This study aimed at estimating mortality risk excesses in workers of specific sectors based on integrating current information systems such as mortality data and detailed work histories. We provided detailed cancer-specific HRs by sector of employment and gender. Our findings indicate that various sectors exhibit elevated cancer mortality rates. While some sectors like construction, food and tobacco product manufacturing, machinery for paper and printing, and wood and leather products are commonly recognized as hazardous, others such as accommodation and food service activities, hairdressing, and cleaning are not typically considered risky. Moreover, among women, we observed increased mortality risks in cleaning activities, the chemical industry, transportation, health services, and agriculture. We focused the attention on neoplasms since cancer is still a leading cause of death worldwide among workers and the quota of occupational cancer, at around 5% of all cancer sites, widely underestimated.
Identification of risk factors is difficult since the occurrence of cancer diagnosis depends on a multiplicity of factors and, there is little distinction between professional and environmental factors. The delay between exposure and clinical diagnosis, particularly for solid tumors, may last for decades, which makes it difficult to identify occupational or other factors that could contribute to the onset of the pathology.
This study, characterized by high statistical power due to the large amount of collected data, enabled the investigation of the excess mortality risks in 25 industrial sectors and 20 cancer causes in men and women.
In addition to the known excess risks reported in the literature, the study has identified sectors that are not usually thought of as hazardous, as they are likely not associated with exposure to carcinogenic agents but involve exposures to potential carcinogens. Examples of such sectors include accommodation and food service activities, hairdressing and the cleaning sector. This excess may be explained by the characteristic of the study population in the municipality of Rome, where public administration and services are primarily present as economic activities rather than typically industrial sectors.
Cleaning workers are exposed to chemical substances that are found in dirt and dust while removing from surfaces, floors, furniture, etc. (25). The main active components of most cleaning agents responsible for health damage are surfactants, acids or bases, disinfectants, solvents, or some complexing agents (substances capable of forming a complex compound with another material in solution) (26). Also, formaldehyde is used in some cleaning products as a disinfectant or preserving agent and cleaners may also be exposed to volatile organic compounds (VOCs) emanating from cleaning products (27). Cleaners, in the past, were found to be affected mainly by: musculoskeletal diseases; skin diseases, dermatitis; respiratory, circulatory and cardiovascular disorders (28). In our study, cleaning workers have reported an increased mortality risk of stomach and lung cancer in men and of liver and lympho-haematopoietic cancer in women. Such results were partially confirmed by scientific literature that evidenced significant standard mortality ratios in lung cancer (29) and significantly elevated risk of chronic lymphocytic leukaemia (30), leukaemia (31), non-Hodgkin's lymphoma (32–33) in cleaners. Excess risk for liver cancer in private household workers, non-domestic cleaning and food and beverage service workers (34) was found in Brooklyn residents. We found, also, an increase of mortality risk with the duration of employment on all results, except for lymphatic and haematopoietic tissue in women where the risk decreased for length exceeding 10 years also resulting in a loss of statistical significance.
For workers in the restaurant and food preparation and cooking sector, cleaning and disinfecting chemicals substances together with cooking fumes and vapours are possible hazards (35). A critical review showed that the consumption of repeatedly heated cooking oils and the inhalation of their fumes in association with exposure to polycyclic aromatic hydrocarbons (PAH) could cause a high incidence of various cancers, among which lung cancer (36–37). The association of malignant neoplasm of digestive organs in restaurant workers is less documented in the literature; only a Spanish case - control study, carried out in 2012, found a statistically significant increased risk of stomach cancer in cooks (38).
Another occupational sector that is not usually thought of as hazardous is the category of hairdresser. The International Agency for Research on Cancer (IARC) has classified such occupation as posing a probable carcinogenic risk, based primarily on the exposure to hair colourants and to hair dyes, hairsprays, formaldehyde, and some solvents (39–40).
Our study results suggested a higher risk of liver cancer in men, which, however, is not supported by the literature. Our findings regarding a higher risk of liver cancer were only mirrored in scientific literature associated with exposure to hair sprays containing vinyl chloride (VC) as a propellant (41) and to azo pigments with limited evidence from studies on rats (42).
Such findings should prompt us to question the dangers of permanent hair colours, as they are not applied only by professionals but also by consumers even if the frequency and duration of exposure, however, may be quite different for consumers and professionals.
The significant mortality risk found in construction workers for lung and pleura cancer is a well-known risk (43), which has been reduced in the last 20 years (44) thanks to strong regulations to reduce exposure to asbestos and silica.
No specific evidence supports the risk reported in our study for liver and stomach cancer in construction as well as for central nervous system and brain cancer. In scientific literature, a Canadian study suggested a possible role of occupational exposure in the aetiology of brain cancer in construction trades associated with exposure to asbestos (45).
The risk of the central nervous system and brain cancer detected in metallurgy is not supported by literature results, while the significant excess risks found for prostate cancer in the leather industries were in line with evidence shown in the scientific literature (46). The risk was not positively related to the length of employment.
Among women, the increased mortality risk for lymphatic and haematopoietic tissue in transport and healthcare and for leukemia in agriculture is noteworthy. All these risks show a consistent association with the length of employment. There is scientific evidence for the risk of leukemia in agriculture (47). In transport, similar results were confirmed in truck and conveyor operators among women, as far as Non-Hodgkin's lymphomas and myeloma are concerned (48–49). In health personnel, the excess risk may be explained by exposure to formaldehyde or ethylene oxide.
Understanding aetiology requires a comprehensive assessment of micro-exposures and potential interactions among multiple hazardous agents, often leading to additive or synergistic effects. The most effective way to assess occupational exposure would be to have an expert industrial hygienist who analyzes individual job descriptions. Although this method produces accurate results, it is time-consuming, costly, and challenging to to be implemented in large-scale studies. Thus, alternative methods have been introduced that consider occupations or job titles as a proxy for substance exposure when there are no estimates for agent exposure. Among them, we may consider self-reported interviews, job exposure matrix, and administrative files collecting work histories. Self-reported interviews, used to reconstruct work history and assess exposure, have been shown to be prone to misclassification (10, 11). Inaccuracies in self-reported work histories from interviews may cause variability in cumulative exposures. The job exposure matrix (JEM) is another valid method to characterize the worker’s exposure to carcinogen agents, but it provides accurate assessments of exposure for groups of workers, not for individual subjects. Moreover, JEMs may not be inclusive of all occupational carcinogens and may have inadequate standardization in classification systems among countries and over time, leading to variability and potential inaccuracies in exposure estimation (12, 13).
Collecting complete employment histories through administrative records is a valid method to gather occupational information retrospectively for epidemiological research (14). It may overcome the limitations of time and cost involved in hygienist evaluation, the imprecision of self-reported interviews and the misrepresentation of individual estimated exposure in JEM’s use. Various attempts were made in many countries to evaluate the possibility of using administrative data from social security records to estimate the occupational component in epidemiological research (15–17).
In Italy, numerous studies have used record linkage methods to estimate the risk of cancer diagnosis and the permanence in a particular economic sector, considering the employment histories acquired from current data sources as a measure of exposure (21, 50, 51). The use of such type of “electronic data linkage” has been demonstrated as a powerful tool in epidemiological studies such as case-control studies (50), mortality studies (51) and cohort studies (21).
Limits and strengths
The most significant limitation of this study, since it is based on an administrative cohort, is the absence of direct information on individual confounders, such as, especially, smoking habits. There are studies in the literature attempting to address this issue by using sophisticated methodologies, but they are not suitable for occupational epidemiological studies (52). To mitigate this bias, educational attainment was used as a surrogate for smoking habits, given the well-known association between smoking and low educational attainment (53): indeed, the results remain unchanged.
An additional constraint of the study, inherent in its design lies in the exposure classification, where subjects are assigned to the specific occupational sector having the maximum length of work. The industrial sector was the only information retrieved in the social security records and it is considered as a measure of the exposure, however, no individual exposure to harmful factors or specific type of work and profession could be included.
We were unable to account for the potential confounding factor known as the healthy worker effect, which results in a lower relative mortality rate. This phenomenon occurs because generally healthier individuals are more inclined to obtain and retain employment. Moreover, the impact of the healthy worker effect can vary depending on variables such as the type of disease, age, intensity and duration of the exposure, and gender (54). Nevertheless, the results of this study confirm excesses of mortality risk in sectors where employees are primarily exposed to risk factors described in the literature, such as wood and leather industry, construction, healthcare, and metallurgy. At the same time, new risks requiring further investigations have been highlighted, such as cleaning activities, accommodation, and food services, even though they are considered at low risk as they involve mainly administrative tasks.
Performing a record linkage of administrative data, and in particular between INPS files and health data derived from the population census conducted in 2011, allowed us to fill the gap of occupational information which refers only to a limited sample of population in the census. Moreover, by using this methodology, it was possible to establish an epidemiological study with high statistical power and to study the excess mortality risks among men and women separately.