Occupational Burden of Asbestos-related Cancer in Argentina, Brazil, Colombia, and Mexico

Findings: Using mortality data of the World Health Organization Health Statistics database for the year 2009 and applying the estimated values for population attributable fractions, the number of estimated deaths in 5 years for mesothelioma and for lung, larynx, and ovary cancers attributable to occupational asbestos exposures, were respectively 735, 233, 29, and 14 for Argentina; 340, 611, 68, and 43 for Brazil; 255, 97, 14, and 9 for Colombia, and 1075, 219, 18, and 22 for Mexico.


INTRODUCTION
The purpose of this study was to tentatively estimate the asbestos occupational burden of cancer in Argentina, Brazil, Colombia, and Mexico, in the frame of scientific cooperation activities envisaged by the Italian National Asbestos Project. The project aimed at developing collaborations with Latin America countries where asbestos use is still permitted or only recently banned, as discussed by Marsili et al. 1 Neoplasms causally associated with asbestos are mesothelioma of pleura, peritoneum, pericardium, and tunica vaginalis testis and cancer of the lung, larynx, and ovary. 2 Asbestos is the only recognized cause of mesothelioma, together with some asbestiform mineral fibers. In industrialized countries such as Italy, more than two-thirds of all mesothelioma cases are associated with documented This study was funded by the Italian National Asbestos Project (www.iss.it/ amianto). RP contributed to all phases of the study: design, data collection, data analysis, and manuscript preparation. BT contributed to discussion of the study design, revising the results, and manuscript preparation. DM contributed to revising the results and manuscript preparation. PC contributed to revising the results and in manuscript preparation. All authors read and approved the final manuscript.
http://dx.doi.org/10.1016/j.aogh.2014.09.003 asbestos exposure in the workplace. 3 In general terms, the number of mesothelioma cases within a population can be directly retrieved from mortality data, if reliable, or from pathology registries, where available. This figure, furthermore, also can be indirectly estimated from data on asbestos consumption in the population of interest. 4 Contrary to mesothelioma, the etiology of pulmonary, laryngeal, and ovarian cancers is multifactorial. For each outcome, the number of cases that would be prevented if exposure to asbestos were eliminated can be estimated by using the population attributable fraction (PAF). Within a project on occupational carcinogens commissioned by World Health Organization, 5 a multistep approach to estimate national PAFs from data on the workforce and a number of assumptions regarding exposure were established. Additionally, limited to lung cancer, PAF can also be estimated using a reasonable ratio between mesothelioma and occupational lung cancer cases. This ratio is time-and place-specific and is largely determined by the effectiveness of antismoking campaigns and by the type of asbestos used. In the United Kingdom, it has been estimated that for every mesothelioma death, between two-thirds and one asbestos-related lung cancer death occur. 6 A more recent meta-analysis of occupational cohort studies concluded that all types of asbestos, except crocidolite, kill at least twice as many people through lung cancer than through mesothelioma. 7 However, for chrysotile, still widely consumed today, the number of asbestosrelated lung cancers cannot be robustly estimated from few mesothelioma deaths.
Both approaches have been used in the present study. A major problem in Latin America is the quality of mortality statistics regarding mesothelioma, whose diagnosis may be problematic. Worldwide, a sizable number of pleural mesotheliomas (C45.0 code in the 10th revision of the International Classification of Diseases) are wrongly certified as "pleural cancers excluding mesothelioma," corresponding to the C38.4 code. 8,9 A few studies performed in Brazil 10 and Mexico 11 suggested that the recognition of deaths caused by mesotheliomas in those countries is far from being satisfactory. Changes in time of the absolute number of death from C45 and C38.4 in the countries of interest are described in Figure 1.

METHODS
As stated earlier, the methodological approach was derived from an earlier study. 5 The number of cases of a given disease due to a given exposure is the product between the PAF for the given exposure and the total number of cases of that disease.
The formula used to calculate the PAF considering different levels of exposure is: where Pi is the proportion of the exposed population in the exposure category i. RRi is the risk for the disease in the exposure category i relative to the risk in the reference exposure category (i.e., the population not exposed).
Exposures to asbestos may occur as direct occupational, indirect occupational, environmental occupational, extra-occupational in living environments, environmental-residential. 12 For occupational settings, it is possible to estimate the proportion of workers exposed in the workplace in different economic sectors. The PAF for cancers of the lung, larynx, and ovary associated with asbestos occupational exposures has been estimated using the following data: Proportion of the workforce employed in each sector Proportion of workers exposed to asbestos in each sector Occupational turnover Levels of exposure Proportion of the population in the workforce Relative risk for each considered disease for different levels of exposure (when available)

Proportion of the Workforce Employed in Each Sector
Data from the ILOSTAT database of the Intenational Labour Organization were used ( Table 1).

Proportion of Workers Exposed to Asbestos in Each Economic Sector
No estimates are available for the countries of interest. For western European and North American countries, data are available from the CAREX (carcinogen exposure) database, which estimated the workforce exposed to a number of carcinogens in different occupational sectors in 1990 to 1993. 13,14 The proportions of the CAREX database for asbestos exposure (Table 2) have been applied to Argentina, Brazil, Colombia, and Mexico, under the assumption of an overall similarity of working procedures in today's Latin American and yesterday's European countries. This assumption is essentially based on the knowledge of asbestos consumption trends in western Europe and in Latin American countries. In western Europe, asbestos consumption began in the first decades of the 20th century and the asbestos industry reached its greatest expansion in the 1960s; although in Latin American countries, asbestos consumption began in the early 1960 reaching its greatest consumption in the 1980s, 1990s. 15 This corresponds to a lag time that can be generally assumed in 20 to 30 years.

Occupational Turnover
The interval between start of exposure and mesothelioma onset is measured in decades. Risk persists well after exposure ceases and well after the minimum latency is reached. For any job position, a turnover of people occurs also for reasons not necessarily related to health. Therefore, at any time, people at risk for developing cancer caused by a certain exposure are all those who have been exposed in the past and have met the minimum latency period. Direct calculation of the turnover is not straightforward and varies depending on the age of the individuals, the annual turnover in each sector, and the life expectancy of population in the country. A turnover factor of 4 was previously proposed 16 and adopted. 17 The same has been adopted as the factor to be multiplied by the present proportions of individuals considered at risk.

Levels of Exposure
The intensity of exposure differs within and between different economic sectors. To the best of our knowledge, in the past, the production of data on asbestos concentration in work settings in the 4 countries of interest has been extremely limited if not negligible. The levels of asbestos exposure that have been assumed in the present exercise were previously presented 17 on the basis of national indicators of human development. Two levels of exposure have been introduced (high and low) assuming that 50% were highly exposed and 50% lowly exposed.

Proportion of the Population in the Workforce
The size of the economically active population is available from administrative data sources. In the present approach, a single value that is the economically active population for the whole population (men and women combined) aged 15 years or older has been used. In 2009, the proportion of the population in the workforce was 0.55 for Argentina, 0.63 for Brazil, 0.59 for Colombia, and 0.57 for Mexico. These data were obtained from the previously mentioned ILOSTAT database dividing the employed population over age 15 by the total population over age 15. Considering only women in order to compute estimates for ovary cancers, the corresponding proportions were 0.44 in Argentina, 0.52 in Brazil, 0.46 in Colombia, and 0.41 in Mexico.

Relative Risk of Each Considered Disease for Each Level of Exposure
The relative risks due to asbestos exposure for cancers of the lung, larynx, and ovary have been retrieved from the scientific literature, in particular from metanalyses (Table 3).
For each of the 4 countries taken into consideration, the total numbers of deaths from mesothelioma and lung cancer in 2009 were retrieved from the WHO health statistics database (http://www.who.int/healthinfo/ statistics/mortality_rawdata/en/index.html). For lung cancer, the number of cases attributable to asbestos also has been estimated using the proportion mesothelioma to lung cancer as 1:1 and 1:2. This exercise is focused on occupational exposures, therefore the estimate of asbestos attributable lung cancers has been reduced by one-third on the basis of evidence from industrialized countries, such as Italy, 3 showing that nonoccupational mesothelioma cases, including those related to environmental and domestic exposures, and those with limited or no exposure information, can sum to about 30%.

RESULTS
For each country of interest and for each cancer site, the number of cases attributable to asbestos exposure in the  workplace was obtained by multiplying the specific PAF by the total number of cases (Table 4). For mesothelioma, the estimated number of deaths in 5 years was based on the official mortality figure provided by the WHO. These figures includes all 4 topographic sites, pleura (the great majority), peritoneum (up to about 10% of total cases), pericardium, and tunica vaginalis testis (extremely rare). 3 For mesothelioma, the absolute number of cases reasonably can be considered as an indicator of asbestos burden of disease. For the other neoplasms, lung, larynx, and ovary cancers, the estimated number of deaths in 5 years is based on the WHO 2009 figure, whereas the number of attributable cases is modeled with PAF method and, only for lung cancer, with the mesothelioma-to-lung cancer ratio method.
The difference between the estimates of the number of occupational lung cancer deaths obtained and the 2 methods are difficult to interpret. In Brazil, the PAF method led to estimates somewhat higher than those produced by the ratio method whereas the reverse occurred in the other 3 countries (with differences up to 5-fold in Mexico). An effect of the weaknesses of both methods and intercountry differences in the reliability of the background statistics cannot be excluded. However, the peculiarity of the estimates in Brazil also might reflect the limited use of amphiboles, which were been made in that country. 1 It is known that the differential in carcinogenic potency between amphiboles and chrysotile is greater for mesothelioma than for lung cancer. 7

DISCUSSION
Estimating the occupational burden of asbestos-related cancer at the national level is a worthwhile exercise. However, when applied to the countries considered in the present study, these estimations must rely on a number of assumptions open to debate. Furthermore, in Latin American countries, the databases required for this exercise are limited or nonexistent and they have not been submitted to adequate control in terms of quality and exhaustiveness. Despite its limitations, the present study suggests that a sizable number of occupational asbestosrelated cancers (in the order of a few to several hundreds per 5 years) occur in each country. As shown in the study, these numbers are almost certainly underestimated.
Two conclusions appear to be warranted in light of this exercise. One is the urgent need for remediation of asbestos contamination of the environment (and a national asbestos ban in Brazil, Colombia, and Mexico). The other is the improvement of the quality of asbestosrelated environmental and health data. The approach 5 that we adopted in this study has not been applied, as far as we know, to countries that have extensively produced and used asbestos. This may depend, presumably, on the fact that political decisions on the ban of asbestos use had been taken before 2004, the year of publication of Driscoll et al.'s work, 5 on the basis of the existing available evidence. Basic figures to be used in burden-ofdisease computations are reported in the literature for mesothelioma and lung cancer. Most population-attributable fractions for lung cancer following occupational asbestos exposure ranged in between 2 and 10, with peak values of 20 to 50, in the male population of European countries; 21 the occurrence of mesothelioma cases is well monitored, and annual incidence rates of 1 to a few cases per 100,000 are generally estimated, 12 indicatively one order of magnitude over those of Latin American countries. Figures of time trend of asbestos consumption in countries where its use is still legal supports the notion of future outbreaks of mesothelioma and other asbestos-related diseases.
The estimates provided in this study must be considered as exploratory, but they also intend to stimulate the construction of "local" databases regarding the extent to which asbestos is or was present in the occupational environment.
The proportion of the workforce exposed to asbestos in different occupational contexts applied to the present case study are those of the CAREX database, which provide prevalence of exposure to asbestos by economic sector for western European countries. In each economic sector, prevalence of exposure may be specific in Latin American countries and may have been different form those in western European countries in the 1990s because of differences in industrial cycles and in use of asbestos products. Therefore, it would be preferable and desirable to have such proportions estimated through surveys implemented in each country. Estimates of the PAFs in the countries of interest would greatly benefit by more precise data on the proportion of workers exposed to asbestos in each productive sector, on the occupational turnover, and on levels of exposure.
Methodological advances in case ascertainment and exposure assessment might improve the validity of the modeled estimates, thus providing a more solid input to decision-making processes aimed at preventing asbestosrelated disease.