohol abuse, anxiety, and illness anxiety, after accounting for relevant demographic variables (Model 1). Greater DHOS exposure was associated with greater anxiety and illness anxiety but not with depression or alcohol abuse (Model 2). Once trauma history was accounted for, there was no significant association between DHOS exposure and depression, alcohol abuse, and anxiety (Model 3). The association between DHOS exposure and illness anxiety decreased in size (β changed from 0.49 to 0.24) but remained statistically significant (P=0.03). Re-running these analyses with a trauma variable that excluded the 1 item about exposure to disasters (ie, “Since you were 18, were you in a hurricane, flood, earthquake, tornado, or fire?”) did not meaningfully alter the results.
Trauma and DHOS Exposure Main Effects on Mental Health Variables
DHOS=Deepwater Horizon oil spill; SE=standard error.
*Models control for gender, age, education, and race/ethnicity; † P<0.001; ‡ P<0.05.
Interaction Effects
Next, we tested the interaction between trauma history and DHOS exposure predicting the 4 mental health outcomes after accounting for these variables’ main effects as well as age, gender, education, and race/ethnicity. The interaction between trauma and DHOS exposure did not significantly predict any of the 4 outcomes – depression: B(SE)=-0.02(0.01), β=-0.03, ns; alcohol abuse: B(SE)=0.00(0.02), β=0.02, ns; anxiety: B(SE)=0.01(0.02), β=0.03, ns; illness anxiety: B(SE)=0.04(0.05), β=0.04, ns.
DISCUSSION
In this study, we examined the associations between degree of self-reported DHOS exposure, trauma history, and mental health outcomes that included depression, anxiety, illness anxiety, and alcohol use 6 years after the oil spill. Whereas previous studies on DHOS exposure and mental health have found that greater exposure was associated with elevated depression and anxiety symptoms, we found that exposure was related to anxiety and illness anxiety only after accounting for relevant demographic variables. However, once trauma history was accounted for, the association between DHOS exposure and anxiety became nonsignificant; DHOS exposure remained significantly related to illness anxiety but did not predict depression, anxiety, or alcohol use above and beyond the effects of trauma history, which remained a significant predictor of all 4 outcomes. These findings suggest that other traumatic experiences are more strongly related to mental health compared with DHOS exposure, and that these other traumatic experiences may be driving previously identified relationships between DHOS exposure and mental health. Another explanation, however, is that our measure of DHOS exposure does not adequately capture the most traumatic or psychologically impactful aspects of oil spill exposure. Future studies on the mental health consequences of oil spills and other disasters could include additional exposure items that are more directly related to mental health, such as the degree of fear and anger experienced during the event.
We also found that DHOS exposure and a number of traumatic experiences were significantly correlated, suggesting that Gulf Coast residents who were more exposed to DHOS were also more likely to have experienced other potentially traumatic events. Thus, those most affected by the DHOS are a particularly vulnerable group as a result of other experiences.
This was the first study, to our knowledge, to examine symptoms of illness anxiety in individuals exposed to the DHOS. Before accounting for trauma history, we found that greater DHOS exposure was associated with elevated illness anxiety in our sample, after controlling for demographics and trauma exposure. Our findings suggest that exposure to an environmental disaster may have sustained effects on anxiety about health (ie, illness anxiety) but not other more general behavioral health symptoms (ie, depression, anxiety, and alcohol abuse). Elevated levels of illness anxiety are associated with increased service utilization, as affected individuals search for medical reassurance and information.Reference Lee, Creed, Ma and Leung 54 , Reference Tomenson, McBeth and Chew-Graham 55 However, particularly when illness anxiety is associated with persistent somatic symptoms (eg, pain and fatigue), resulting service use may expose those affected to unnecessary testing, ineffective treatments, and harmful side effects.Reference Hollifield, Paine, Tuttle and Kellner 56 It may also create strain on providers in resource-poor environments, particularly at times immediately after disasters when demand may be elevated.
We hypothesized that DHOS exposure and trauma history would interact such that those with high exposure to both would report more symptoms compared to those with less exposure. However, the interactions we tested were nonsignificant. In light of the previously discussed findings on the main effects of these variables, it appears that years after the oil spill, the degree of DHOS exposure may not be as relevant to Gulf Coast residents’ behavioral health compared to other traumatic events they may have experienced. Moving forward, this research suggests that the degree of trauma exposure may be a better indicator of vulnerability to the long-term mental health impacts of an oil spill disaster than the degree of exposure to the oil spill itself.
Our study has some limitations that must be considered. This was a cross-sectional study conducted 6 years after the DHOS, and thus we can make assumptions only about “baseline” mental health problems or problems in the year immediately following the spill based on previous studies. We were not able to examine change in symptoms over time within our sample or causal associations. Further, we relied on self-reports of DHOS exposure that may have become less reliable over time.Reference Coughlin 57 We also did not assess childhood trauma exposure. There is a great deal of evidence that exposure to traumatic events in childhood is particularly detrimental to mental health,Reference Edwards, Holden, Felitti and Anda 58 , Reference Horwitz, Widom, McLaughlin and White 59 so it is possible that the effects of trauma on mental health that we identified in this study may have been stronger if childhood events had been accounted for. We also were not able to account for participation in behavioral health treatment (medication or behavioral) or for substance use, which can affect behavioral health symptoms.
CONCLUSIONS
Residents of the Gulf Coast have been exposed to an inordinate number of disasters in recent decades, including hurricanes, flooding, and oil spills. It is reasonable to expect more of such disasters in the future, and policy-makers and health care providers in the area are preparing for such events. 60 The present study emphasizes the importance of measuring and considering not only the index event in disaster-related mental health work, but also other potentially traumatic exposures. Further, anxiety about health problems (ie, illness anxiety) is a relatively understudied phenomenon that is worthy of additional exploration in future work.
Acknowledgments
The authors would like to thank the Gulf Coast residents who participated in this research.
Funding
This research was made possible by a grant from The Gulf of Mexico Research Initiative [231501-00]. Data are publicly available through the Gulf of Mexico Research Initiative Information and Data Cooperative (GRIIDC) at https://data.gulfresearchinitiative.org.
