Burden of occupational cancer in Brazil and federative units, 1990-2019

ABSTRACT: Objective: To analyze the spatiotemporal distribution of the burden of occupational cancer in Brazil and federative units between 1990 and 2019. Methods: Data were extracted from the Global Burden of Disease (GBD) study. Deaths from cancer whose attributable risk factor was occupational carcinogens were considered. Spatial analysis was performed with the first and last years of the series (1990 and 2019). Age-adjusted mortality rates were used to estimate the global Moran’s Index (Moran’s I), and the local indicator of spatial association (LISA) to identify clusters in the country with the respective statistical significance. The occupational cancer mortality rate, adjusted for age, was analyzed based on its trend for Brazil and federative units, in the period between 1990 and 2019. Results: Between 1990 and 2019, occupational cancer mortality rate showed a decreasing trend (R2=0.62; p<0.001) as well as the burden of disease indicator — DALY (R2=0.84; p<0.001). However, mortality is increasing in most states, suggesting that a minority of federative units induce the country’s global trend. There is also the development of a spatial pattern of autocorrelation, indicating clusters of states with low mortality and DALY rates in the Northeast and high values in the South of the country. Conclusion: The overall decreasing trend in the trend of occupational cancer masks the heterogeneity across states. This scenario may be associated with the diversity of economic activities, and suggests a decentralized and equitable plan for occupational cancer surveillance.


INTRODUCTION
Cancer is a noncommunicable chronic disease with worldwide distribution and a still poorly-defined etiology, being considered a public health issue in developed and developing countries 1 . In 2018, global burden estimates showed that 18 million new cases of cancer occurred worldwide, with an adjusted incidence rate of 204.7 cases per 100 thousand men and 175.6 cases per 100 thousand women. Regarding mortality, there were 9.6 million deaths from neoplasms, and cancer was considered the fourth leading cause of premature death (before 70 years of age) in different regions of the world 2 . In Brazil, in 2018, 243,588 deaths from cancer were recorded, with a standardized mortality rate of 109.7 deaths per 100 thousand men and 77.9 deaths per 100 thousand women. Estimates of cancer incidence for the triennium 2020-2022 point to the emergence of more than 600 thousand new cases 3 . Therefore, it is one of the most complex problems that health systems face, considering its epidemiological, social, and economic scope 4 .
The etiology of cancer is multifactorial. Approximately 30% of cancer types are related to environmental factors, which includes occupation. Thus, the increased interest in occupational exposures has grown in the scientific community 5 . One of the reasons for this increased interest is the fact that, over time, estimates show that the occurrence of occupational cancer is close, in magnitude, to deaths due to risk factors (RF) traditionally known and studied 6 .
In 2016, it was estimated that there were 349 thousand deaths and 7.2 million years of life lost due to disability (disability-adjusted life years -DALYs) due to these exposures, with relative variations between regions and ages 7 . Nonetheless, a major challenge in the prevention of occupational cancer is the lack of knowledge of where cancer exposures are taking place and how many workers are affected 8 .
Compared with traditional indices, the DALY indicator combines metrics that incorporate life expectancy and quality of life or disability, in addition to mortality itself. In this sense, changes in public policies can improve not only the measures of death, but also the burden of the disease, which is related to the time that a person can live with a certain disease and how much this can impact their quality of life 7 . For example, diabetes, musculoskeletal disorders, and mental disorders, overall, have low mortality and incidence. However, if these disorders occur early, they can provide a survival with many years of low quality of life to the affected individuals, as they are chronic conditions that directly interfere in the physical, mental, and social capacity of those affected. Thus, the aim of this study was to analyze the burden of occupational cancer in Brazil and federative units between 1990 and 2019.

Study design
This is a study on the burden of disease associated with occupational cancer, for the period from 1990 to 2019, in Brazil and federative units.

Data sources
Data were extracted from the Global Burden of Disease (GBD) study (https://www.healthdata.org/gbd/data-visualizations), available from 2022, concerning the 1990-2019 period, at the subnational level for Brazil. Deaths from cancer whose attributable RF was occupational carcinogens were considered for data extraction. The GBD study investigates RF groups. For each of those that were selected, the population attributable risk (PAR) is calculated, which assesses the burden of disease attributable to certain exposures. PAR allows identifying how much of the total risk for a given disease in the general population is due to a certain risk group. With this information, data on cancer deaths whose risk factor is attributed to carcinogenic occupational agents were obtained. Relative risk estimates are based on consistent research results, such as randomized controlled trials, cohort studies, and others, provided they are developed with appropriate methods. In turn, exposure levels and relative risk for each of the listed factors are estimated according to the available literature 9 .

Inclusion criteria
According to the GDB, the definition of occupational carcinogens includes the proportion of individuals inserted in groups identified as exposed (high and low exposure) to recognized carcinogens (asbestos, arsenic trioxide, acids, benzene, beryllium, cadmium, chromium, diesel, formaldehyde, nickel, polycyclic aromatic hydrocarbons, passive smoking, silica, trichloroethylene), having as reference the distribution of the population in 17 economic sectors 10 . In turn, occupational cancers listed by the GBD include mesothelioma, lung, tracheal and bronchial, laryngeal, ovarian, nasopharyngeal, kidney, esophageal, liver, pancreatic, leukemia, and bladder cancers 11 . It is worth mentioning that the included exposures, as well as cancer locations, are aligned with the classification of the International Agency for Research on Cancer (IARC) and were recently addressed for the Brazilian context 12 .

Spatial analysis
Spatial analysis was conducted with the first and last years of the series (1990 and 2019). Age-adjusted mortality rates were used to estimate the global Moran's index (Moran's I) to analyze the global spatial autocorrelation; and the local indicator of spatial association (LISA), to identify local https://doi.org/10.1590/1980-549720230001 clusters in the country with their statistical significance 13 . Moran's global and local indices aim to identify patterns of spatial distribution of the presented indicators. The following quadrants are presented to describe the univariate spatial correlation: high-high, low-low, high-low, and lowhigh. Moreover, in order to observe the spatial pattern for the two indicators concomitantly, a bivariate analysis was conducted and the spatial correlation of DALY and mortality rate was presented.

Time series
The occupational cancer mortality rate, adjusted for age, was analyzed based on its trend for Brazil and federative units, in the period between 1990 and 2019. It is assumed that the outcome has no seasonality. Thus, this component of the time series was not included in the model. The analysis followed three steps 14 . The stationarity of the series was evaluated by the Wald-Wolfowitz test. Then, to evaluate the trend direction, the Cox-Stuart test was used. Finally, the trends were analyzed by the polynomial model, whose dependent variable (Y) is represented by age-adjusted mortality rates, and the independent variable (X) is represented by the calendar years, modified from the midpoint of the series (year-centralized variable) to avoid multicollinearity. For choosing the best model, the analysis of residues was performed and the assumption of homoscedasticity and adherence to the normal distribution was evaluated. In addition, the scatter plot and the value of the coefficient of determination (R 2 ) were analyzed.
The software R 4.0.0 was used for the analysis of time series, and the GeoDa, version 1.14.0, was used for spatial analysis.

Ethical considerations
This study is exempt from consideration by the Research Ethics Committee for using secondary, public, and unidentified databases, pursuant to Resolutions No. 466/2012 and 510/2016.

RESULTS
In Brazil, from 1990 to 2019, the mortality rate from occupational cancer, as well as the DALY indicator, showed a decreasing trend. We can verify this difference between 1990 and 2019 (Table 1). It is worth highlighting that this discrepancy is observable when comparing the rates adjusted for age.
When verifying the crude rates, there is an opposite trend, corroborating the idea that cancer is an outcome dependent on the age structure. There are still notable differences between federative units, which comprise not only the comparison between the extreme years of the historical series, but also the trend. In Table 2 we present the results of trend analysis according to federative units. It should be noted that, although the mortality rate continues to decline, this is not the profile of the federative units. In fact, 15 of them tend to increase and are mostly located in the North and Northeast regions. Except for the states of Pará and Amapá, which did not show a significant trend, all other states presented models with adjustments of variable quality, but statistically significant (R 2 =0. 275  In Figure 1 we present the results of the analysis of the local spatial autocorrelation (Moran's LISA). Moran's I showed statistical significance (p-value<0.001) for global spatial autocorrelation of occupational cancer in Brazil, both for mortality and DALY. By employing the LISA method, we observed autocorrelation at the local level and spatial clusters. There are high-high clusters between states of the Southern region, whereas the low-low and low-high clusters are located in the Northeast region.
Finally, still in Figure 1, we analyzed the bivariate spatial correlation between mortality and DALY due to occupational cancer in Brazil in the two studied periods, by federative unit. The general results show spatial heterogeneity over the period. Regarding regional inequality, high mortality and high DALY correlation groups persist in the Southern regions and in the Northern Amazon region of the country. Conversely, federative units with correlation between low mortality and low DALY are grouped in the Northeast region.

DISCUSSION
Mortality rates and DALY due to occupational cancer tended to decrease in Brazil throughout the period. It is worth mentioning that, over the three decades, there has been an increase in metabolic risks. In addition, it should be noted that the study on global burden of diseases has a hierarchy of risk factors. Occupational risks are at the first level, along with other behavioral and metabolic factors. However, age-standardized overall mortality for some groups of chronic conditions has decreased, as documented in the GBD 2019 Study. This paradox has been attributed to the effect of access to health care, social determinants of health, cohort effects, and other behavioral risks, including changes in occupational health surveillance practices 15 .
Although decreasing in the country, we observed heterogeneity in the trends for the federative units. This evidence allows us highlighting the need for decentralized policies, under the sole command of the Federal Government, providing a level-equitable approach throughout the country 16 .
First, it is important to recognize that Brazil is a country of continental size, whose productive activities are widely https://doi.org/10.1590/1980-549720230001 spread throughout the geographic space. Activities of the primary, secondary, and tertiary sectors are developed in the country's territory. Particularly the industry, which is part of the secondary sector, has a higher concentration in the South and Southeast regions, especially the metropolitan region of São Paulo. Conversely, the primary sector of Brazil has prominence in the following regions: Midwest, with a more qualified standard in agribusiness; North, with plant and mineral extraction; and Northeast, with standard of monoculture agricultural production and low-tech mining 17 .
Regarding locations, recent estimates show that, even in more conservative scenarios, between 3.9 and 4.2% of all incident cases of cancer can be attributed to occupational exposure, with most of them being mesothelioma-type cancers, non-melanoma skin cancer, lung, female breast, and urinary bladder 18 . The study conducted by Purdue 19 showed that the potential impact of work on the genesis of cancer ranged from 2 to 8% (men, 3-14%; women, 1-2%).
Furthermore, in more specific studies, the fraction attributable to occupational cancer identified in the literature can reach over 30%, as is the case of lung cancer 5 . In fact, these locations are included in the study, along with others that GBD identifies as work-related 11 .
It is likely that the increase in mortality rates in more than half of the federative units reflects the improvement in the quality of the completion of Death Certificates, especially in the North and Northeast regions of the country 20 ; and that the increase in these rates among women and the decrease among men is due to the greater favoritism of the entry and increase of women in the labor market from the 1970s onwards, when there was an expansion of the economy, increasing urbanization and industrialization at an accelerated pace 21 . Thereafter, women started sharing with men the burden of occupational exposure to carcinogens.
Our results demonstrated a high-high spatial autocorrelation of occupational cancer both for mortality and DALY in the states of Santa Catarina and Rio Grande do Sul. A rea-  sonable explanation for these findings is that the Southern region is one of the most industrialized in the country, especially in the processing of primary products and in the production of parts and metallurgy 22 , and these industries may be involved in occupational exposure to carcinogens. Occupational cancers are largely preventable 23 . Nevertheless, it is worth emphasizing that the patterns of disease screening vary according to the characteristics of the occupation. It is known, for example, that the adherence of workers from small companies is lower than that of workers from large organizations as well as occupational activities that require lower level of education. It is also known that, when adjusted, these associations cease to exist, which suggests, therefore, that the type of occupation is a proxy for socioeconomic status and access to healthcare services 24 . These findings underscore the need for innovative public health strategies to improve cancer screening in vulnerable populations. As the decentralization of the productive process ends up exposing a population that, in itself, is already vulnerable, the social and political context of labor relations must be recognized, especially the fact that most developing countries lack political mechanisms to ensure the protection of workers 25 .
Moreover, it is remarkable that the detection of occupational cancer is partly difficult due to characteristics typical of natural history. Occupational exposures are often of low intensity and long duration, increasing the latency period to the disease 26 . It is quite common to discover it only after the work activity itself has been completed. Although levels of many exposures have been reduced in recent years, the long latency means that past high exposures will continue to result in substantial numbers in the near future 27 . Thus, despite controversies about the accuracy of quantitative estimates, there is a certain consensus that occupational cancer tends to be concentrated among relatively small groups of people, but who have a high risk in the development of the disease 28 .
It is noteworthy that all these estimates about the burden of occupational cancer are somewhat vulnerable to biases that may lead to underestimation of occupational burden, such as the exclusion of possible or probable carcinogens (following the IARC classification), the exclusion of cancer locations that are not emphatically described in the literature as being related to work, or a gap in the evidence of association with substances that have not yet been studied regarding carcinogenic potential 29 .
In fact, there are opportunities to revitalize comprehensive global cancer control policies, incorporating primary interventions against environmental and occupational carcinogens 30   that pose a risk for cancer, promoting an improvement in occupational health surveillance. This atlas provides the analysis of 18 cancer locations that are work-related. Spatial descriptions of mortality data are presented, as well as age-cohort-period analyses for all locations, based on standardized and corrected mortality rates, considering the fraction of these cancers that is attributed to occupation 31 . This product was developed in the creation of a list that includes carcinogenic agents, established or probable, present in the productive processes of some economic activities in Brazil, aiming at implementing monitoring actions and, ultimately, devising an action plan for the control of occupational cancer 32 .
More recently, the Brazilian Ministry of Health published a new edition of the Atlas, seeking to promote a national surveillance of occupational cancer 33 . However, the update was limited to presenting regionalized analyses of mortality by locations associated with work in its entirety. First, the adopted analysis strategy makes a comparison with the virtually limited attributable fraction, as it does not associate occupational exposure with the occurrence of cancer, but death due to this cancer. This event, nevertheless, is permeated by a number of other circumstances such as timely access to diagnosis and treatment. In addition, the Atlas analysis does not discriminate, in the historical series of mortality, what fraction of these deaths is attributable to work, either precisely, year by year, or in the trend. This measure is dependent on other characteristics besides the outcome such as the prevalence of occupational exposure to carcinogens. Only the historical series of the locations of cancers with some evidence of being work-related is observed, without necessarily having a causality. Hence, it would not be possible, by analyzing the Atlas, to evaluate the volume of cancer that could be avoided with the cessation of exposure. While the evidence provided by the Atlas is very limited to sectoral action on occupational cancer control -which was its original purpose -, the surveillance effort to make this field a priority should be recognized.
Our findings illustrate the repercussions of occupational exposure on cancer burden as one of the effects of work on health. Despite the fact that cancer prevention requires cessation of exposure to individual risk factors, such as smoking and consumption of processed foods, it is necessary to understand the contextual effect of exposure, especially those attributed to vulnerable populations such as occupational exposure 34 .
Therefore, in addition to the evident implication for the quality of life of workers, it is worth highlighting that productivity losses associated with cancer in the workplace are significant. At the same time that we used secondary data standardized by international methodology in this research, we sought to promote the visibility of this important public health issue, already stimulated by the Brazilian Ministry of Health in recent years.
All in all, the detection of occupational risks should be a priority in occupational health surveillance, reinforcing the need to develop strategies for preventing and controlling occupational cancer in Brazil from the perspective of public health and workers' health.