Baird, Coleen P. MD, MPH; DeBakey, Samar MD, MPH; Reid, Lawrence MPH; Hauschild, Veronique D. MPH; Petruccelli, Bruno MD, MPH; Abraham, Joseph H. ScD
On June 24, 2003, US military field reports indicated a large fire started at the state-run Al-Mishraq Sulfur Plant near Mosul, Iraq.1 The fire burned continuously for almost a month, until approximately July 21, 2003, emitting dense clouds of sulfur dioxide (SO2), a by-product of the combustion of elemental sulfur piles.
The overall amount of SO2 released into the atmosphere during the fire was later estimated at approximately 600 kt, with a daily average of approximately 21 kt. In comparison with highly polluting plants in the United States that produce 20 kt per year, the Misraq sulfur fire was considered an exceptionally strong point source of SO2.2 These measurements are based on the Earth Probe Total Ozone Mapping Spectrometer, a National Aeronautics and Space Administration satellite-based instrument that has been used to measure and map daily ozone levels and map volcanic SO2 clouds from space since 1982. The Earth Probe Total Ozone Mapping Spectrometer estimated daily volumes of SO2 emissions, together with satellite thermal infrared radiance imagery and US military and Iraqi news reports of fire conditions, clearly demonstrated that, over the weeks of this incident, the smoke plume varied in direction, length, and opacity over parts of Iraq and even western Iran. This event was reported to have been the largest nonvolcanic SO2 emission incident ever measured by any Total Ozone Mapping Spectrometer.2
At the time of this incident, thousands of US military personnel were deployed to the area in support of Operation Iraqi Freedom. Some of the troops in the area were called upon to assist local Iraqis fighting the sulfur fire. Others assisted in evacuating civilians from local towns nearby. Some continued various military missions and transport operations in the area. Military reports noted that odors characteristic of sulfur were reported at a base camp referred to as Q-West, which was 25 km to the southwest of Al-Mishraq, but also as far as the Mosul Airfield area, which was approximately 50 km to the north. Medical personnel at Q-West reported that medical visits potentially associated with the fire, mostly associated with respiratory irritation, increased during the period of the fire.1
When incidents occur in deployed settings, the ability to monitor may be limited by equipment at hand. Most routine monitoring of ambient air quality consists of the collection of samples for assessment of particulate matter and associated metals. Such samples are collected on filters that are shipped to military laboratories in the United States for analysis, and results are not available in real time. Nevertheless, during the fire, a team of military preventive medicine personnel (the 62nd Medical Brigade Preventive Medicine staff) recognized the potential for acute hazards from the fire. The team used available direct-reading monitoring equipment (eg, GASTEC Detector Tubes and Grab Sample Pump, GASTEC Corporation, Kanagawa, Japan) to obtain a limited number of grab samples for SO2 and hydrogen sulfide (H2S) to try to provide some real-time, semiquantitative context of the potential exposures. The sample results demonstrated extremely variable concentrations over a broad area and time and thus provided some ground evidence of the variability of exposures that may have occurred. Some of the grab sample concentrations were high enough to be consistent with significant acute effects such as eye and respiratory tract irritation and were compatible with some of the physical complaints reported by field personnel. These levels are potentially associated with chronic respiratory conditions. SO2, and H2S were, therefore, determined to be the primary contaminants of concern.1 Because of the limited availability of equipment and personnel for this extensive area, these real-time samples were taken at only a few locations and only on a few days. Therefore, these sample results do not provide an adequate basis to characterize the exposure received by any particular person or group.
Both SO2 and H2S can be acutely fatal at high levels of exposure, but no deaths were documented secondary to the fire.3,4 High but nonfatal exposures can result in notable acute respiratory effects and ocular and skin irritations. Neither SO2 nor H2S are known carcinogens.
Although the SO2 and H2S samples were obtained at limited times and locations, they demonstrated a broad range of potential concentrations and geographic area of potential concern. For example, SO2 levels ranged from nondetectable to a notable high level of 283 mg/m3 (108 ppm, based on 25ºC and 1 atm). These included samples obtained from base camp Q-West. The data indicate that a large population of interest was potentially exposed (at least intermittently) to levels plausibly associated with health effects.
SO2 is a colorless gas that can be detected by taste at concentrations of 0.35 to 1.05 ppm and has a pungent, irritating odor with an odor threshold of 0.67 to 4.75 ppm.5 SO2 is an upper-respiratory-tract and eye irritant; other effects include rhinorrhea, choking cough, and reflex bronchoconstriction. According to the National Academy of Science's National Advisory Committee for Acute Exposure Guideline Levels, exposure to SO2 concentrations more than 0.75 ppm can induce moderate bronchoconstriction during exercise in patients with asthma.5 Volunteers exposed to SO2 demonstrated increased airway resistance and decreases in forced expiratory volume during pulmonary function testing. The presence of respirable particles, dry air, exercise, and mouth breathing may increase the severity of adverse effects caused by SO2.6 Occupational studies of workers exposed to SO2 over extended periods have identified chronic health effects, including cough, sputum production, difficulty breathing, and reduction in pulmonary function.7 Single exposure to very high concentrations of SO2 has been associated with bronchial hypersensitivity, reactive airway dysfunction syndrome, bronchitis, and constrictive bronchiolitis.7–9 Case studies of severe acute occupational exposures and epidemiological studies of occupational and ambient air exposures to SO2 pollution described by the National Academy of Science provide a thorough overview of the respiratory health consequences associated with this hazard. The weight of evidence of a potential association between exposures possibly experienced by military personnel during the sulfur fire incident and chronic respiratory conditions raised sufficient concern to warrant an evaluation.5
H2S is a colorless gas with a characteristic odor of rotten eggs, exposure to which can cause acute eye, ear, nose, and throat irritations.4 H2S can be detected by smell at even very low concentrations, ranging from 0.0005 to 0.3 ppm. Exposure to high concentrations of H2S can cause acute respiratory arrest and pulmonary edema, although these effects are likely secondary to effects of H2S on the central nervous system. Exposure to H2S does not seem to result in significant alterations in lung function, but it may cause bronchial hyperresponsiveness, especially in sensitive subgroups of the population, for example, among patients with asthma.10–12 Although this constituent was considered in the assessment as consistent with the emission source and combustion chemistry, this hazard seems to possibly have been more localized to the vicinity of the fire based on data from direct-reading equipment in the field and thus may not have dispersed as broadly as the SO2.
The primary aim of this retrospective cohort study was to characterize the postdeployment respiratory health status of the US Army personnel potentially exposed to emissions from the Al-Mishraq Sulfur Plant Fire and to compare the risk of plausible adverse health outcomes among this group with unexposed personnel. The authors assessed and compared self-reported health status and postdeployment medical encounters through March 2007. Only active-duty personnel were included in the cohort. The data were limited to active duty members because the capture of medical encounters among Reserve and National Guard members may not be complete. In 2003, the time of the fire, medical visits in theatre were not captured electronically, so these visits could not be analyzed.
Defining the Potentially Exposed Population of Interest
Although individuals' SO2 and H2S exposure levels were not obtained, information from measurements made in the field, satellite imagery, and field reports of odors and symptoms were used to provide a basis for defining the potentially exposed population. The population of interest (POI) was defined by two separate exposure groups: firefighters (FFs) and potentially exposed personnel (PEP).
The FF group includes 191 US Army personnel who, based on a field roster of their names, were known to have been exposed during specific activities fighting the fire. These individuals were primarily from the 101st Airborne Division, the 52nd Engineer Battalion, the 326th Engineer Battalion, and the 887th Engineer Company. Although these individuals are assumed to have been exposed to relatively high levels of SO2 and H2S at various times during the fire due to their location at the fire site and the nature of their tasks, the specific individual exposure levels, duration, and frequency are not known or obtainable. On the basis of notable irritation effects initially reported, these personnel were provided personal protective equipment in the form of the military M40 mask. The initial use of the mask was reported to have provided inadequate protection (irritant effects of exposure were still noted).1 Procedures to increase the frequency of replacement of the mask's filter were reported to have improved the mask's effectiveness once these schedules were established. This information supports an assumption that high exposures were likely, but it does not permit specification of individual exposures, even in this group of personnel who were directly exposed during their activities.
Unlike the FF group that was listed on a field roster for specific tasks directly at the fire location, much less is known about the various personnel exposure experiences of the many other units in the area. As previously noted, the evidence provided via satellite imagery and field data and reports regarding the smoke plume variance over time and a broad area indicated exposure was, quite possibly, to thousands of personnel in the area. We determined that a reasonable boundary of the exposure area would be based on the distance of the farthest troop locations which reported awareness of the fire from the sulfur plant. This 50-km distance was used as the radius to define the potential exposure area because the evidence demonstrated shifts in plumes from northerly to southerly directions during the weeks of the fire. The military personnel determined to be within the 50-km area are referred to as the PEP group. To identify the individuals to be included in the PEP group, we abstracted data from unit deployment records; 6341 US personnel (not including the FF group) were identified as the PEP group. The PEP group included units at different locations, experiencing different levels of exposure to the plume's hazardous constituents. Some personnel closer to the vicinity of the fire included medics and others who assisted in evacuation of civilians. Many individuals were at a base camp 25 km away from the Al-Mishraq Sulfur Plant where medical visits reportedly increased by about 20% during the period of the fire, with one patient with asthma suffering an exacerbation during that period.1
Defining Unexposed Comparison Groups
To estimate the frequency of health outcomes in personnel not exposed to the sulfur fire smoke plume, we identified two separate comparison groups. One group, referred to as the contemporaneous cohort (CC), included 1869 personnel who had been deployed at the time of the Al-Mishraq Sulfur Plant fire but who were based at other locations in southwest Asia. The second comparison group, referred to as the geographical cohort (GC), includes 2284 personnel who were deployed to the same geographic area as the exposed groups but whose tours began after the fire had been extinguished (eg, between November 2004 and mid-July 2005). The Deployed Theater Accountability Software, a military software system that tracks personnel locations while deployed, was used to define this later group.
To determine and confirm deployment data locations and history, we abstracted deployment records using Deployed Theater Accountability Software. Demographic records for cohort members were obtained from the Defense Manpower Data Center. Pre- and postdeployment health questionnaire data, postdeployment inpatient hospitalization records, and outpatient clinical medical encounter data were obtained from the Defense Medical Surveillance System.
Self-Reported Questionnaire Data
Health questionnaire data were based on a mandatory Department of Defense health self-assessment, completed by service members before deployment and again within 3 months of return from deployment.13,14 These standardized predeployment health assessments and PDHAs include questions regarding current health status, health concerns, and medical problems.
Defining Index Deployment Time
An index deployment for each cohort member was defined as the deployment that either coincided or closely surrounded the sulfur fire incident (June 24, 2003, to July 21, 2003) for both exposed and unexposed groups. A timeframe up to a year before this index deployment was defined as the predeployment period, and the time period from the index deployment to the end of the study was classified as the postdeployment period. Times greater than 365 days before the index deployment were censored. Similarly, postdeployment time was censored at dates corresponding either to the end of the study period (March 31, 2007) or the deployment start date of the most recent nonindex deployment occurring before the end of the study. The person-time at risk in this study thus represents the cumulative time cohort members would be available in-garrison to seek health care and have clinical medical encounters captured in the inpatient and outpatient record systems. Medical encounters were not available for periods of deployment outside the United States.
Selected Disease Categories
During the year before, and the time after, the index deployment, the authors identified the following disease categories based on International Classification of Diseases–Ninth Revision–Clinical Modification (ICD-9-CM): diseases of the circulatory system (ICD-9-CM, 390 to 459); diseases of the respiratory system (ICD-9-CM, 460 to 519); and symptoms, signs, and ill-defined conditions (ICD-9-CM, 780 to 799). The authors specifically assessed chronic obstructive pulmonary disease (COPD) (ICD-9-CM, 490 to 496), asthma (ICD-9-CM, 493), other chronic bronchitis (ICD-9-CM, 491.8), pneumoconiosis and other lung disease due to external agents (ICD-9-CM, 500 to 508), ischemic heart disease (ICD-9-CM, 410 to 414), other forms of heart disease (ICD-9-CM, 420 to 429), cerebrovascular disease (ICD-9-CM, 430 to 438), symptoms involving cardiovascular system (ICD-9-CM, 785), and symptoms involving the respiratory system (ICD-9-CM, 786).
SAS version 9.1 (SAS Institute Inc, Cary, NC) was used for all statistical analyses. Rates of medical encounters (per 1000 person-years) were compared between predeployment and postdeployment periods, stratified by the two groups in the POI (the FF and PEP groups) and two comparison unexposed groups (the CC and GC). Inpatient and outpatient data were combined because the numbers of inpatient encounters were insufficient for separate analysis. Clinical medical encounters within 1 year before the index deployment were used to establish preexisting medical conditions. Analyses were based on the primary (first-listed) diagnosis for each encounter. For any pair or series of encounters with the same ICD-9-CM code within 30 days of each other, only the first encounter was included.
Unadjusted incidence rates and standardized morbidity ratios (SMRs), accounting for age variation across the comparison groups, were calculated. For large numbers, 95% confidence intervals were estimated using methods described by Breslow and Day.15 For small numbers (n < 100), confidence intervals were derived from the Poisson distribution according to methods described by Cain and Diehr.16
Table 1 summarizes the stratified demographic and occupational characteristics of the overall study population. The study population includes the two groups (FF and PEP) that make up the sulfur fire POI and the two unexposed comparison groups (the CC and GC). Over half of the study population was younger than 29 years at the time of deployment. Nevertheless, a larger proportion of the POI were younger than 29 years, compared with the unexposed CC and CG comparison groups. The study population was comprised predominantly of enlisted men. Differences in the proportion of women between the FF group and both the unexposed comparison groups were not statistically significant, although the proportion of women in both of the unexposed groups was significantly higher than that among the PEP group. The study population was almost exclusively comprised of the US Army personnel, with the exception of two US Air Force individuals in the PEP. Just less than a third of the study population had primary occupations in combat arms positions.
Self-Reported Health Status
Postdeployment health questionnaires were completed and available for 85% of the potentially exposed population. The availability was significantly less for the comparison populations, so only the potentially exposed population will be described here. About 25% of personnel in the POI reported a change in health for the worse during deployment (26% of FFs and 25% of PEP). Health concerns were reported by 39% of the FFs and 23% of the PEP. Self-reported symptoms were common in the PEP group with 24% of the FFs reporting cough, 34% reporting runny nose, and 31% reporting difficulty breathing. In the broader PEP group, 16% reported cough, 28% reported runny nose, and 14% reported shortness of breath.
Figure 1 summarizes respiratory disease crude incidence rates for the two POI groups and the unexposed comparison groups (by selected ICD-9 code categories), comparing pre- and postdeployment periods. Overall, the incidence of encounters for respiratory diagnoses in each of the four groups (those in the POI as well as the two unexposed comparison groups) decreased from the predeployment period to the postdeployment period. Nevertheless, the decrease was not statistically significant in any of the groups, and it was driven exclusively by the dominant outcome category (acute respiratory infections). Contrary to this overall trend, the frequency of encounters for “other upper respiratory tract outcomes” (ICD-9-CM, 470 to 478), COPD and allied conditions (ICD-9-CM, 490 to 496), and respiratory signs, symptoms, and ill-defined conditions (ICD-9-CM, 786) increased after deployment, relative to the predeployment period for all four groups. The increase in the incidence of encounters for COPD from the predeployment period to the postdeployment period was statistically significant among the PEP and the CC groups; the increase for respiratory signs, symptoms, and ill-defined conditions was statistically significant among the PEP group.
Table 2 presents postdeployment medical encounter SMRs and 95% confidence intervals for the sulfur fire POI (191 FFs and the 6341 members of the PEP). Compared with the unexposed CC group, the age-adjusted SMR for diseases of the respiratory system overall (ICD-9-CM, 460 to 519) was significantly elevated in the PEP group (SMR = 1.13; P < 0.05), but the SMR in the FF group was not statistically significant (SMR = 1.09). The age-adjusted SMR for COPD was not significantly different from unity in the FF group (SMR = 0.73) nor in the PEP group (SMR = 1.1). Compared with the unexposed GC group, the age-adjusted SMR for diseases of the respiratory system overall was significantly less than unity among both the PEP (0.64) and FF (0.62) groups. The age-adjusted SMRs for COPD were below the null for both the FF (SMR = 0.41; P < 0.05) and the PEP groups (SMR = 0.58; P < 0.05).
Self-Reported Exposures and Health Status
Subjective, postdeployment health data indicate that the persons defined as being known or potentially exposed to smoke from the Al-Mishraq Sulfur Fire (the sulfur fire POI) self-reported medical problems, respiratory symptoms, and health concerns. Although exposure to the sulfur fire smoke is a possible cause of such symptoms, deployment in general has been associated with an increase in respiratory symptoms. Smith et al17 reported an increase in self-reported respiratory symptoms in those who deployed (14%) compared with those who had not (10%), although both groups had similar rates of physician-diagnosed respiratory complaints (1%). Particulate matter, burning trash, and fuel have been implicated in postdeployment respiratory and other conditions.17
We observed an elevated rate of postdeployment medical encounters for acute respiratory infections and ill-defined conditions among the sulfur fire POI when using the unexposed CC group as a standard. Nevertheless, we observed no such increase when using GC group as the standard population.
Postdeployment health care utilization data did not reveal substantial increases in the frequency of chronic respiratory problems associated with exposure to the Al-Mishraq sulfur fire. Specific outcomes of interest included asthma and COPD. We observed fewer medical encounters for asthma than expected. This observation was statistically significant when using unexposed GC group as the standard population. Although some assume that deployed military personnel are medically screened out for asthma, individuals with asthma before the age of 13 years or with a waiver can join the military and deploy. Exacerbation of existing asthma at the time of the fire would be plausible, but as discussed, in-theatre medical visits and diagnoses were not available. Roop et al18 surveyed redeploying Army personnel and found that 5% of troops deployed to southwest Asia reported a previous diagnosis of asthma. If new-onset asthma had been diagnosed during deployment, we would expect to see at least a follow-up visit for this condition after deployment.
The frequency of clinical medical encounters for COPD among the exposed groups were similar to, or again lower than expected, depending on the unexposed population used to standardize the incidence rates. Unlike asthma, COPD is a chronic condition that might not be diagnosed for several years after deployment. Mild COPD or subclinical decreases in pulmonary function may be detected through the use of longitudinal individual spirometry; however, the Department of Defense does not routinely conduct surveillance spirometry on all service members.19
One of the units deployed to Q-West during the fire returned to Fort Campbell, where screening spirometry was offered. Individuals with abnormal screening spirometry, symptoms, or concern about the fire were referred to specialty care. Some of the referred individuals were symptomatic but had normal pulmonary function. It was noted that baseline pulmonary function tests of individuals would have been useful due to considerations that athletic individuals might have declines that would be significant for them, yet still within the predicted normal range.1 From this group, 38 of 80 soldiers were subsequently diagnosed with constrictive bronchiolitis as reported by King et al.20 Within this group, a reported history of exposure to the 2003 sulfur fire was common but not universal.20 The screening program described would be expected to generate more specialty visits and likely more respiratory diagnoses than the comparison groups due to surveillance bias, but this was not seen. The frequency of postdeployment respiratory clinical encounters among the groups considered exposed to the sulfur fire plume was generally not greater than that observed among the unexposed comparison groups.
When individuals' postdeployment medical encounters were compared with their predeployment encounters, both exposed FF and PEP groups showed postdeployment increases in rates of various respiratory-related clinical outcomes. This trend, however, was not statistically significant. Perhaps more importantly, the increase in postdeployment respiratory-related clinical outcomes was also demonstrated in both the unexposed control groups and was even a statistically significant increase within the GC group. This could be due to a surveillance bias related to increased concerns and referrals at the time of the postdeployment assessment. Nevertheless, this demonstrated increase in respiratory disease was associated with deployment in general and not deployment during the time of the sulfur fire.
A primary limitation of this assessment is the absence of individual-level SO2 or H2S exposure data. Our analysis assumed that all individuals in the PEP and FF groups were equally exposed. We assigned exposure status to personnel who were present in the area where the smoke plume had generally been identified. Although we feel that the qualitative exposure assignments used in this evaluation are reasonable, the prospect of nondifferential misclassification of exposure among persons in the sulfur fire POI is real. The greatest potential for exposure misclassification exists for the PEP group, which was defined by a broad geographic and temporal proximity to the sulfur fire. Given that actual exposure levels of personnel in the POI cannot be verified, this assumption was a substantial but unavoidable limitation to this evaluation. In all cases in this study, errors in exposure classification are unlikely to be associated with the assignment of the health outcomes. This error is therefore likely to be nondifferential in nature.
A portion of this study utilized self-reported characterizations of health status from Department of Defense–required PDHAs. The recall associated with PDHAs is undoubtedly imperfect because reporting is subject to lapses in memory, and the motivation for reporting (or not reporting) exposures and health concerns on PDHAs is undoubtedly complex. As such, some researchers14 have suggested that survey results be interpreted with caution. Because differential misclassification bias would be induced if the quality (accuracy and completeness) of postdeployment survey data was not independent of exposure to the sulfur fire in this population, we did not compare the results of the exposed group with the unexposed group as their rates of completion differed.
The prospect of confounding also limits our inference. It is possible that uncharacterized differences in risk factors for adverse health outcomes exist between the sulfur fire POI and unexposed groups. Differences in smoking behavior and other environmental, occupational, or hobby-related exposures are examples of such potential confounders. Members of exposed groups were generally younger compared with the unexposed groups, so we adjusted for age. Reservists are likely to be older and may have different levels of baseline health, but we did not include Reservists in the analysis. Members of the exposed group were more likely to be male because these are combat arms units, but we did not adjust for sex.
In this study, postdeployment medical encounters for respiratory conditions were not associated with exposure to the sulfur fire. Troops deployed to an area not far from where the sulfur fire had burned, but well after the fire had been extinguished, were more likely than personnel exposed to the sulfur fire to have an initial, postdeployment respiratory disease medical encounter. Nevertheless, all groups showed an increase in respiratory visits after deployment compared with that before deployment, although this increase was statistically significant in only one of the unexposed comparison groups. At least some of the increase in health care encounters after redeployment is expected due to referrals generated from self-reported symptoms and exposure concerns identified in the PDHAs. Other authors17 have reported increase in respiratory symptoms after deployment.
Although the findings of this exploratory analysis cannot be used to rule out the possibility of an association between sulfur fire exposure in Iraq and respiratory disease, they do not demonstrate a definitive link. Nevertheless, the follow-up time in this study (July 2003 to March 2007) does not permit the evaluation of associations between sulfur fire plume exposures and diseases with a latency of longer than 4 years. To evaluate potential chronic health effects, as well as potential progression from symptoms to disease in all groups, we recommend repeating this study with a longer duration of follow-up.
1. Baird CP. Mishraq Sulfur Fire Environmental Exposure Assessment, June 2003–March 2007. Aberdeen Proving Ground, MD: US Army Public Health Command (Provisional); 2010.
2. Carn S, Krueger A, Krotkov N, Gray M. Fire at Iraqi sulfur plant emits SO2 clouds detected by Earth Probe TOMS. Geophysical Res Lett. 2004;31:1–4.
3. US Department of Health and Human Services, Public Health Service Agency for Toxic Substances and Disease Registry. Toxicological Profile for Sulfur Dioxide. Atlanta, GA: US Department of Health and Human Services, Public Health Service Agency for Toxic Substances and Disease Registry; 1998.
4. US Department of Health and Human Services, Public Health Service Agency for Toxic Substances and Disease Registry. Toxicological Profile for Hydrogen Sulfide. Atlanta, GA: US Department of Health and Human Services, Public Health Service Agency for Toxic Substances and Disease Registry; 2006.
5. National Research Council. Acute Exposure Guideline Levels for Selected Airborne Chemicals, Volume 8. Washington, DC: National Academy Press; 2010.
6. World Health Organization. Air Quality Guidelines Global Update 2005. Geneva, Switzerland: World Health Organization; 2006.
7. Skalpe IO. Long-term effects of sulphur dioxide exposure in pulp mills. Br J Ind Med. 1964;21:69–73.
8. Charan NB, Myers CG, Lakshminarayan S, Spencer TM. Pulmonary injuries associated with acute sulfur dioxide inhalation. Am Rev Respir Dis. 1979;119:555–560.
9. Epler G, ed. Fume-Related Bronchiolitis Obliterans. New York, NY: Raven Press; 1994.
10. Bhambhani Y, Burnham R, Snydmiller G, MacLean I, Lovlin R. Effects of 10-ppm hydrogen sulfide inhalation on pulmonary function in healthy men and women. J Occup Environ Med. 1996;38:1012–1017.
11. Hessel PA, Herbert FA, Melenka LS, Yoshida K, Nakaza M. Lung health in relation to hydrogen sulfide exposure in oil and gas workers in Alberta, Canada. Am J Ind Med. 1997;31:554–557.
12. Jappinen P, Vilkka V, Marttila O, Haahtela T. Exposure to hydrogen sulphide and respiratory function. Br J Ind Med. 1990;47:824–828.
14. Mancuso JD, Ostafin M, Lovell M. Postdeployment evaluation of health risk communication after exposure to a toxic industrial chemical. Mil Med. 2008;173:369–374.
15. Breslow N, Day N. Statistical Methods in Cancer Research: The Design and Analysis of Cohort Studies. Lyon, France: International Agency for Research on Cancer; 1987.
16. Cain KC, Diehr P. Testing the null hypothesis in small area analysis. Health Serv Res. 1992;27:267–294.
17. Smith B, Wong CA, Smith TC, Boyko EJ, Gackstetter GD. Newly reported respiratory symptoms and conditions among military personnel deployed to Iraq and Afghanistan: a prospective population-based study. Am J Epidemiol. 2009;170:1433–1442.
18. Roop SA, Niven AS, Calvin BE, Bader J, Zacher LL. The prevalence and impact of respiratory symptoms in asthmatics and nonasthmatics during deployment. Mil Med. 2007;172:1264–1269.
19. Townsend MC and the ACOEM Occupational and Environmental Lung Disorder Committee. Evaluating pulmonary function changes over time. American College of Occupational and Environmental Medicine web site. Available at: http://www.acoem.org/EvaluatingPulmonaryFunctionChange.aspx
. Accessed March 29, 2012.
20. King MS, Eisenberg R, Newman JH, et al. Constrictive Bronchiolitis in Soldiers Returning from Iraq and Afghanistan. New England Journal of Medicine. 2011;365:222–230.