The 2004 Florida hurricane season was unprecedented. Within a period of 7 weeks, 4 hurricanes and 1 tropical storm made landfall.Reference Acierno, Ruggiero and Galea1, 2 The $4.85 billion in costs incurred for hurricane relief accounted for nearly 88% of the total disaster aid in 2004.3 State and local public health workers played a critical role as first responders in this and other disasters.
Concern for public health response to natural disasters has increased during the past decade in the aftermath of 9/11, the tsunamis in Asia and Japan, Hurricane Katrina, and the Haiti earthquakes. Public health and disaster workers experience acute and longer term posttraumatic stress disorder (PTSD) and health risk behaviors such as increased alcohol and increased tobacco use.Reference Benedek, Fullerton and Ursano4 However, few studies have addressed large populations regarding these important components of our emergency public health response.
Work-related exposure to disasters has been associated with PTSD and depression in disaster workers.Reference Bills, Levy and Sharma5 A meta-analysis of 38 studies of rescue workers (N = 20 000) found that the worldwide pooled prevalence of PTSD was 10%.Reference Berger, Coutinho and Figueira6 The prevalence of PTSD in rescue workers enrolled in the World Trade Center health registry was 12.4% (range, 6.2% for police to 21.2% for volunteers).Reference Perrin, DiGrande and Wheeler7 Firefighters and mortuary workers have high levels of PTSD symptoms.Reference Fullerton, McCarroll and Ursano8-Reference North, Tivis and McMillen10 In an epidemiological study of first responders, symptoms of PTSD and depression increased with the number of critical work-related exposures.Reference Ward, Lombard and Gwebushe11 Firefighters working at the Oklahoma City bombing were more likely to have PTSD if they had longer exposure to the bomb site.Reference North, Tivis and McMillen10 Symptoms of major depression were reported by 27% of firefighters 13 weeks after responding to Hurricane Katrina.Reference Tak, Discoll and Bernard12
Few studies have examined the risk factors for increased alcohol and tobacco use in disaster workers. Almost half of the disaster workers reported drinking more alcohol than usual during the time that they worked at the World Trade Center, and about one-third continued increased use.Reference Stellman, Smith and Katz13 Smoking in ambulance personnel following the fireworks disaster in Enschede, the Netherlands, predicted symptoms of PTSD and depression 18 months after the disaster.Reference Van der Velden, Kleber and Koenen14 Increased tobacco use has been reported in 29% of disaster workers following 9/11, and 23% of former smokers resumed cigarette smoking.Reference Bars, Banauch and Appel15
The 2004 hurricane season provided a unique opportunity to examine public health workers of the Florida Department of Health (FDOH) who experienced repeated deployments, high work demand, and both work and family stressors in responding to the hurricanes. Specifically, we examined the relationship of probable PTSD, probable depression, and increased alcohol and tobacco use 9 months posthurricanes to disaster exposure and work demand in public health workers.
Methods
Participants and Procedures
Approximately 9 months after the 2004 hurricane season, questionnaires were distributed to FDOH employees using e-mail distribution lists. Participation was voluntary and anonymous, and completed questionnaires were transmitted electronically. Two versions of the questionnaire (ie, A and B) were distributed randomly so that each potential participant received only one version or the other. Questionnaire versions contained some of the same items and some unique items; version A focused on mental health items and version B emphasized workplace issues. Of an estimated 8564 FDOH personnel who worked during the 2004 hurricanes and were available at the time of the survey, we were able to contact and invite 6637 individuals to participate. After reading a description of the study and the informed consent, 4323 agreed to participate, and they completed and returned the questionnaire (version A = 2249; version B = 2074); the estimated response rate was 65.1%. This study used version A respondents. The participants ranged in age from 20 to 78 years (median = 49 years). The majority were women (80.4%; N = 1787) and currently married (66.5%, N = 1482). The majority were White (73.9%, N = 1623); 13.3% (N = 291) were African American; 8.7% (N = 190) were Hispanic; and 4.2% (N = 92) marked Other. Nearly half of the participants had less than a bachelor's degree (48.9%, N = 1091).
Measures
Probable PTSD
Probable PTSD was assessed with the 17-item PTSD checklist (PCL-17).Reference Weathers and Ford16 The PCL-17 lists all symptoms of PTSD according to the Diagnostic and Statistical Manual (Fourth Edition) (DSM-IV). Respondents rated how much they were bothered by those symptoms in the past month as it relates to their experience with the hurricanes. Possible responses were 1, not at all; 2, a little bit; 3, moderately; 4, quite a bit; and 5, extremely. Responses were summed with a range from 17 to 85. Studies in primary care settings have validated the use of PCL-17 scores of 30 or greater as indicative of probable PTSD (sensitivity = .86, specificity = .76).Reference Lang and Laffaye17, Reference Walker, Newman and Dobie18 This study combined the aforementioned criterion with a more clinically based indicator by using DSM-IV criteria: 1 intrusion symptom, 3 avoidance symptoms, and 2 hyperarousal symptoms, each present at the level of moderate or higher during the previous month. Respondents meeting both the total score and DSM-IV criteria were classified as having probable PTSD.
Probable Depression
Probable depression was assessed with the Patient Health Questionnaire Depression Scale (PHQ-9).Reference Spitzer, Kroenke and Williams19 Probable depression was present when at least 5 of the 9 symptoms were present “more than half the days” or “nearly every day” in the past 2 weeks, and 1 of the symptoms was depressed mood or anhedonia. One of the symptoms (“thoughts that you would be better off dead or of hurting yourself in some way”) was considered present if the respondent indicated presence regardless of duration. When compared to structured clinical evaluations in primary care settings, the PHQ-9 demonstrated 73% sensitivity and 94% specificity.Reference Kroenke, Spitzer and Williams20, Reference Spitzer, Williams and Kroenke21
Alcohol Use
An increase in alcohol use by alcohol users was assessed with 1 questionReference Grieger, Fullerton and Ursano22: “Since the hurricanes did you change your drinking habits for any 2-week period of time or more?” Possible responses were 1, I do not drink alcohol; 2, the amount I drank remained the same; 3, I drank more than usual; 4, I drank less than usual; and 5, I had stopped drinking but started again. Increased alcohol use was indicated if response choices 3 or 5 were selected and was compared to decreased or no change in alcohol use (response choices 2 and 4) in subsequent analyses. Persons indicating no use of alcohol were removed when evaluating change in alcohol use.
Tobacco Use
An increase in tobacco use by tobacco users was assessed with 1 questionReference Grieger, Fullerton and Ursano22, Reference Biggs, Fullerton and Reeves23: “Since the hurricanes did you change any tobacco habits (cigarettes, pipe, cigars, chewing tobacco) for any 2-week period of time or more?” Possible responses were 1, I do not use tobacco; 2, my tobacco use remained the same; 3, I increased my use of tobacco; 4, I decreased my use of tobacco; and 5, I had stopped using tobacco but started using it again. Scoring of this item was completed in the same way as the use of alcohol.
Hurricane Exposure
Exposure at the time of the hurricanes was assessed by being in the actual path of any of the 5 storms and amount of injury and damage. Injury and damage were assessed by experience of any of the following 6 events during each of the 5 hurricanes: relocation due to damage to one's home, damage to vehicle, injury or harm to self, injury or harm to spouse/significant other, injury/harm to children, and injury/harm to pets. A high level of hurricane exposure was indicated if respondents were in the path and experienced injury or damage for at least 1 storm.
Hurricane Work Demand
Work demand at the time of the hurricanes was assessed using 3 items: the number of hurricanes during which hurricane-related work was performed, the number of days during the hurricane response that at least 12 of 24 hours were worked, and the number of weeks that the hurricane response lasted before returning to a prehurricane work schedule. The work demand scale was dichotomized; high indicated that at least 2 of the following 3 criteria were met: 1, working at least 2 hurricanes; 2, working more than 12 hours on at least 7 days; and 3, a hurricane response lasting at least 4 weeks.
Statistical Analysis
Potential risk factors for probable PTSD, probable depression, increased alcohol use, and increased tobacco use at 9 months posthurricane in FDOH employees were evaluated using univariate logistic regressions and χ2 analyses. Odds ratios were defined as the likelihood of experiencing probable PTSD, probable depression, increased alcohol use, and increased tobacco use for individuals with versus without a risk factor. Hurricane exposure, work demand, and each potential demographic risk factor for probable PTSD, probable depression, and increased alcohol and tobacco use were also evaluated while controlling for the remaining risk factors using logistic regressions. The estimate of the odds ratio and its 95% confidence interval (CI) were reported. The likelihood ratio (LR) χ2 test was used to determine any difference between the odds for individuals with versus without a risk factor. Hosmer-Lemeshow goodness-of-fit χ2 analyses assessed how well the model predicted the data.Reference Hosmer and Lemeshow24 Statistical analyses were conducted using SPSS, Release 16.0.225 and Stata, Release 9.2.26
Results
The total mental and behavioral health burden (ie, those with 1 or more of the following: probable PTSD, probable depression, increased alcohol use, and/or increased tobacco use) was 11% (N = 229). More than 4% (4.5%, N = 97) of FDOH employees had probable PTSD, and 3.8% (N = 81) had probable depression. Of those with probable PTSD, 43.6% (N = 41) had probable depression compared to 1.8% (N = 37) without PTSD (χ2 = 39.72, df = 1, P < .001). Of those with probable depression, 52.6% (N = 41) had probable PTSD compared to 2.6% (N = 53) without depression (χ2 = 39.72, df = 1, P < .001).
Of those who drink alcohol, 7.0% (N = 70) reported increased alcohol use for at least a 2-week period since the hurricanes. Among those with probable PTSD who drink alcohol, 29.2% (N = 14) reported increased alcohol use compared to those without PTSD (5.9%, N = 55; χ2 = 38.10, df = 1, P < .001). Among those with probable depression who drink alcohol, 34% (N = 12) reported increased alcohol use compared to those without probable depression (5.9%, N = 56; χ2 = 41.92, df = 1, P < .001).
Of those who use tobacco, 22.1% (N = 73) reported increased tobacco use for at least a 2-week period since the hurricanes. Among those with probable PTSD who use tobacco, 50% (N = 9) reported increased tobacco use compared to those without PTSD (20.6%, N = 64; χ2 = 8.53, df = 1, P = .003). Among those with probable depression who use tobacco, 55.6% (N = 10) reported increased tobacco use compared to those without depression (20.6%, N = 63; χ2 = 11.91, df = 1, P = .001).
Sixteen percent (N = 359) of the respondents reported high levels of hurricane exposure, including being in the path of at least 1 storm and experiencing damage to self, significant others, or property. Nearly 30% (29.8%, N = 663) reported high levels of work demand during the hurricane including at least 2 of the following: working at least 2 hurricanes, working more than 12 hours on at least 7 days, and a hurricane response lasting at least 4 weeks.
Probable Posttraumatic Stress Disorder
Those not married were 1.78 times more likely to meet criteria for probable PTSD in the past month (6.3%, N = 45) than those who were married (3.6%, N = 51) (see Table 1). Those whose home was in the path of at least 1 storm and who experienced damage to self, significant others, or property were 3.30 times more likely to meet criteria for probable PTSD in the past month (10.4%, N = 36 vs 3.4%, N = 61) (see Table 1). Work demand was not associated with probable PTSD.
Relationship of Possible Risk Factors to Probable PTSD in FDOH EmployeesFootnote a
Abbreviations: FDOH, Florida Department of Health; LR χ2, likelihood ratio χ2; PTSD, posttraumatic stress disorder.
a Ns for the univariate analyses vary across analyses due to differing levels of missing data across the different variables.
b Adjusted for gender, race, education, marital status, hurricane exposure, and work demand; LR χ2 = 36.61, df = 7, P < .001;
Hosmer-Lemeshow goodness-of-fit χ2 = 4.71, df = 7, P = .695
To further investigate the relationship of hurricane exposure and work demand to probable PTSD, we examined a regression model that included all demographic variables, both hurricane exposure and work demand, and the interaction between hurricane exposure and work demand. In this and all subsequent analyses, the interaction was not significant (P > .10) and was therefore dropped from the final model. Those whose homes were in the path of at least 1 storm and who experienced damage to self, significant others, or property continued to be at increased risk of probable PTSD (OR = 3.21; LR χ2 = 22.41, df = 1, P < .001, 95% CI = 2.04-5.08) (see Table 1). Those who were not married were at a greater risk for probable PTSD (OR = 1.94; LR χ2 = 8.98, df = 1, P = .003, 95% CI = 1.26-2.99).
Probable Depression
Those with a lower level of education (less than a bachelor's degree) were 1.82 times more likely to have probable depression in the past 2 weeks (4.9%, N = 51) than those with a higher level of education (2.8%, N = 30) (see Table 2). Those with high levels of hurricane exposure were 3.06 times more likely to have probable depression (8.4%, N = 29 vs 2.9%, N = 52) (see Table 2). Work demand was not associated with depression. A regression model including all demographic variables and both hurricane exposure and work demand further examined the relationship of hurricane exposure and work demand to probable depression. Those with high levels of hurricane exposure remained at greater risk for probable depression (OR = 3.13; LR χ2 = 18.52, df = 1, P < .001, 95% CI = 1.92-5.10) (see Table 2). Those with less education also continued to be at greater risk of probable depression (OR = 1.78; LR χ2 = 5.46, df = 1, P = .020, 95% CI = 1.09-2.91).
Relationship of Possible Risk Factors to Probable Depression in FDOH EmployeesFootnote a
Abbreviations: FDOH, Florida Department of Health; LR χ2, likelihood ratio χ2.
a Ns for the univariate analyses vary across analyses due to differing levels of missing data across the different variables.
b Adjusted for gender, race, education, marital status, hurricane exposure, and work demand; LR χ2 = 27.97, df = 7, P < .001;
Hosmer-Lemeshow goodness-of-fit χ2 = 5.45, df = 8, P = .708
Increased Alcohol Use
Those with high levels of hurricane exposure were 3.01 times more likely to report increased alcohol use for at least a 2-week period (14.9%, N = 24 vs 5.5%, N = 46) (see Table 3). Similarly, those with high work demand during the hurricanes were 1.98 times more likely to report increased alcohol use (10.0%, N = 37 vs 5.3%, N = 33). The regression model, which included exposure, work demand, and the demographic variables, revealed that those with high hurricane exposure continued to be at increased risk of increased alcohol use for at least a 2-week period (OR = 3.09; LR χ2 = 14.47, df = 1, P < .001, 95% CI =1.78-5.38) (see Table 3). Similarly, those with greater work demand at the time of the hurricanes remained at an elevated risk (OR = 1.70; LR χ2 = 4.01, df = 1, P = .045, 95% CI = 1.01-2.85). There were no significant demographic risk factors. Those with increased alcohol use did not differ from those with either no change or decreased alcohol use over a 2-week period on gender, race, age, education, or marital status (see Table 3).
Relationship of Possible Risk Factors to Increased Alcohol Use in FDOH EmployeesFootnote a
Abbreviations: FDOH, Florida Department of Health; LR χ2, likelihood ratio χ2.
a Ns for the univariate analyses vary across analyses due to differing levels of missing data across the different variables.
b Adjusted for gender, race, education, marital status, hurricane exposure, and work demand; LR χ2 = 27.46, df = 7, P < .001;
Hosmer-Lemeshow goodness-of-fit χ2 = 13.25, df = 8,P = 0.104
Increased Tobacco Use
Those with high levels of hurricane exposure were 3.40 times more likely to report increased tobacco use over at least a 2-week period (41.9%, N = 26 vs 17.5%, N = 47) (see Table 4). Similarly, those with high work demand during the time of the hurricanes were 2.10 times more likely to report increased tobacco use (31.7%, N = 32 vs 18.1%, N = 41). Those with increased tobacco use over at least a 2-week period did not differ from those with no change or a decrease in tobacco use on any demographic characteristic (see Table 4).
Relationship of Possible Risk Factors to Increased Tobacco Use in FDOH EmployeesFootnote a
Abbreviations: FDOH, Florida Department of Health; LR χ2, likelihood ratio χ2.
a Ns for the univariate analyses vary across analyses due to differing levels of missing data across the different variables.
b Adjusted for gender, race, education, marital status, hurricane exposure, and work demand; LR χ2 = 27.32, df = 7, P < .001; Hosmer-Lemeshow goodness-of-fit χ2 = 2.83, df = 8, P = .945
The regression model, which included hurricane exposure, work demand, and the demographic variables, revealed that those with high levels of hurricane exposure remained at a heightened risk (OR = 3.23; LR χ2 = 12.66, df = 1, P = .004, 95% CI = 1.71-6.12) (see Table 4). Similarly, those with greater work demand continued to be at increased risk (OR = 2.11; LR χ2 = 5.77, df = 1, P = .163, 95% CI = 1.15-3.85). There were no significant demographic risk factors.
Discussion
The 2004 hurricane season provided an opportunity to examine public health workers at the FDOH who responded to multiple hurricanes and experienced both high work demand and personal exposure. The total mental and behavioral health burden was 11% (ie, people with probable PTSD, probable depression, increased alcohol use, and/or increased tobacco use); however, it is unclear whether the increased alcohol and/or tobacco use was a transient change for 2 weeks during or in the 9 months following the hurricanes or a more permanent change. Nine months after the hurricanes, 4.5% had probable PTSD and 3.8% had probable depression as compared to the 1-year population prevalence of 3.5% for PTSD and 6.7% for depression taken from a nationally representative sample of US adults.Reference Kessler, Chiu, Demler, Merikangas and Walters27 Personal hurricane damage and injury was related to a higher probability of PTSD, depression, and alcohol and tobacco use. Those with probable PTSD and probable depression were also more likely to report increased alcohol use and increased tobacco use. Interestingly, higher work demands were associated with increased more alcohol and tobacco use but not with probable PTSD or probable depression.
Postdisaster mental health comorbidity is related to decrements in disaster response capacity in the public health workforce. These findings at 9 months after the hurricanes indicate the likelihood of substantial mental health problems in the months immediately after the hurricanes. The higher the number of disorders, the greater the effects on functioning and poor outcome.Reference Kessler, Barker and Colpe28, Reference Kessler, Ormel and Petukhova29 Comorbidity following trauma supports the interrelationship of various disorders (as well as underlying related endophenotypes) to environmental stressors.
Importantly, greater exposure to hurricane stressors is not only associated with probable PTSD but also with probable depression. Those with greater hurricane exposure were 3.06 times more likely to report probable depression. Depression as a trauma-related outcome has not been examined as extensively as posttraumatic stress disorders. Future research should specifically examine depression following trauma as new onset depression. After adjusting for demographics and work demand, those with greater exposure remained at risk for probable depression (OR = 3.13).
In contrast to probable PTSD and probable depression, both high hurricane exposure and work demand during the hurricanes were associated with increased alcohol use and tobacco use. These remained strong predictors even after adjusting for demographics; adjusting for work demand in the case of exposure; and adjusting for exposure in the case of work demand. Increases in alcohol use are associated with negative health outcomes including family violence, increased motor vehicle accidents, and increased tobacco use.Reference Benedek, Fullerton and Ursano4, Reference McCarroll, Fullerton and Ursano9, Reference Grieger, Fullerton and Ursano22 Our findings highlight the need to better understand disaster-related workplace demands. These stressors are distinct from other personal injury or damage that can affect health risk behaviors of alcohol and tobacco use, although perhaps are not related to PTSD and depression. Interventions directed specifically to workplace demands may alter alcohol and/or tobacco use but not PTSD and depression.
Limitations
Several limitations of this study should be taken into account. This is a cross-sectional study, therefore further research using a longitudinal design is recommended to better determine the course of probable PTSD and probable depression over time and its relationship to health risk behaviors. Collecting data on early posttraumatic response and systematic sampling using multimethod approaches allows for larger and more representative samples; however, it is challenging working with highly mobile sample populations such as participants who may have left the FDOH shortly after the hurricane season. Collecting information on previous psychiatric history and history of alcohol and/or tobacco use in future studies is important to better understanding response to trauma. In addition, future research should examine the relationship of symptom severity and limitations on the ability to perform work. Because these data were collected electronically, it would be useful to examine the possibility of nonresponse bias.
Conclusions
The health of public health workers is critical to sustaining the nation's health. Work-related disaster mental and behavioral health burdens, such as, high rates of probable PTSD, probable depression, and health risk behaviors suggest the need for mental health interventions in the workplace. These findings have implications for disaster planning in the workplace, such as, shift rotation to minimize exposure and the role of disaster work demands on risk for increased alcohol and/or tobacco use. The costs of postdisaster PTSD, depression, and health risk behaviors are significant and should be included in disaster planning for primary care and public health workers.
About the Authors
Department of Psychiatry, Uniformed Services University of the Health Sciences, Bethesda, Maryland (Drs. Fullerton, McKibben and Ursano); National Institute for Occupational Safety and Health, Office of the Director, NIOSH, Washington, D.C. (Dr. Reissman); National Institute for Occupational Safety and Health, Work Organization and Stress Research Team, Cincinnati, Ohio (Dr. Scharf); National Institute for Occupational Safety and Health, Pittsburgh Research Laboratory (Dr. Kowalski-Trakofler); University of Miami Miller School of Medicine, Center for Disaster and Extreme Event Preparedness (Dr. Shultz).
