Estimated burden of disease attributable to ... - Wiley Online Library

Download PDF
1 downloads 0 Views 767KB Size Report
Comment
May 10, 2012 - Background As part of an effort to strengthen occupational safety and health pro- grams, the United Arab Emirates (UAE) commissioned a study ...
AMERICAN JOURNAL OF INDUSTRIAL MEDICINE 55:940–952 (2012)

Estimated Burden of Disease Attributable to Selected Occupational Exposures in The United Arab Emirates Tiina J. Folley,

Leena A. Nylander-French, PhD, CIH,1 Darren M. Joubert, and Jacqueline MacDonald Gibson, PhD, MS1

1 MS, MSPH,

2 MSc, COH,

Background As part of an effort to strengthen occupational safety and health programs, the United Arab Emirates (UAE) commissioned a study to estimate the burden of disease attributable to occupational exposure to carcinogens, particulate matter, and noise. Methods We developed an innovative simulation model to estimate the occupational disease burden and facilitate future assessments as more field-based quantitative data become available. Results We determined that, in 2008, an estimated 46 deaths (95% CI: 27–71) and 17,000 health-care facility visits (95% CI: 16,000–18,000), along with 4,500 cases of noise-induced hearing loss, were attributable to the occupational risk factors covered in this study. Lung cancer and leukemia were associated with the highest number of deaths (38), whereas asthma and chronic obstructive pulmonary disease contributed most to the health-care facility visits (nearly 16,900). The highest estimated occupational disease burden is in construction. Conclusion These results will help the UAE to institute new policies for environment, health, and safety management. Am. J. Ind. Med. 55:940–952, 2012. ß 2012 Wiley Periodicals, Inc.

KEY WORDS: occupational burden of disease; occupational exposure assessment; risk assessment; relative risk; United Arab Emirates

INTRODUCTION The United Arab Emirates (UAE), a federation of seven emirates on the Arabian Gulf, has been developing at

Additional supporting information may be found in the online version of this article. 1 Department of Environmental Sciences and Engineering, Gillings School of Global Public Health, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 2 Health Authority-Abu Dhabi, Occupational and Environmental Health Section, Health, Safety and Environment Department, Abu Dhabi, United Arab Emirates Disclosure Statement: The authors report no conflicts of interests. *Correspondence to: Leena A. Nylander-French, PhD, CIH, Department of Environmental Sciences and Engineering, Gillings School of Global Public Health, University of North Carolina at Chapel Hill, CB# 7431, Rosenau Hall, Chapel Hill, NC 27599-7431. E-mail: [email protected] Accepted 3 March 2012 DOI10.1002/ajim.22043. Published online10 May 2012 in Wiley Online Library (wileyonlinelibrary.com).

ß 2012 Wiley Periodicals,Inc.

record speed since its founding in 1971. A British agent visiting the UAE’s capital city, Abu Dhabi, not long before 1971 observed that the city ‘‘consists of just barasti (reed) huts, a broken down market . . . and a few buildings put up by the oil company’’ [Davidson, 2009a]. Today, the capital is a modern, steel-and-glass metropolis, with advanced transportation networks, high-end shopping malls, and world-class museums, including branches of the Louvre and the Guggenheim under development. While in the 1960s the UAE was one of the world’s poorest countries, today it boasts a diversified economy that has given its citizens one of the world’s highest per-capita incomes [Rizvi, 1993; Davidson, 2009a,b]. The country has, in essence, transformed from a developing nation to a first-world economic power in just one generation. Davidson [2009a] has written that the UAE, led by Abu Dhabi, ‘‘powered into the 21st century on the back of

Occupational Disease Burden

ever-increasing oil revenues, well-established petrochemical industries, and massive oil-financed overseas investments.’’ While the UAE has by all accounts leapfrogged into the post-industrial age, its environmental and occupational health and safety programs have struggled to keep pace. Growing public concerns about occupational and environmental risks to health, including occupational health studies revealing unsafe work practices [Gomes et al., 1998, 1999, 2001, 2002; Al-Kaabi and Hadipriono, 2003; Al Nahyan, 2009], prompted the Abu Dhabi government to commission a comprehensive assessment of the risks to health of exposures to hazardous physical, chemical, and biological substances in air, water, soil, food, and workplaces [Li et al., 2010; Willis et al., 2010; Davidson et al., 2012; MacDonald Gibson et al., in press]. As part of this process, government agency representatives, scientists, industry representatives, and others in the UAE participated in a focus group process to prioritize environmental risks to health [Willis et al., 2010]. Participants in these focus groups ranked occupational exposures as a very high priority for the country, behind only air pollution in importance, relative to other exposure risks [Willis et al., 2010]. As a consequence of the increasing recognition of the importance of reducing occupational risks, the Environment Agency—Abu Dhabi (EAD) commissioned the University of North Carolina (UNC), Chapel Hill, to quantify the burden of disease due to occupational exposures to hazardous substances, based on available information. The goal was to identify the industrial sectors and health end points that contribute the most to the preventable disease burden, in order to inform the development of improved policies for environmental and occupational health protection. UNC collaborated with the Health Authority—Abu Dhabi (HAAD) on this project as HAAD maintains the most comprehensive public health data available in the UAE. Here, we report the results of our assessment. This assessment provides the first comprehensive estimate of the burden of disease due to occupational exposures to hazardous substances available for the UAE. The assessment is based on a computer simulation model developed for this research that combines data from multiple local and international sources. This model enables computation not only of the expected risks in different economic sectors but also uncertainty in the risk estimates and sensitivity of the estimates to changes in model input data. The sensitivity and uncertainty analyses we provide can help the UAE prioritize the types of occupational health data that are most important to collect in the future to fill gaps in knowledge. Due to the large number of harmful substances present in work environments and the lack of scientific data on exposure prevalence and risk arising from exposure, quantifying the burden of disease resulting from exposure

941

to all possible agents is not feasible. Therefore, the exposures included in this study are based on World Health Organization (WHO) guidelines: the study covers common occupational carcinogens, occupational airborne particulates, and noise [Concha-Barrientos et al., 2004a,b; Driscoll et al., 2004a,b; Nelson et al., 2005a,b]. These three categories of exposures (carcinogens, particulates, and noise) encompass the most common occupational exposures to harmful pollutants and hence are the only occupational pollutant categories for which the WHO has developed guidelines. Based on the strength of evidence and the availability of data on exposure, as well as the number of cases in the UAE population, three well-documented occupational cancers were included: lung cancer, leukemia, and malignant mesothelioma [Concha-Barrientos et al., 2004b]. Similarly, the most prevalent work-related respiratory conditions were included: asthma, chronic obstructive pulmonary disease (COPD), and two types of pneumoconioses (asbestosis and silicosis) [Concha-Barrientos et al., 2004b]. The most significant health outcome resulting from occupational exposure to noise is noise-induced hearing loss. Importantly, even though occupational injuries resulting from accidents, such as falling from a height or being hit by an object, are an important contributor to the occupational burden of disease, work-related accidents, ergonomic problems, and musculoskeletal disorders are outside the scope of this study. In this study, we focused on health risks due to releases of hazardous substances and noise into the environment resulting from human activities.

METHODS Our method follows closely the approach that the WHO outlines in its guidelines for assessing the environmental burden of disease at the national and local levels, which is described in detail in a series of documents [e.g., Concha-Barrientos et al., 2004a; Driscoll et al., 2004a,b]. This method has been applied to a wide variety of settings and studies and published extensively in the public health literature for estimating the public health burden attributable to different kinds of environmental exposures [e.g., ¨ stu¨n Driscoll et al., 2005a; Nelson et al., 2005a,b; Pru¨ss-U et al., 2005; Li et al., 2010]. Even though many health conditions have been associated with exposure to occupational hazards, only a few diseases are caused exclusively by work-related exposures. Examples of the latter health outcomes are pneumoconiosis, such as asbestosis and silicosis; accordingly, for this study, we assumed all reported cases arose from occupational exposures. For all other health outcomes, the proportion of the disease burden caused by work-related exposure must be separated from that attributable to other causes. The WHO approach for calculating this

942

Folley et al.

work-related fraction is based on a long-standing concept from public health: the ‘‘attributable fraction,’’ defined as the proportion of documented cases of a particular health outcome that could be prevented if exposure to the particular risk factor were reduced or eliminated [Coughlin et al., 1994; Rowe et al., 2004; Steenland and Armstrong, 2006]. For any particular risk factor (e.g., occupational exposure to benzene), the attributable fraction (AF) of a given health outcome (e.g., leukemia) is estimated from the following equation [Coughlin et al., 1994]: P Pi RRi  1 (1) AF ¼ P Pi RRi where Pi ¼ proportion of the population in each exposure category ‘‘i’’ (unexposed, low exposure, high exposure) and RRi ¼ relative risk at the exposure category ‘‘i,’’ compared to the reference level (unexposed population). Of note, this calculation is often carried out relative to some counterfactual exposure scenario (e.g., under a new system of occupational exposure controls). In this analysis, we use as a counterfactual scenario the elimination of all occupational exposures to carcinogens, particulate matter, and noise. The observed number of cases of the specific health outcome (e.g., leukemia) in the general population is then multiplied with the appropriate AF to estimate the disease burden attributable to the particular risk factor: Dattrib ¼ AF  Dtotal

(2)

where Dattrib ¼ number of deaths or illnesses attributable to the given risk factor and Dtotal ¼ total number of deaths or illnesses observed in the UAE population as a whole. Our reference year for this assessment is 2008. However, when information for 2008 was not available, we used data that were recorded as close to 2008 as possible. The attributable fraction approach represented in Equations (1) and (2) requires three types of data: (1)

(2)

(3)

the total number of deaths and illnesses in the population of interest, organized according to international classification of disease (ICD) code (i.e., Dtotal for each potential environmentally related disease); the fraction of the population exposed to a given hazard and, within the exposed population, the distribution of exposure levels (e.g., number of people exposed to particulate matter in air at specified concentration ranges), needed to estimate Pi; and the relative risk of developing a specific illness (organized by ICD code) for a specific exposure at a specific level (e.g., the relative risk of developing

asthma per 10 mg/m3 increase in particulate matter exposure concentration), in order to specify RRi. For each category of data, substantial uncertainty and/ or variability in the population may exist. Ideally, the burden of disease predictions, then, will represent the range of possible values, given the uncertainty and variability in the parameters used to estimate the burden of disease. We constructed a simulation model using Analytica software (Version 4.1, Lumina Decision Systems, Los Gatos, CA) to perform the computations for this analysis, so that uncertainty in input variables could be carried through the computations to estimate uncertainty in our predictions. Wherever possible, we considered uncertainties in variables related to the fraction of the population exposed to a given hazard and the relative risks of illness upon exposure by representing variables as probability distributions. Notably, although there is uncertainty in our estimates of the baseline illness rates, primarily due to extrapolating data from Abu Dhabi to the other emirates, characterizing this uncertainty was not possible due to the lack of previous statistical comparisons of baseline health status in the different emirates by demographic group. Monte Carlo simulation with the Median Latin Hypercube sampling method with 1,000 iterations (which produced stable results) produced mean estimates as well as 95% confidence intervals on the predicted health burden. Sensitivity analyses were carried out by replacing one input at a time, while holding all others at their baseline values. The model is designed to be modular, with different modules for different types of exposures, and can be updated easily as improved data become available. The Online Supplement shows the complete structure of the model, lists all its input variables and the values for each used in this analysis, and provides references to the original sources of input data. The following sections provide details on how we developed the three broad categories of data used as inputs to the model (i.e., the data used to estimate Dtotal, Pi, and RRi).

Total Deaths and Illnesses in the Population (Dtotal), By Disease Category No central database tracking illnesses among workers currently exists in the UAE. Furthermore, health tracking is inconsistent across the seven emirates. Hence, we relied on public health data for Abu Dhabi emirate, which has a modern public health record-keeping system and which comprises approximately one-third of the UAE population and more than 80% of its land area. This study did not require an approval from an Institutional Review Board because no human subjects were involved and only publicly available health records were used. We obtained

Occupational Disease Burden

complete death records, with causes of death recorded by ICD-10 code, for the emirate for the year 2008 from HAAD [2009]. In addition, we obtained health insurance records documenting (by ICD code) every visit to a healthcare facility in the year 2008 by Abu Dhabi residents enrolled with the emirate’s major health insurance provider, which covers 73% of the total population [HAAD, 2009]. We used these latter data as a surrogate for morbidity estimates because this was the most accurate and comprehensive database on incidences of illnesses by health outcome available. We extrapolated the patient encounter incidence data to Abu Dhabi as a whole by dividing the rate in each demographic group by 0.73 (reflecting the percentage of the population included in the database). We were not able to express results as illness cases per year, because in some cases a patient might have visited a health provider multiple times for the same illness, but without patient identifiers we could not screen out these repeat visits. We extrapolated healthcare facility visit estimates and mortality estimates to the other emirates based on Abu Dhabi per-capita incidence rates (separated by gender and ethnicity) and population data from the different emirates. We adjusted the projections for the six emirates other than Abu Dhabi to account for demographic differences, since the ratio of expatriates to nationals and males to females varies by emirate (Table I). The assumption that illness rates among demographic groups across the UAE would be similar to those in Abu Dhabi is not unreasonable, since all Emiratis are covered by a government health insurance program; all non-Emiratis employed in Abu Dhabi (where the bulk of expatriates are employed) receive mandatory health insurance coverage through their employers; and climatic conditions are very similar across the emirates. Table II summarizes the total TABLE I. UAE Population by Emirate

Emirate

Total population

Male/female ratio

National/expatriate ratio

Abu Dhabi Dubai Sharjah Ajman Umm Al-Quwain Ras Al-Khaima Fujairah Totals

1,493,000 1,478,000 882,000 224,000 52,000 222,000 137,000 4,488,000

1.92 3.14 1.93 1.80 1.60 1.64 1.69 2.20

0.33 0.11 0.19 0.22 0.44 0.69 0.78 0.24

Source:UAE Ministry of Economy [2008].These population data are based on extrapolations by the Ministry of Economy from a 2005 census. Population data are now maintained in the UAE National Bureau of Statistics (http://www.uaestatistics.gov.ae/ EnglishHome/tabid/96/default.aspx). A population census is currently underway in the UAE (October 2011); when completed, the results will allow for more accurate and current population data.

943

TABLE II. Baseline Cases in the UAE in 2008 for Diseases Considered in This Study

Lungcancer Leukemia Malignantmesothelioma Asthma COPD Asbestosis Silicosis

Number of deaths

Number ofhealth-care facility visits

Males

Females

Males

Females

84 75 6 4 9 0 0

36 54 0 6 28 0 0

389 1,080 28 49,137 19,861 3 8

54 440 0 23,164 7,352 0 0

COPD, chronic obstructive pulmonary disease.

number of deaths and illnesses due to the health endpoints considered in this study (i.e., Dtotal for each health endpoint) other than noise-induced hearing loss. Since noise-induced hearing loss was not included in the insurance claims records, we estimated the baseline prevalence using data from Mathers et al. [2000], which presents the prevalence of adult-onset hearing loss in the WHO Eastern Mediterranean Region based on a study conducted in Oman [Khabori et al., 1996], the UAE’s eastern neighbor, where occupational conditions are expected to be comparable. The hearing loss prevalence presented in Mathers et al. [2000] does not differentiate between the two main causes of adult-onset hearing loss, noise-induced hearing and age-related. We therefore used data from Thorne et al. [2008] to estimate the fraction of hearing loss attributable to noise, rather than age, by age group: 50% for the age group of 30–44 years, 38% for age group of 45–59 years, and 30% for the age groups of 60–69 years and 70–79 years. It should be noted that in the data set by Mathers et al. [2000] used in this study, baseline hearing loss in the age group of 15–29 years was zero. Table III shows the resulting estimates.

Fraction of Population Exposed (Pi) In the ideal scenario, exposure information is determined from measurements in the affected population, for example using personal breathing-zone monitoring of workers’ exposure to air contaminants. However, the UAE does not yet require the routine collection of occupational exposure data. Indeed, a goal of this project was to provide information on the types of occupational exposure information that might be the most important for the UAE to gather, as Abu Dhabi emirate is currently leading a process of substantial development and implementation of systems for occupational health and safety management.

944

Folley et al.

TABLE III. Estimated Baseline Cases of Noise-Induced Hearing Loss in the UAE Population Age group

Males Females

15^29

30^44

45^59

60^69

70^79

0 0

12,600 2,723

6,937 2,023

2,176 1,059

1,302 968

Due to the lack of local exposure data, we used methods recommended by the WHO to estimate UAE occupational exposures; Driscoll et al. [2004a,b] and ConchaBarrientos et al. [2004a] describe these methods in detail, and below we outline the methods for each category of harmful chemical and physical stressors considered in this analysis. In brief, exposures are estimated from a combination of UAE economic and employment data, along with exposure information collected from large studies in Europe and the United States for different economic sectors and job classifications. The latter estimates were adjusted to reflect the fact that exposures in the WHO Eastern Mediterranean Region are likely to be at higher levels than in Europe and the United States, due to the differences in occupational health and safety laws and enforcement mechanisms. While this approach is less than ideal, it is the best that can be accomplished given the lack of exposure data for the UAE. Moreover, it is reasonable to assume, as a starting point, that workers in the UAE are exposed to similar hazardous substances as their counterparts in similar industries and job classifications in other parts of the world—especially since industrial facilities and construction sites in the UAE are frequently operated as joint ventures between Emirati and multinational corporations. For example, FERTIL, a UAE-based agrichemical manufacturer, is co-owned by the Abu Dhabi National Oil Company (ADNOC) and TOTAL, the multinational energy company [Davidson, 2009a]. In the oil and gas industry, all major operations are partnerships between Emirati companies (led by ADNOC) and international firms, including British Petroleum, Camagnie Francaise des Petroles, Royal Dutch Shell, Exxon Mobil, TOTAL, and the Japan Oil Development Company [Davidson, 2009a].

Carcinogen exposure estimates: Occupational carcinogens other than asbestos To estimate occupational exposures to carcinogens, we followed the method of Driscoll et al. [2004a]. Specifically, we estimated the proportion of workers exposed to carcinogens by determining the proportion of male and

female workers in each economic sector (using data from the UAE Ministry of Economy [2009]) and then multiplying these proportions with the estimated proportion of workers exposed to specific carcinogens within each sector. The latter estimates are derived from a large international CARcinogen EXposure database (CAREX) covering over 32 million European Union workers [CAREX, 2006], since currently no carcinogen exposure database exists for the Middle East. The CAREX data are based on exposure information collected in two reference countries, Finland and the United States, and adjusted for 19 European Union countries by a panel of national experts. Following the guidelines of Driscoll et al. [2004b], we assumed that the proportion of workers exposed to each carcinogen in the UAE is similar to that in the CAREX database. Mean proportions of workers exposed to the selected carcinogens within each economic sector in the database are presented in Driscoll et al. [2005b]. Both the UAE Ministry of Economy and CAREX classify the labor force according to the International Standard of Industrial Classification of All Economic Activities (ISIC), a widely used classification scheme developed by the United Nations [2008]. However, the CAREX database is organized according to the Revision 2 of the ISIC scheme, whereas UAE employment data use Revision 3.1. The only major difference is that Revision 3.1 divides the services sector into several subsectors, whereas Revision 2 groups these. Hence, we had to make minor adjustments to the UAE data, grouping several sectors (including hotels and restaurants, real estate, public administration, and others) into the ‘‘Services’’ category. Many diseases, such as cancers, have long latency periods. Therefore, workers exposed in the past are at risk even after moving to another job or retiring. Therefore, past exposures must be considered in the burden of disease estimates, not just exposures from the reference year (2008). Past exposures are accounted for by using an ‘‘occupational turnover (OT)’’ factor, discussed in Concha-Barrientos et al. [2004b]. The proportion of the workforce ever exposed to carcinogens is estimated by multiplying the proportion currently exposed by OT: OT ¼

Pt P0 þ P0  t  ATR  xðP0 þ P0  t  ATRÞ ¼ P0 P0 (3)

where Pt ¼ proportion of workers ever exposed, during time period t, who are still living; P0 ¼ proportion of workers occupational exposed at t ¼ 0; ATR ¼ annual turnover rate; t ¼ working time period; x ¼ estimated death rate over time period t. We calculated a weighted-average mortality rate (x) of 0.0013 in the working age population (15–65 years) using the death rate and population proportion in each age

Occupational Disease Burden

group. Due to the large number of expatriate workers in the UAE, a relatively high range of 10–40% was used for the annual turnover rate and a wide range of 10– 40 years was assumed for the working time, resulting in a mean OT factor of 7. Note that one significant limitation to this approach is that expatriate workers must return to their home countries after their work contracts have expired, so the estimates presented here may not account adequately for occupationally induced illnesses and deaths that occur after workers leave the UAE. However, it is common for workers to stay for many years. According to Al Awad [2010], on average, expatriate workers stay for 7.4 years, with 25% staying for 10 or more years. Following the WHO approach, two levels of exposure were used for the lung carcinogens and leukemogens: high exposure, above the relevant US Occupational Safety and Health Administration (OSHA) Permissible Exposure Limit (PEL), and low exposure, below the PEL [ConchaBarrientos et al., 2004b; Driscoll et al., 2004b]. Since no measured data exist on the levels of carcinogens in UAE workplaces, the proportion of workers in each exposure group was estimated following the approach of ConchaBarrientos et al. [2004b]. Based on the lower prevalence of occupational health and safety programs in the industrializing regions (countries in the WHO subregions B-E, including the UAE), 50% of the exposed workers were estimated to be exposed to high levels and 50% to low levels of carcinogens. To determine the fraction of the total population in each exposure group, the proportions of male and female workers in the low- and high-exposure groups were then multiplied with the proportion of the population in the work force, that is, the economic activity rate of 89% for males and 42% for females [UAE Ministry of Economy, 2009]. The proportion of the population outside the work force was considered the unexposed background group.

Carcinogen exposure estimates: Asbestos A simpler method was used for malignant mesothelioma, since exposure to asbestos is essentially the only cause. However, it must be borne in mind that not all asbestos exposure is occupational, even though most of it is work-related. Calculating an AF using relative risk in the exposed and unexposed populations is not feasible since mesothelioma does not occur in populations not exposed to asbestos. However, the AF for this health outcome has been analyzed in numerous studies, and the results indicate that approximately 90% of cases in males and 25% in females are related to occupational asbestos exposures [Nurminen and Karjalainen, 2001]. We used these AF estimates to calculate the disease burden attributable to occupational asbestos exposures.

945

Particulate matter exposure estimates A wide range of respiratory conditions is associated with work-related exposure to particulate matter in air. The most important non-malignant respiratory diseases include asthma, COPD, and pneumoconiosis. Of the three main pneumoconioses, only asbestosis and silicosis are covered here because exposure to coal dust, causing coal workers’ pneumoconiosis, is not likely to be significant in the UAE. Whereas asbestosis and silicosis are caused exclusively by exposure to asbestos and silica, countless sources of particles may cause occupational asthma and/or COPD, and it would be impossible to estimate exposure patterns and relative risks for all of these substances. Consequently, relative risks are presented in the epidemiology literature for job classifications within industrial sectors rather than for specific occupational agents [Driscoll et al., 2004a]. Overall, the approach for calculating the proportion of UAE workers exposed to particulate matter follows the WHO approach [Driscoll et al., 2004a] and is similar to that used for lung cancer and leukemia, discussed above. For asthma, the proportion of UAE workforce in different occupations within each industry sector was calculated based on information from the UAE Ministry of Economy [2009]. The data were reclassified to match the grouping used in the literature from which the relative risks were derived, following the approach by Driscoll et al. [2004b]. The figures were then adjusted with the UAE economic activity rate to obtain the total male and female population in each occupation-industry category (e.g., sales workers in the construction industry, administrative staff in the financial industry).

Noise exposure estimates Noise is a common risk factor in many work environments, particularly in manufacturing, transportation, mining, construction, agriculture, and military operations. Even though noise-induced hearing loss is completely preventable, it is one of the most prevalent occupational adverse health conditions worldwide. Exposure to noise is commonly measured as A-weighted decibels, dB(A), which takes into consideration the sensitivity of the human ear to sound at different frequencies. Occupational noise exposure is usually divided into three categories that reflect the common regulatory limits, which are 85 dB(A) in most developed countries and 90 dB(A) in many developing nations over an 8-hr work day [Concha-Barrientos et al., 2004a]. Exposure to sound levels less than 85 dB(A) is considered low, exposure to levels of 85– 90 dB(A) is considered moderately high, and exposure to levels above 90 dB(A) is considered high [ConchaBarrientos et al., 2004a]. In the epidemiology literature,

946

Folley et al.

exposure to occupational noise has been estimated for different occupations, shown to be the most important determinant of exposure level [Concha-Barrientos et al., 2004a]. The method for quantifying the disease burden due to occupational noise exposure follows the WHO approach, as described in detail in Concha-Barrientos et al. [2004a]. Similar to the method for estimating exposure to particulate matter, the method is based on multiplying the proportion of UAE workers in each occupational category within each economic sector with the estimated proportion of workers that are exposed to different noise levels in each occupation-industry group. In a study conducted at a Dubai foundry, almost all workers were observed to have exposure above the recommended exposure limit of >85 dB(A) for 8 hr [Gomes et al., 2002]. However, since occupational noise exposure data are not routinely available for UAE workers, we estimated noise exposure prevalence for different occupations in different industries following the WHO recommendations for estimating noise exposures in nations of the WHO’s Eastern Mediterranean Region. The WHO’s recommended exposure estimates for this region are based on extensive data from the United States [CDC, 1986], covering over 9 million production workers, and are modified to reflect working conditions in different WHO subregions, as described in Concha-Barrientos et al. [2004b]. The proportions of UAE workers in each occupation-industry category were derived from information from the UAE Ministry of Economy [2009] and were reclassified to match the occupational categories in the noise exposure data. We estimated that 3.4% of male and 5.4% of female workers were exposed to moderately high [85–90 dB(A)] levels of noise and that 6.6% of male and 1.6% of female workers were exposed to high [>90 dB(A)] levels of noise in the UAE. These worker exposure estimates were then adjusted to reflect the exposure in the general UAE population by multiplying them with the economic activity rate, 0.89 for males and 0.42 for females.

Relative Risks (RRi) Table IV summarizes the relative risk estimates used in this study and provides sources for those estimates. In each case, we followed WHO guidance in choosing relative risk estimates from the international epidemiologic literature. Details on the meta-analyses used to derive these relative risks are summarized in the reference documents listed in Table IV.

RESULTS Figure 1 shows the estimated number of deaths attributable to the occupational exposures considered in this

research. In total, we estimate that about 46 deaths (95% CI: 27–71) per year are attributable to occupational exposures, out of which 39 were in males and 7 in females. In total, 0.52% of the 8,865 deaths from all causes in the UAE in 2008 thus can be attributed to occupational exposures to carcinogens and airborne particulate matter. This estimate can be compared to the estimated number of occupational deaths derived from a previous WHO study focused on the Eastern Mediterranean-B Region, which includes the UAE [Driscoll et al., 2005a,b]. Extrapolating the WHO’s estimate for the region to the UAE by applying the occupational mortality rate that the WHO estimated (for the same end points as are included in our study) to the UAE population produces an estimate of 60 annual occupational deaths—somewhat higher than our estimate but within the 95% confidence interval. Of the health outcomes covered in the study, lung cancer and leukemia were shown to be associated with the highest number of deaths (25 and 12, respectively), accounting for 0.43% of deaths from all causes (Fig. 1). Figure 2 shows the contribution to the burden of excess lung cancer cases by economic sector and contaminant. As shown, the leading sector of concern for fatal lung cancer is construction, with the most important construction exposures (from a health standpoint) estimated to be silica, diesel fumes, and asbestos. Other sectors potentially contributing to the burden of excess lung cancer deaths include mining (due mainly to silica and diesel exposures from quarrying activities), transportation (due to diesel exposures), and manufacturing (due to a variety of carcinogen exposures). In reference to Figure 2, it is worth noting that the contributions of the oil and gas industry, which drive the UAE’s economy, are represented in the manufacturing sector. Unfortunately, the UAE Ministry of Economy data available for this analysis do not divide manufacturing workers into subcategories that would enable a more careful investigation of the contributions of the oil and gas sector, in comparison to other manufacturing industries. Figure 3 shows the number of annual medical visits attributable to the exposures considered in this study, along with the number of cases of noise-induced hearing loss (which may or may not involve medical visits). In total, we estimate that 17,000 (95% CI: 16,000–18,000; 16,000 in males, and the remainder in females) medical visits and approximately 4,500 cases of noise-induced hearing loss are attributable to occupational exposures. The leading health outcomes necessitating medical visits are asthma and COPD, with 12,000 and 5,000 visits, respectively. Occupationally related medical visits for asthma and COPD account for 28% of the medical visits among adults to treat these conditions (59,600 visits in total). Figure 4 compares our AFs for several of the health endpoints to WHO estimates for the Eastern

Occupational Disease Burden

947

TABLE IV. Relative Risk Estimates Used inThis Study Exposure variable Occupational carcinogens andleukemogens

Health endpoint Lungcancer mortality and morbidity

Leukemia mortality and morbidity

Particulatematter

Relative risk Background:1 Low exposure: normal (1.21,0.0148), High exposure: normal (1.77,0.0316), Background:1 Low exposure: normal (1.9,0.15) High exposure: normal (4,0.2)

Sources of data Driscolletal.[2004b,2005b], Nurminen and Karjalainen [2001], Steenland et al. [1996, 2003] Driscoll et al. [2004b],IARC [1997],Lyngeetal.[1997], Steenland et al. [2003]

Malignantmesothelioma

A relative risk for malignantmesothelioma in exposed versusnon-exposedpopulation is not available since mesothelioma does notoccurin populationsthathave notbeen exposed to asbestos.Previous studies have estimatedthat90%ofmesotheliomainmalesand25%in females is relatedto occupational exposure to asbestos

Nurminen and Karjalainen [2001], Steenland et al. [2003]

Asthma mortality and morbidity

Occupation group Administration Male:1 Female:1 Technical Male: normal (1.05,0.0357) Female: normal (1.06,0.0204) Sales Male: normal (1.1,0.0663) Female: normal (1.13,0.0255) Agriculture Male: normal (1.41,0.3112) Female: normal (1.41,0.3112) Mining Male: normal (1.95,0.2296) Female: normal (1,1.5408) Transportation Male: normal (1.31,0.0459) Female: normal (1.22,0.0459) Manufacturing Male: normal (1.56,0.0459) Female: normal (1.33,0.0306) Services Male: normal (1.53,0.1226) Female: normal (1.41,0.0255)

Driscoll et al. [2004a], Karjalainen et al. [2001], Kogevinas et al. [1999]

Chronic obstructive pulmonary disease (COPD) mortality and morbidity

Background:1 Low exposure Male:1.2 Female:1.1 High exposure Male:1.6 Female:1.4

Driscoll et al. [2004a], Korn et al. [1987]

(Continued )

948

Folley et al.

TABLE IV. (Continued ) Exposure variable

Noise

Health endpoint

Relative risk

Sources of data

Asbestosis and silicosis

100% attributablefraction (asbestosis and silicosis are almostsolely causedbyoccupational exposure to asbestos and silica,respectively)

Driscoll et al. [2004a]

Hearingloss

Age groups 15^29 years Moderately high [85^90 dB(A)]:1.96 High [>90 dB(A)]:7.96 30^44 years Moderately high: 2.24 High: 5.62 45^59 years Moderately high:1.91 High: 3.83 60^69 years Moderately high:1.66 High: 2.82 70^79 years Moderately high:1.66 High: 2.82

Concha-Barrientos et al. [2004a]

 Normal (a, b) indicates a Gaussian normal distribution with mean a and standard deviation b. 

Following the WHO method [Driscoll et al., 2004a, 2005b], a mean relative risk of1.49 was calculated for lung carcinogens by weighting each relative risk with the proportion of the UAE workforce exposed to that carcinogen and summing these weighted relative risks. Avariance for the mean relative risk was calculated as a weighted sum of the variances of the relative risk for each carcinogen, where the weights are the squares of the proportion of the workforce exposed to that carcinogen (assuming the relative risks for each carcinogen are uncorrelated). Separate relative risks for low (1.21) and high (1.77) exposure are then computed from the mean (1.49) relative risk following the approach by Driscoll et al. [2005b], in which the ratios of the low and high relative risks to the average U.S. relative risk are applied to produce low and high relative risks for other regions [Driscoll et al., 2005b].  Information was insufficient to characterize the uncertainty in this relative risk. Hence, the relative risk is modeled as a deterministic variable rather than as a random variable.

FIGURE 1. Estimated numbers of deaths attributable to occupational exposures considered in this study for the year 2008. Error bars indicate 95% confidence intervals. (For some health outcomes, data were insufficient to characterize uncertainty in input variables, and hence the estimates are deterministic.)

Mediterranean-B region [Concha-Barrientos et al., 2004b]. UAE AFs for males are higher than for males elsewhere in the region, as estimated by the WHO, while estimates for females are generally comparable, with the exception of leukemia, for which the UAE AF appears higher than regional estimates. Figure 5 shows a similar comparison for noise-induced hearing loss, except that the comparison population is global, rather than regional. As shown, overall, the AFs for noise-induced hearing loss for males in the UAE are comparable to those worldwide; for females, noise-induced hearing loss AFs are lower in the UAE than globally, probably because females in the UAE do not commonly work in occupations with potential noise exposure, and the economic activity rate for UAE females (41.8%) is comparatively low. Figure 6 shows estimated contributions of various economic subsectors to occupational asthma, the leading occupational cause of medical visits in the UAE. As shown, the leading sectors of concern are the services and construction industries, with construction workers at highest risk compared to other occupational groups.

Occupational Disease Burden

FIGURE 2. Estimated contribution to fatal lung cancer cases by economic subsector and contaminant.

949

FIGURE 4. Comparison of UAE attributable fractions (i.e., fraction of the total disease burden attributable to occupational exposures) estimated for this study with those estimated by the WHO for the Eastern Mediterranean-B region (EMR-B) [Concha-Barrientos et al., 2004b]. Error bars show 95% confidence intervals.

To determine which types of information would have the greatest effect on the occupational disease burden estimates, we conducted sensitivity analyses to see how our estimates for the most common occupational health outcomes would change if the key input variables were to change. Figure 7 shows the results for occupational lung cancer deaths. The figure indicates the change in the median estimated number of deaths with a 25% decrease or increase in the value of each key input variable—the wider the bar, the more responsive is the estimate to the value of the input variable. As shown, changing the assumed relative risk values would change the overall estimate by the most as compared to other input variables; a 25% increase in relative risk over the values used in this analysis would increase the estimated number of deaths by 19 (an

increase of 72% over the original estimate), whereas a decrease in relative risk by 25% would decrease the estimated number of deaths by 15 (nearly a 60% decrease). Results for the other health outcomes are similar. That is, local relative risk estimates would have the biggest impact on the ability to estimate the occupational burden of disease. Developing such local estimates would require epidemiologic studies, comparing workers exposed to different levels of pollutants and in different occupations with unexposed groups having similar demographic characteristics. Such studies would require both exposure monitoring and health status monitoring for large groups of workers in the key economic sectors.

FIGURE 3. Estimated medical visits attributable to occupational exposures considered in this study for the year 2008. For noise-induced hearing loss, the estimates indicate expected number of cases, as medical visit data on this health outcome were unavailable. Bars indicate 95% confidence intervals, for outcomes with sufficient data to

FIGURE 5. Comparison of UAE attributable fractions for noise induced hearing loss

characterize uncertainty.

with global estimates prepared by the WHO [Driscoll et al., 2005a,b].

950

Folley et al.

FIGURE 6. Contributions of economic subsectors and job classifications within those subsectors to medical visits for asthma attributable to occupational exposures.

DISCUSSION These results can assist the UAE as it begins to identify the economic sectors, job categories, and pollutants for which more stringent occupational health and safety mechanisms or stronger enforcement of existing requirements would have the greatest public health benefit. Laudably, UAE’s leadership has expressed a commitment to strengthening its occupational health and safety protection systems. A royal decree 2009 required any entity (company, facility, establishment, or group) undertaking ‘‘an activity that potentially impacts the environment and/or the workforce and community’’ to ‘‘develop and implement an Environment, Health, and Safety Management System

FIGURE 7. Results of sensitivity analysis for occupational lung cancer mortality estimates. The figure shows the amount by which the estimated number of deaths would change if the indicated input variable were changed by 25% (left side) or þ25% (right side). As shown, the relative risk estimate has by far the greatest impact on these estimates.

(EHSMS) within their scope of work to protect workers, society and the environment from any adverse impacts that may result from their activities’’ [Al Nahyan, 2009]. Abu Dhabi subsequently established the Environment, Health, and Safety Center to coordinate implementation of this decree. According to the results reported here, the economic sectors with the highest potential occupational burden of disease (other than accidents and injuries) are the construction and services sectors. According to the ISIC classification, the services sector is broadly defined to include hotels and restaurants, real estate, business services, personal services (such as beauty salons), public administration, military services, health and social work, community organizations, activities of private households as employers, and unstated activities [United Nations, 2008]. While this classification is very broad, it appears that those involved in production and manufacturing of goods in this sector are most at risk (see the top-most bar in Fig. 6). As Abu Dhabi leads the development and implementation of EHSMS programs, it may be worthwhile to emphasize the inclusion of workers involved in service industry activities, due to the large potential share of the estimated occupational disease burden arising from them. The estimates presented here are based on local UAE workforce data from the UAE Ministry of the Economy. However, observed mortality and morbidity rates for emirates other than Abu Dhabi had to be estimated using documented rates from Abu Dhabi, scaled to reflect demographic differences between Abu Dhabi and the other emirates. Furthermore, the exposure levels and relative risks had to be estimated, following WHO guidelines, from meta-analyses in the international occupational exposure and epidemiologic literature, because the UAE does not as yet systematically collect occupational exposure or occupational health data. As a result, as part of improving its occupational health and safety system, the UAE could identify economic sectors in which to begin collecting such exposure and health information across all emirates. Once such information is available, these estimates can be updated by changing the parameter values in the burden of disease model (detailed in the Online Supplement) to reflect the new estimates. Ideally, as well, in the future the UAE would further stratify its occupational exposure data by subsector so that, for example, the disease burden attributable to specific industries (such as the oil and gas industry) could be estimated. This additional stratification could be accomplished by adding new subsector categories to the existing burden of disease model. Despite these limitations, a main advantage of this study is that it employs a risk assessment framework consistent with that employed in other studies around the world and thus provides a basis for the UAE to compare itself against other countries, as well as to compare the

Occupational Disease Burden

burden of disease within economic sectors in the UAE. Indeed, the Abu Dhabi Executive Council, which is comprised of ministers of key government departments and others appointed by the Ruler of Abu Dhabi, has declared that its vision is ‘‘to facilitate Abu Dhabi to be regarded as one of the best five governments in the world.’’1 These results highlight specific areas in the occupational sector in which the UAE can take action to make progress towards this ambitious goal. Taking steps to control and better characterize occupational health risks in the UAE could pay off in the long run not only by helping the UAE achieve its vision of excellence in governance but also be reducing overall government spending on healthcare. For example, Al Awad [2010] estimates that the UAE government spends $5.4 billion on health services of citizens and another $1 billion for non-citizens each year. Reducing exposure to harmful substances in occupational environments could help to decrease these costs in the future and certainly would reduce the burden on the health sector in the UAE.

ACKNOWLEDGMENTS This study was funded by Environment Agency—Abu Dhabi (EAD). We would like to thank H. E. Majid Al Mansouri for initiating this work and H. E. Razan Al Mubarak and Dr. Fred Launay for their continuing support. We wish to express special thanks to Dr. Jens Thomsen at HAAD and Dr. Tar-Ching Aw and Dr. John Schneider at the UAE University for their invaluable insight on worker exposures in the UAE. We thank the UAE Ministry of the Economy for providing employment data. Appreciation is also due to Dr. Zeinab Farah and the EAD staff members for assisting with data collection in the country. We would also like to thank Chris Davidson and other colleagues at UNC for their technical advice.

REFERENCES Al Awad M. 2010. The cost of foreign labor in the United Arab Emirates. Dubai: Zayed University, Institute for Social & Economic Research. p 51. Al Nahyan HEM. 2009. Abu Dhabi Environment, Health and Safety Management System Regulatory Framework (EHSMS) Decree. Abu Dhabi: Environment Agency, Executive Council, Abu Dhabi. Al-Kaabi N, Hadipriono FC. 2003. Construction safety performance in the United Arab Emirates. Civ Eng Environ Syst 20:197–212. CAREX. 2006. CAREX International Information System on Occupational Exposure to Carcinogens. Helsinki: Finnish Institute of Occupational Health. Available: http://www.ttl.fi/en/chemical_safety/ carex/Pages/default.aspx (accessed October 24, 2011).

951

CDC. 1986. Perspectives in disease prevention and health promotion, leading work-related diseases and injuries—United States: MMWR Weekly 35:185–188. Concha-Barrientos M, Campbell-Lendrum D, Steenland K. 2004a. Occupational Noise: Assessing the Burden of Disease from Workrelated Hearing Impairment at National and Local Levels. Environmental Burden of Disease Series, No. 9. Geneva: World Health Organization. Available: http://www.who.int/quantifying_ehimpacts/ publications/en/ebd9.pdf (accessed October 25, 2012). Concha-Barrientos M, Nelson D, Driscoll T, Steenland N, Punnett L, ¨ stu¨n A, Corvalan C, Leigh J, Tak S. Fingerhut M, Pru¨ss-U 2004b. Selected occupational risk factors. In: Ezzati M, Lopez A, Rodgers A, Murray C, editors. Comparative quantification of health risks: Global and regional burden of disease attributable to selected major risk factors. Geneva: World Health Organization. p 1651–1802. Coughlin SS, Benichou J, Weed DL. 1994. Attributable risk estimation in case-control studies. Epidemiol Rev 16:51–64. Davidson CA, Krometis LA, Al-Harthi SS, Gibson JM. 2012. Foodborne exposure to pesticides and methylmercury in the United Arab Emirates. Risk Anal 32:381–394. Davidson CM. 2009a. Abu Dhabi’s new economy: Oil, investment and domestic development. Middle East Policy 16:59–79. Davidson CM. 2009b. Abu Dhabi: Oil and beyond. New York: Columbia University Press. 256 p. Driscoll T, Steenland K, Nelson DI, Leigh J. 2004a. Occupational Airborne Particulates: Assessing the Environmental Burden of Disease at National and Local Levels. Environmental Burden of Disease Series No. 7. Geneva: World Health Organization. Available: http:// www.who.int/quantifying_ehimpacts/publications/9241591862/en/ index.html (accessed October 25, 2011). ¨ stu¨n A, Nelson DI, Leigh J. 2004b. Driscoll T, Steenland K, Pru¨ss-U Occupational Carcinogens: Assessing the Environmental Burden of Disease at National and Local Levels. Environmental Burden of Disease Series No. 6. Geneva: World Health Organization. Available: http://www.who.int/quantifying_ehimpacts/publications/9241591471/ en/index.html (accessed October 25, 2011). Driscoll T, Nelson DI, Steenland K, Leigh J, Concha-Barrientos M, ¨ stu¨n A. 2005a. The global burden of nonFingerhut M, Pru¨ss-U malignant respiratory disease due to occupational airborne exposures. Am J Ind Med 48:432–445. Driscoll T, Nelson DI, Steenland K, Leigh J, Concha-Barrientos M, ¨ stu¨n A. 2005b. The global burden of disease Fingerhut M, Pru¨ss-U due to occupational carcinogens. Am J Ind Med 48:419–431. Gomes J, Lloyd O, Revitt MD, Basha M. 1998. Morbidity among farm workers in a desert country in relation to long-term exposure to pesticides. Scand J Work Environ Health 24:213–219. Gomes J, Lloyd OL, Revitt DM. 1999. The influence of personal protection, environmental hygiene and exposure to pesticides on the health of immigrant farm workers in a desert country. Int Arch Occup Environ Health 72:40–45. Gomes J, Lloyd OL, Norman NJ, Pahwa P. 2001. Dust exposure and impairment of lung function at a small iron foundry in a rapidly developing country. Occup Environ Med 58:656–662. Gomes J, Lloyd O, Norman N. 2002. The health of the workers in a rapidly developing country: Effects of occupational exposure to noise and heat. Occup Med (Lond) 52:121–128. HAAD. 2009. 2008 health data for Abu Dhabi emirate. Abu Dhabi: Health Authority—Abu Dhabi.

1

See http://www.ecouncil.ae/Sites/GSEC/Navigation/EN/root.html (accessed January 23, 2012).

IARC. 1997. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans: Silica, Some Silicates, Coal Dust and Para-aramid

952

Folley et al.

Fibrils. Lyon: International Agency for Research on Cancer, World Health Organization. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Available: http://monographs.iarc.fr/ ENG/Monographs/PDFs/index.php (accessed October 25, 2011). Karjalainen A, Kurppa K, Martikainen R, Klaukka T, Karjalainen J. 2001. Work is related to a substantial portion of adult-onset asthma incidence in the Finnish population. Am J Respir Crit Care Med 164:565–568. Khabori M, Mohammed A, Khandekar R, Prakesh N. 1996. National Survey for Causes of Deafness and Common Ear Disorders in Oman. Oman Ear Study (OES ’96) survey report. Kogevinas M, Anto JM, Sunyer J, Tobias A, Kromhout H, Burney P. 1999. Occupational asthma in Europe and other industrialised areas: A population-based study. European Community Respiratory Health Survey Study Group. Lancet 353:1750–1754. Korn RJ, Dockery DW, Speizer FE, Ware JH, Ferris BG Jr. 1987. Occupational exposures and chronic respiratory symptoms. A population-based study. Am Rev Respir Dis 136:298–304. Li Y, Gibson JM, Jat P, Puggioni G, Hasan M, West JJ, Vizuete W, Sexton K, Serre M. 2010. Burden of disease attributed to anthropogenic air pollution in the United Arab Emirates: Estimates based on observed air quality data. Sci Total Environ 408:5784– 5793.

Nelson DI, Nelson RY, Concha-Barrientos M, Fingerhut M. 2005b. The global burden of occupational noise-induced hearing loss. Am J Ind Med 48:446–458. Nurminen M, Karjalainen A. 2001. Epidemiologic estimate of the proportion of fatalities related to occupational factors in Finland. Scand J Work Environ Health 27:161–213. ¨ stu¨n A, Rapiti E, Hutin Y. 2005. Estimation of the global Pru¨ss-U burden of disease attributable to contaminated sharps injuries among health-care workers. Am J Ind Med 48:482–490. Rizvi SNA. 1993. Tents to high rise: Economic development of the United Arab Emirates. Middle East Stud 29:664–678. Rowe AK, Powell KE, Flanders WD. 2004. Why population attributable fractions can sum to more than one. Am J Prev Med 26:243–249. Steenland K, Armstrong B. 2006. An overview of methods for calculating the burden of disease due to specific risk factors. Epidemiology 17:512–519. Steenland K, Burnett C, Lalich N, Ward E, Hurrell J. 2003. Dying for work: The magnitude of US mortality from selected causes of death associated with occupation. Am J Ind Med 43:461–482. Steenland K, Loomis D, Shy C, Simonsen N. 1996. Review of occupational lung carcinogens. Am J Ind Med 29:474–490.

Lynge E, Anttila A, Hemminki K. 1997. Organic solvents and cancer. Cancer Causes Control 8:406–419.

Thorne PR, Ameratunga SN, Stewart J, Reid N, Williams W, Purdy SC, Dodd G, Wallaart J. 2008. Epidemiology of noise-induced hearing loss in New Zealand. N Z Med J 121:33–44.

MacDonald Gibson JA, Launay F, Thomsen J, Brammer A, Folley T, Davidson C. (in press). Environmental burden of disease assessment: A case study in the United Arab Emirates. Dordrecht, The Netherlands: Springer.

UAE Ministry of Economy. 2008. UAE in Figures 2007. Abu Dhabi: United Arab Emirates Ministry of Economy, Central Statistics Department.

Mathers C, Smith A, Concha M. 2000. Global burden of hearing loss in the year 2000. World Health Organization Health Statistics and Health Information Systems: The Global Burden of Disease (GBD). Nelson DI, Concha-Barrientos M, Driscoll T, Steenland K, Fingerhut ¨ stu¨n A, Leigh J, Corvalan C. 2005a. The globM, Punnett L, Pru¨ss-U al burden of selected occupational diseases and injury risks: Methodology and summary. Am J Ind Med 48:400–418.

UAE Ministry of Economy. 2009. Statistic Abstract 2008—Employment. Abu Dhabi: United Arab Emirates Ministry of Economy. United Nations. 2008. International Standard Industrial Classification of All Economic Activities, Revision 3.1. Available: http://unstats.un. org/unsd/cr/registry/regcst.asp?Cl¼17 (accessed October 25, 2011). Willis HH, Gibson JM, Shih RA, Geschwind S, Olmstead S, Hu J, Curtright AE, Cecchine G, Moore M. 2010. Prioritizing environmental health risks in the UAE. Risk Anal 30:1842–1856.