In this nationwide study, we analyzed the causes of death of more than 64k workers employed in the agriculture in comparison with a control group of 107K workers of the service sector. We found agricultural workers to be associated with mortality for neurological and malignant cancer-specific diseases. In particular, we identified in the group of neurological diseases, associations with mortality for spinal muscular atrophy and Parkinson’s disease, while in the lymphoid, hematopoietic and related tissue cancer group, we found associations with lymphoma, myeloma and leukemia diseases. In the malignant solid neoplasm group, we also identified positive associations in agricultural workers with mortality for cancer in stomach, colon-rectum, skin, connective and soft tissue, prostate, kidney and brain and central nervous system. An evident tendency towards an increased risk for cancer of lip was found for men only, likely caused by exposure to solar radiation.
Pesticides are considered the main agent of exposure for agricultural workers. They are specifically manufactured and used to eradicate or control undesired organisms. Insecticides, herbicides, fungicides, nematocides, rodenticides and other preparations are the pesticides for which there is a higher concern for occupational health [28]. Best-known toxic mechanisms of pesticides are: interference with axonal nerve conduction; synaptic transmission; mitochondrial respiration; steroid biosynthesis; blood coagulation [28]. The main target organs are the skin, the eyes, the mucosae of the respiratory tract, the digestive apparatus, and the nervous system. Other chronic effects involve liver and kidney toxicity, as well as chronic effects on the skin, on immune, respiratory, and endocrine systems, on blood, and on peripheral and central nervous system [28].
The results of this study are mostly in line with the ILO’s list of occupational diseases caused by pesticides exposure, and with the past and recent literature in this field [5–7, 15, 17, 29, 31, 35–41, 51, 52].
As for the diseases of the nervous system, we found farmers, and male in particular, to be at risk for Parkinson’s diseases. The evidence of the association between working in agriculture and Parkinson’s diseases is large but also controversial. In a review by Nandipati and Litvan [37], the authors addressed that among the different pesticides, rotenone, paraquat, and organochlorines have been well-documented in human epidemiological studies to be associated with Parkinson’s diseases, while organophosphates, pyrethroids, and polychlorinated biphenyls require further study. In a more recent systematic review, Sturm et al. [35] examined 22 studies about the relationship between agricultural work and Parkinson’s disease or Parkinsonism but they did not all agree on the presence of an association. Positive associations were found in studies carried out in France [41], Norway [53], US [54] and Spain [36]. In Italy one study concerning farming and Parkinson’s disease found no association [47]. Positive association was instead found in a case-control study in north-east of Italy [55]. An important genetic influence seems to be a factor [56–58].
Our study found an association between working in agriculture and mortality for the group of spinal muscular atrophy diseases (SMA) without gender difference. A few studies found an association between pesticide exposure and the development of ALS, a disease part of SMA group [59–61]. In Italy there is poor information about this disease in connection with working in agriculture. Filippini et al. [62] in a population-based case-control study in four Italian provinces, found no statistically significant positive association with ALS risk in agricultural workers. The results of our study, are in line with the scientific literature and add further evidence to this field of study. However, when stratified by sex this association loss of statistical significance.
We did not find associations with mortality for Alzheimer’s disease, multiple sclerosis, and epilepsy, although these health effects were positively associated with exposure to pesticides in Spain [36, 38] and in a review by Gangemi et al. [59]. At the same time, we did not find any association with mental, behavioral and neurodevelopmental disorders as a whole, and in vascular and unspecified dementia. One possible reason was the use of mortality data instead of hospitalization data of these diseases, which could have masked the associations.
As for cancer effects caused by long-term exposure to agents during working in agriculture, some studies have suggested an association between pesticide exposure and Hodgkin's disease, leukaemia, non-Hodgkin lymphoma, multiple myeloma (5,6, 10–14) as well as tumors of the lip, stomach, prostate, skin, brain, and connective tissues [5, 6, 31, 35]. The International Agency for Research on Cancer (IARC), evaluating the cancer risk associated with non-arsenical insecticide spraying, has concluded that this activity is probably carcinogenic to humans (Group 2A) and recently some active ingredients have been evaluated as certain or probable carcinogens for humans [27, 63], likely causing an increased risk for skin, lymphatic, and lung cancer [27, 28, 63]. However, no pesticide has sufficient evidence in humans to be classified as a brain or central nervous system carcinogen by IARC [27, 64].
The results of our study confirm the above evidence of cancer effects related to agricultural workers, providing evidence of risk for specific cancer types.
As for brain cancer, our results are in line with a study about mortality in male farmers licensed to use pesticides [24], with an Italian case-control study on farmer work and the cancer morbidity [65], and with some recent reviews [31, 35]. Our study provides additional information about its prevalence in men workers, and in farmers who worked for more than 30 years.
In Italy both cohort and case-control studies were conducted to investigate cancer effects among agricultural workers.
Faustini et al. [66] carried out a cohort-based mortality study of farm workers, in which no statistically significant excesses were observed in cancer mortality. A cohort of licensed pesticide users was investigated by Figà-Talamanca et al. [25] resulting in a standardized mortality ratio (SMR) of 72 for all cancers. An increase of risk for melanomas and eye tumors were observed in a cohort of farmers and for lymphoma and tumors of the connective tissue in the sub cohort of subjects living in villages with mainly arable land [67].
By comparing cancer risk mortality among two cohorts of Danish and Italian farmers, Ronco et al. [68] found a consistent risk reduction for cancer of the lung, bladder, small intestine, colon, rectum, and prostate, but an excess risk for leukemia in female only, and a slight excess risk for non-Hodgkin's lymphoma.
Among the case-control studies, an early mortality study about farmers in central Italy found no association with NHL, Hodgkin's lymphoma, multiple myeloma, and leukemia but when the different crops are considered some increasing risks are observed [69]. The findings of other Italian studies were more in agreement with our results about lymphoid, hematopoietic and related tissue cancer group. Among them a study carried out in a north area of Italy found increasing risk for NHL and chronic lymphocytic leukemia [19]. A multicenter case-control study suggested an increased risk for NHL and leukemia, and some chemical classes of pesticides, although few are statistically significant and some are based on few exposed cases [21]. As in our study, no statistically significant increased risk of NHL was found in a population-based case-control study carried out in Italy and involving many agricultural and mixed areas, in which exposure in agriculture was assessed by the use of a specific questionnaire given to subjects who were exposed to phenoxy herbicides not using protective equipment [23]. A hospital-based case-control study was conducted in five Italian rural areas by Settimi et al. [20], which investigated the association between cancer and farming among male agricultural workers. They found increased risks of cancer associated with agricultural workers for stomach, rectum, larynx and prostate, in agreement with the findings of this study out of the larynx one.
The above Italian studies show a large spectrum of specific positive associations results, often with inconsistencies and disagreement likely due to specific peculiarities of each study. Among them we could mention: type of study (eg. cohort and case-control), peculiarities in the sample of farmers (eg. population, pesticide licensed users), studied area (cohort limited to specific Provinces, multicenter) and their representativeness at national level, different job activities (eg. cultivation, harvesting of crops), methods of farmer identification (questionnaire, registries, census), exposure assessment and heterogeneities of cultivations at regional level. It's worth noting that the healthy worker effect or, the healthier lifestyle of farm families, can be easily observed for agricultural workers and can lead to an attenuation of the risk of death.
Although this study provides new and additional associations between health effects and agricultural workers in Italy, it contains few limitations that must be considered. The first is that occupational administrative data cannot adequately characterize occupational exposure profiles. INPS data report the industrial sector but do not contain information on workers’ roles in a specific sector or exposure variables such as exposure intensity. In addition, the complexity of exposure pattern, the difficulty in reconstructing past exposure, considering the change over time in specific chemical use, and the variation in work practices, should be taken into account. Consequently, occupation data of the present study only concern about the working-time spent in specific sectors, which might be involved with exposure to agents possibly causing diseases. Therefore, we used a rather broad classification of agricultural workers, not considering important factors for the exposure assessment like: specific job-related activities; heterogeneity of agricultural work itself (eg. crop type, animal type); type of pesticide applied; complex synergies among risks; and actual dose of exposure. We also not considered individual factors like genetic, lifestyles, socioeconomic status and co-morbidities, as well as external factors like environmental differences. We could not account for latency of the specific outcomes related to the exposure agents due to lack of information.
Nevertheless, this study provides a national picture of neurological and cancer risks that could be experienced by workers in agriculture in Italy, overcoming the limited information and representativeness provided by former case-control and cohort studies. Performing a record linkage of administrative data, and in particular between INPS files and mortality data, allowed us to fill the gap of the misclassification of the occupational information derived by questionnaire or census. To enhance the etiological relationship between the sector of employment and health outcome, we adopt a selection criterion based on a prevalent sector of employment, and a sector specific duration of it. Moreover, it was possible to establish an epidemiological study with high statistical power and to study the excess mortality risks among men and women separately.