The terrorist attacks of September 11, 2001, resulted in the complete destruction of the World Trade Center (WTC) complex in New York City. Almost immediately, an enormous workforce was mobilized to conduct rescue and recovery and begin the process of cleaning up the 16-acre field of debris. It is likely that several thousand workers were involved in the cleanup operation that transformed the disaster site from a 1.8 million ton mountain of rubble to an area ready for reconstruction in under 10 months.1 Central to the cleanup operation was a labor force that represented a wide range of skills and that performed debris removal 24 hours per day, 7 days per week. During the first months after the attacks, working conditions were especially hazardous. In addition to the potential for physical injury resulting directly from the removal of debris, workers were exposed to a complex mixture of mechanically generated dust, combustion particles, gases, and fumes.2–4 Analysis of settled dust samples collected shortly after the attack showed the dust to be composed of a broad range of building materials, such as cement, vermiculite, plaster, mineral wool fibers, paint particles, and paper and glass fragments. The dust was found to be highly alkaline. Polycyclic aromatic hydrocarbons and polychlorinated biphenyls were also detected. In addition, on-site measurements of airborne particles showed that the particle size distribution changed from day to day depending on on-site activities and the intensity of the burning fires.5
There is evidence that inhalation exposure to these contaminants was hazardous to the respiratory system. A study carried out at the beginning of October 2001 of 362 New York City firefighters involved in the disaster response reported a prevalence of lower respiratory symptoms (LRS) ranging from approximately 20% to 50%.6 Approximately 1 month after the attack, 8% of 1636 firefighters present during the collapse of the first tower developed a persistent and debilitating cough labeled “WTC cough.”7 Sixty-two percent of surviving Emergency Services Police had a cough two and a half months after the event.8 During December 2001, we assessed 183 construction workers on-site, of whom 65% started work at the WTC without LRS. Of the symptom-free group, 34% developed cough, 24% developed phlegm, and 19% developed wheeze while working at the site.9
Results from these studies suggested that inhalation exposure at the WTC may have been unusually hazardous. Since the end of the cleanup operation, there have been several reports in the press about WTC workers experiencing respiratory problems many months after leaving the disaster site, but more than 6 years after the attacks our understanding of the impact of exposure at the WTC on respiratory health still is limited. Herbert et al report on 9442 workers who self-referred between July 2002 and April 2004 to a clinical screening program specifically set up for WTC workers and volunteers. Of the group that reported no LRS in the year before work at the WTC (85%) 44% said that they developed a new LRS while working at the WTC. At the time of the examination 32% of these workers said they were still experiencing symptoms.10 Banauch et al examined 179 firefighters recruited from approximately 1500 New York City firefighters eligible for enrollment in the study. They reported that of the 77 who had been present during the collapse of the first building, 23% exhibited bronchial hyperreactivity 12 months after their initial exposure.11
The present study was conducted to address concerns that construction workers involved in the cleanup operation could be at risk for respiratory health effects that persisted after completing work at the WTC. The goal of this study was to test the hypotheses that workers involved with the cleanup operation at the WTC are at elevated risk for respiratory health effects several month after the initial exposure, and that the risk is related to the duration of work at the site and the location of work.
Materials and Methods
Study Design and Population
This study was designed to evaluate and compare respiratory health symptoms of two groups—workers involved in the cleanup operation at the WTC and workers who were never involved with the disaster site. Participation in this project was explored with representatives from six New York City local labor unions, as well as the New York City Department of Sanitation (NYCDOS). Three of the local unions approached, as well as the NYCDOS, agreed to participate. The local unions were affiliated with the International Brotherhood of Teamsters, the International Union of Operating Engineers, and the Laborers International Union of North America. Representatives of each organization were asked to identify members who had participated in the cleanup and recovery effort at the WTC disaster site, as well as a group of workers who were never at the WTC. The local International Brotherhood of Teamsters and International Union of Operating Engineers Unions agreed to both requests; the local Laborers International Union of North America Union and the NYCDOS agreed to identify only workers who had been involved at the WTC. Workers contacted to participate in this study included truck drivers, heavy equipment operators, laborers, and sanitation employees.
Study Material and Procedures
Data used for analysis of respiratory health effects were collected by a survey specifically designed for this study. The World Trade Center Clean Up and Recovery Worker Health Assessment Survey included questions covering demographic information, work history, work experience at the disaster site, exposure at the WTC, general health history, and respiratory health. Two versions of the survey were created. One version was designed to be administered to workers who were involved at the WTC. A second version, which did not ask questions specific to the disaster site, was designed to be administered to workers who were never at the WTC. Nevertheless, in this version of the survey, a set of questions was included to allow a respondent to indicate whether they had been at the WTC. Both versions were printed in a scannable format (NCS Pearson, Columbia, PA). Each survey was labeled with a unique serial number that was printed on every page along with a page number, making every page of the survey unambiguously identifiable.
Surveys were administered by mail. Packets contained the survey, a cover letter introducing and supporting the study from the local labor union or the NYCDOS on letterhead, a letter of disclosure about the purpose of the study and requirements of participation on Johns Hopkins Bloomberg School of Public Health (JHSPH) letterhead, brochures about health services specifically for WTC workers, a self-addressed stamped envelope to be used to return the completed survey to JHSPH, and a sharpened pencil. Consent was considered given if a worker chose to complete and return the survey. The JHSPH Institutional Review Board approved the study. Mailings were conducted either directly from JHSPH, if a mailing list had been provided, or directly from the organization if contact information could not be released. In the latter case, assembled packets with postage were delivered to the organization where they were labeled and mailed. Two organizations agreed to provide mailing lists. The surveys were mailed between March and June 2003. Approximately 2 weeks after each mailing, a post card was sent to remind workers to fill out the surveys. Opportunities to educate members of the participating organizations about the survey and its purpose were limited. All organizations were provided with information about the survey to give to their members, as well as posters to advertise the survey. The study team did not have direct access to organization members; however, members were provided with a toll-free number they could call with questions about the survey.
Demographics and Health.
Participants were asked date of birth, race/ethnicity, martial status, income level, and educational achievement. Health history was assessed by a 34-item checklist based on an evaluation of Gulf War veterans.12 Included in this checklist were specific questions about respiratory diseases, including asthma, emphysema, chronic bronchitis, pneumonia, and allergies. Additional questions concerning cardiovascular and circulatory disease, as well as skin disease and other disorders were included. Current respiratory health, including cough, phlegm production, shortness of breath, and wheeze, was assessed using a modification of the American Thoracic Society ATS-DLD-78-A respiratory health questionnaire.13 Current health was defined as experiencing a symptom at the time the participant filled out the survey. Smoking status was assessed by number of cigarettes smoked per day.
Experience at the WTC.
Experience at the disaster site was reported as “first day on-site,” and “last day on-site,” as well as work location. Participants were also asked to identify their work location at the WTC according to the definitions found in Table 1. Work location definitions were identical to those used in the World Trade Center Health Registry survey.14 Participants were asked to identify their work location for each month worked at the WTC. Months were defined as September 2001, October 2001, November 2001, December 2001, January to June 2002. Participants were asked to describe respirator use for each month worked at the WTC according to the categories of “never used,” “rarely used,” “used some of the time,” “used most of the time,” “used always,” and “no need.”
Exposure was defined as 1) “never” worked at the WTC; 2) “ever” worked at the WTC; 3) “duration” at the WTC; and 4) “location of work” at the WTC by month worked at that location. Never worked at the WTC was defined for participants who completed the questionnaire designed for non-WTC workers. Ever worked at the WTC was defined for participants who completed the questionnaire designed for WTC workers. Duration was determined by subtracting the reported first day of work from the reported last day of work plus 1 day. If participants reported only the month they started and/or ended work at the site, a start and/or end date was assigned as the 15th of the month. If that month was September 2001, then September 20, 2001 (mid-date between September 11 and 30) was assigned.
Logistic regression models were used to assess the relationship between exposure at the WTC and the risk of LRS. Rate ratios (RRs) and 95% confidence intervals (CIs) were determined for current LRS looking independently at cough, phlegm, shortness of breath, and wheeze and any LRS or LRS combinations. LRS combinations included cough and/or phlegm, shortness of breath and/or wheeze. All multivariate analysis models were adjusted for age, gender, race, marital status, education, household income, respiratory history, and smoking by number of cigarettes per day (1 to 9, 10 to 19, 20+).
A total of 4546 workers were identified as having been involved at the WTC (ever), of which 1114 (25%) responded to the survey. Of these, six respondents gave ambiguous information regarding their role at the WTC and were excluded from the analysis. An additional 15 respondents did not give information about their first day on-site, their last day on-site, or the months they worked at the site. These respondents were excluded from all analyses involving duration of time at the WTC. A total of 2103 workers were identified as having never been involved at the WTC (never) by their union or employer. Of this group, 248 (12%) responded. Twenty nine reported that they had been at the WTC at some point. Because information for key WTC exposure variables was not collected in the survey designed for non-WTC workers, including first and last day on-site, these respondents were excluded from the analysis. General characteristics of the ever and the never groups are shown in Table 2.
Information resulting from reported respirator use was incomplete, but did provide a qualitative assessment of the use of personal protective equipment over the course of the cleanup effort. Forty six percent of respondents reported wearing a respirator at least some of the time during September. For the time period between January and June only 29% of respondents reported wearing a respirator at least some of the time. Nevertheless, information provided by respondents also varied by month. Seventeen percent of respondents provided no information about respirator use during September. This increased to 53% who did not provide any information about respirator use for January through June.
Relative to those who were never involved at the WTC, the risk of respiratory symptoms for all outcome categories was found to be associated with exposure at the WTC after adjusting for confounding variables (Table 3). All LRS and LRS combinations showed a 2-fold or greater risk among exposed workers approximately 20 months after their first exposure. Overall, the RR for an exposed worker to develop any LRS was 3.40 (95% CI: 2.33–4.94).
Further analysis showed that the duration of exposure was also significantly associated with the development of LRS after adjusting for the confounding variables given above (Table 4). When compared with those never exposed at the WTC, it was found that the RR for developing any LRS increased from 2.04 (95% CI: 1.28–3.24) for those exposed at the site 1 to 7 days to 5.79 (95% CI: 3.59–9.35) for those exposed over 240 days at the WTC.
Airborne contaminant concentrations were not homogenous across the disaster site. Concentrations measured in the middle of the debris field were an order of magnitude greater than those measured near the perimeter, suggesting that work location was a factor in exposure.5 Therefore, the influence of work location at the WTC on the risk of LRS was also explored. Table 5 shows the RR of LRS by work location adjusting for the variables listed above, as well as duration of exposure. When compared with those who were never at the site, work locations that presented additional risk of exposed at the WTC included “Outdoors on the WTC debris pile/pit” and “In an enclosed area beneath the WTC debris pile” with the enclosed area presenting the greatest additional risk. The RR of any LRS was 1.78 (95% CI: 1.07–2.94) for “In an enclosed area beneath the WTC debris pile” and 1.58 (95% CI: 1.20–2.08) for “Outdoors on the WTC debris pile/pit.”
The concentration of airborne contaminants at the site was dynamic. On September 11, concentrations of airborne particles were likely in the thousands of milligrams per cubic meter; unfortunately no data exist for this day and several days after. Limited measurements collected in October 2001 and later in April 2002 showed that airborne contaminant concentrations decreased until they had approached background near the end of the cleanup operation.5 To evaluate the potential impact of differences in air pollution concentrations over time on the risk of LRS we used start date at the site as a surrogate for exposure intensity. In this analysis, start day 1 equals September 11. Table 6 presents RR for LRS as related to start date. Over the first 4 months the impact of start date on LRS was similar. However, the risk doubled if a worker started at the site around the beginning of 2002: RR = 4.82 (95% CI: 1.80–12.94).
The World Trade Center Clean Up and Recovery Worker Health Assessment Survey was initiated to address concerns that respiratory health effects would result from involvement with the cleanup and recovery effort at the WTC. These concerns were expressed both by workers at the disaster site and their representatives. The goal of this study was to examine the impact of exposure to airborne pollutants during the cleanup effort on the risk of experiencing LRS several months after completing work at the WTC.
One of the most challenging issues for exploration of health effects resulting from work at this disaster site has been the estimation of exposure. Daily monitoring by the US Environmental Protection Agency for particulate matter at multiple locations near the disaster site was not initiated until late September 2001.15 As a result, no air quality data are available for the period when dust exposures were the highest, including the day of the attacks and the first few days after. In addition, feasibility issues related to access to power and instrument security resulted in no systematic monitoring being conducted directly within the debris field or at the debris perimeters. Data resulting from monitors located near the site are not representative of concentrations present directly on-site during the cleanup effort.
Among studies examining health effects associated with work at the WTC, most have defined exposure by timing at the site. In a study of ironworkers involved in the cleanup effort, exposure was defined as presence at the site during specific time periods and was used as an eligibility requirement for participation in the study.16,17 In a study of firefighters responding to the disaster, Prezant et al also used arrival time as a surrogate for exposure.7 Exposure was ranked as “high” if arrival occurred the morning of the attacks, “moderate” if arrival occurred that afternoon or the next day, “low” if arrival occurred 3 to 7 days later, and “not exposed” if arrival at the site occurred after 14 days. Similar exposure classifications were used in a study of emergency service police officer and additional studies of firefighters. Herbstman et al categorized exposure of their study population of trades workers by the number of days spent at the WTC with exposure groups defined as having been on-site 1 to 32 days, 33 to 64 days, and 65 to 96 days.9 In all of these studies, respiratory health effects were then reported as the percentage of the study population that was symptomatic.
These studies provide provocative evidence that inhalation exposure at the site was inducing problems with respiratory health. However, estimations of exposure based on time at the site alone may not completely capture aspects of engagement with the site that could result in respiratory symptoms. In this study, we expanded the definition of exposure by developing a series of exposure variables that allowed us to examine whether different interactions with the disaster site had differential impacts on the risk of respiratory symptoms. In our most basic model we compared workers who were never at the disaster site with a group that had been there for any period of time. In this analysis we found that WTC workers were more than 3 times as likely to have respiratory symptoms as compared with those who had never worked at the site after adjusting for age, gender, race, martial status, education, income, respiratory disease history, and smoking. We then stratified exposure by the number of days spent on-site to assess the impact of work duration on the risk of symptoms. We again found that when compared with workers who had never been at the WTC, all workers, regardless of the duration of their employment, were at significantly elevated risk for LRS. In addition, the risk increased with the number of days at the site. Workers who had spent 1 to 7 days on-site were twice as likely to be still experiencing symptoms (RR = 2.04, 95% CI: 1.28–3.24) as those who were never there, whereas workers who were present during the entire cleanup operation were more than 5 times as likely (RR = 5.79, 95% CI: 3.59–9.35).
The potential for inhalation exposure to airborne contaminants was not uniform across all WTC-related job sites. Measurements of particle concentrations made on-site during October 2001 showed a strongly decreasing concentration gradient from the middle of the debris outward.5 Our direct experience on-site between October 2001 and April 2002 also provided us with insight into differences in airborne concentrations by location. Particle concentrations were clearly higher within the debris pile than at the WTC dock or on the East Side Pier, suggesting that not all work locations presented the same opportunity for exposure. In addition, airborne particles concentrations changed over time. It was expected that as the cleanup operation progressed, air quality would improve. From September 11, 2001, through the end of December 2001 air quality at the site was impacted by debris removal activities, which disturbed the rubble and resuspended dust that had already settled to the ground, and underground fires, which continued to burn until December 19 when they were officially declared extinguished. In the beginning of 2002, debris removal had moved into the substructure of the WTC complex. By April 2002, the rubble pile was reduced to a three-story mound where the North Tower formerly stood. Although clear evidence of the disaster remained, the 16-acre area had become more similar to a normal construction site, and the air quality had significantly improved.
To explore the impact of work location on the risk of LRS we stratified our population by work location. Work locations are described in Table 1 and are illustrated in Fig. 1. We found that the site did not present a uniform risk. Working outdoors on the debris pile or in the pit, and in an enclosed area beneath the debris was more hazardous than working in other areas, including the Fresh Kills landfill where the use of personal protective equipment was more rigorously enforced. It was also found that arrival time at the disaster site impacted risk. However, surprisingly, the greatest risk of symptoms was associated with starting work at the site after a significant portion of the debris had been removed and the fires had been extinguished. With the fires out and the debris in the substructure, at the beginning of 2002, the disaster site had become more similar to a typical construction site. It is possible that those beginning work at the site after the first of the year were less conscious of the potential for hazardous exposures, viewed working there as more of a typical construction job, and therefore took even fewer protective precautions against inhaling WTC dust.
In this analysis, we recognize that the low response rate may have resulted in selection bias within our population, in that a worker experiencing symptoms may have been more likely to respond to the survey than a worker who felt asymptomatic. In addition, there were significant differences in the response rates between the ever and never exposed groups. To examine the impact of possible selection bias, we estimated theoretical RRs for cough under progressively stringent assumptions taking advantage of our specific information about the number of individuals contacted to define a denominator for each group. We evaluated the RR of cough by ever/never exposure groups based on the following assumptions: 1) nonresponders had the same rate of cough as responders in both groups (this is a no-selection bias assumption postulating that the RR of cough for the two underlying populations of responders plus nonresponders should be equal to the RR for responders only); 2) responders had 1.5 times the rate of nonresponders for cough in both groups indicating that responders experiencing cough were 50% more likely to respond than those who did not experience the symptom; 3) responders had 2 times the rate of nonresponders in each group; 4) responders had 4 times the rate of nonresponders in each group. A comparison of the RR results, summarized in Table 7, under different assumptions, gives an indication of how selection bias would alter the study results. The results indicate that the potential selection bias would not cause an overestimation of RRs. Under the strictest assumption that the case rate of responders was 4 times of that of nonresponders in both groups, we found that the RR for the underlying population increased slightly to 2.03 from 1.57. This suggests that the underlying RR could be higher than 1.57 if the selection bias was a factor. The explanation of this result lies in the fact that the response rate for the ever exposed group (25%) was more than double the response rate for the never exposed group (12%). Because the selection bias assumption would reduce the estimated case numbers in the nonresponders group, this case number reduction would be more proportionately obvious in unexposed group although the assumptions were the same for both groups.
In addition to the potential for selection bias, there was possibility of information bias that could have resulted if the exposed group had reported differently on their health issues. To address this concern, a comparison of positive responses to a likely non-WTC related health outcome was conducted. For this comparison we selected responses to the question “Have you ever had vision problems?” Of the ever exposed group, 28.4% (59 of 219) reported vision problems whereas 26.9% (315 of 1108) of the never exposed group reported vision problems. The difference was not statistically significant, suggesting that response bias may be limited in this study. An additional source of bias with self-reported data is recall bias, which cannot be quantified from the data itself. However, in this study the respiratory health outcomes reported by the respondents were current LRS, so that recall bias was likely limited. In addition, it is unlikely that the major exposure indicator, ever/never at WTC, was recalled incorrectly.
Monitoring of respiratory health effects during the WTC cleanup operation suggests that inhalation exposure to the debris provoked a response from the upper and lower airways in some workers. Our results suggest an impact on respiratory health related to work experience at the WTC. These results are consistent with the limited information from other studies that indicate that some individuals have suffered respiratory symptoms related to the exposure at the WTC many months after completing work at the site. The extent of the long-term impact, however, remains unknown and indicates continued follow up of WTC worker populations.
This work was supported by supplemental funds from the National Institute of Environmental Health Sciences (NIEHS) to the NIEHS Environmental Health Science Center (P30ESO3819).