• pesticides
• epidemiology

In this issue of Occupational and Environmental Medicine, Ohlander and colleagues review methods used to assess pesticide exposure in occupational epidemiology studies over the last 25 years .1 Their findings are striking for two reasons. The first is the very large number of studies published during this time period. Pesticides have been linked to almost every category of chronic disease, including cancer, respiratory, neurological and autoimmune diseases, as well as reproductive outcomes. The authors identified 1271 articles, averaging over 50 publications per year. Second, as the authors note, the methods used to assign pesticide exposure are relatively crude in many instances, limiting the aetiological inferences that can be made from these data.

Pesticide exposure is ubiquitous not only in occupational settings but also in the environment. Worldwide, in 2012, the last year for which data are available, almost six billion pounds of pesticide active ingredients were applied.2 These chemicals are important in agriculture and food production, public health (including vector control) and for aesthetic reasons. The use of pesticides has increased dramatically, with an estimated 11% growth per year between 1950 and 2000.3 This growth continues, particularly in low-income and middle-income countries. In the USA, it is estimated that >90% of the population has detectable concentrations of pesticide biomarkers in their urine or blood.4 It is therefore critical that we understand how they may impact human health. However, the word ‘pesticide’ is an umbrella term that describes hundreds of different chemical compounds, including herbicides, insecticides, fungicides, fumigants, rodenticides and nematocides, used alone and in combination. Without additional information, the term pesticide is not adequate for studying health effects.

Epidemiological studies dating back to the 1970s based on job titles, such as farmer, or tasks, such as pesticide applications, have suggested links between pesticides and a variety of health outcomes.5 These historical associations were intriguing and further detailed studies have led to important aetiological insights. For example, it has long been observed that both prostate cancer and lymphohaematopoietic malignancies are elevated in many farming populations.5 In follow-up studies, specific pesticides have been linked to these cancers, including some organophosphate insecticides (prostate cancer), organochlorine insecticides (non-Hodgkin's lymphoma) and the insecticide permethrin (multiple myeloma).6–8 Similarly, non-malignant respiratory diseases have been observed to be elevated among farmers for centuries,9 and more recent studies have found that specific pesticides, including some organophosphate insecticides, have been linked to respiratory outcomes, including wheeze, chronic bronchitis and chronic obstructive pulmonary disease.10 Going forward, studies that further characterise risks and provide mechanistic insights into these and other associations with specific pesticide exposures are critical.

Contemporary studies based on crude assignments of job title or tasks may also suggest areas in need of additional study, particularly in geographical areas or specific populations that have not yet been studied, such as low-income and middle-income countries. However, of the 1271 articles included in this review, Ohlander and colleagues note that 32.1% report exposure as ‘pesticides in general’ and another 15.9% assigned pesticide exposure based on job title only. Clearly, given the large number of pesticides used, the misclassification of exposures based on these assignments is enormous and would limit the ability to detect associations. Such broad classifications provide little information to identify specific aetiological agents and cannot provide insights into biological mechanisms or help understand disease processes. It is important that future epidemiological studies of these chemicals become more sophisticated.

Although occupational exposure assessment is never easy, the assessment of pesticide exposures presents some unique challenges, particularly for the investigations of chronic diseases. First, there is temporal variability, both within a year and across time. In the agricultural setting, where most epidemiological studies of pesticides have been conducted, pesticides are applied at specific times of the year, often for a very short period of time. Depending on the specific circumstances, the workers applying the pesticides may not be informed what pesticides they are applying. In other instances, farm workers engage in re-entry tasks that have a high potential for pesticide exposure, but the workers may have no idea what chemicals have been applied to these crops. In addition, older chemicals are replaced with new ones, meaning that the assessment of chemicals in current use may be of little utility in estimating past exposures. Work practices, including the amount of pesticides being applied, application methods and the use of personal protective equipment vary widely around the world and by agricultural commodity, and greatly influence the actual exposure a worker experiences. To capture exposure–response relationships, it is critical to study exposures in various settings. As Ohlander and colleagues note, the vast majority of studies in the past 25 years have been conducted in a small number of countries .1

Similarly, there are some pesticides that have been studied much more extensively than others. Some of the earliest manufactured pesticides are the organochlorine insecticides, which are mostly now banned due to their persistence in the body and the environment. However, this persistence makes it possible to conduct exposure assessment at an individual level based on serum biomarkers with decades-long half-lives. For example, the availability of easily measured serum metabolites of the organochlorine insecticide dichlorodiphenyltrichloroethane (DDT) represents a cumulative measure of lifetime exposure, which has contributed to it being studied extensively. There were over 100 studies of DDT and cancer when the International Agency for Research on Cancer (IARC) reviewed the carcinogenicity of this chemical in 2015.8 Among the handful of other pesticides recently reviewed by IARC, many of the evaluations relied on data from just a few epidemiological studies with sufficient detail on specific chemicals.8 11 These evaluations represent only a small percentage of the hundreds of pesticides in use. Most pesticides used today have short biological half-lives, measured in days and hours, precluding the use of exposure biomarkers to reflect the long-term exposures typically associated with chronic diseases, and which contribute to the paucity of data.

The use of pesticides provides important societal benefits, including the reduction of vectorborne diseases, and increased agricultural productivity, and it is clear that they will continue to be widely used in the foreseeable future. However, given the ubiquity of these chemicals, the potential for both occupational and environmental exposures, and the sheer variety of chemicals used, more research is needed to understand how pesticides may impact human health, both individually and as part of mixtures. There is a critical need for high-quality studies, including the identification of specific chemicals, alone and in combination. Both will require the development and implementation of novel exposure assessment methods or combinations of methods that reflect the situation of the population under study and the validation of these methods. Only then can we fully understand the potential impact of these chemicals on health.