A qualitative characterization of meso-activity factors to estimate soil exposure for agricultural workers

Meso-activity framework

Our approach builds on an occupational hygiene methodology, whereby specific job titles are used as surrogates to estimate exposures among workers [23]. Epidemiologists have also employed job titles as a surrogate exposure approach in retrospective studies of occupational hazards (e.g., asbestos) [24]. Exposure assignment based on job titles, however, may be imprecise when workers with the same job title complete a variety of tasks within a single occupational setting [25]. This approach has previously been used in the agricultural context to assess pesticide exposures [26] via dermal contact and inhalation [27], but not soil exposures. In pesticide risk assessment at the US EPA, the Pesticide Handler Exposure Database (PHED) provides estimates of exposures for specific scenarios, defined by a combination of job function, chemical formulation, and level of personal protective equipment [28]. The PHED is a valuable tool which incorporates task as a key variable, while also integrating both quantitative and qualitative behavioral considerations on exposure. Agricultural work is often not limited to pesticide application, however, and given the diversity of tasks farm workers may engage in [29], we propose that a meso-activity-centered framework that focuses on better characterization of each task and pathway is a more appropriate tool for soil exposure estimation for this population.

Participant recruitment and in-depth interviews

We used purposive sampling to identify and recruit growers via email and direct networking. Growers were eligible if they were currently a farm owner/manager, farm employee, or community gardener in Maryland, ≥ 18 years of age and had completed farm activities directly related to food production (e.g., planting, harvesting, weeding, mulching) within the past 12 months, and expected to be engaged in some agricultural tasks in the upcoming 12 months. We use the term “grower” intentionally to maintain the focus on the actions conducted by agricultural workers in this study; Using the terms “farmers” or “gardeners” evoke connections to the place (e.g., farm or garden) where the person works and may evoke misperceptions about the work environment or activities. In-depth interviews (IDI) were conducted by SL with 16 fruit and/or vegetable growers in Maryland at their farms between January and February 2020 using methods described previously [8]. All participants provided informed consent prior to the interview. All study tools and protocols were reviewed and approved by the Johns Hopkins Institutional Review Board (IRB00009866).

We used a semi-structured guide designed to gather information about farm tasks and soil contact. It began with questions asking growers to describe a typical workday and the farm operation (i.e., the distribution of the labor onsite), and included questions that asked growers to describe in greater detail the conduct of specific tasks (e.g., planting, irrigation, weeding, harvesting) mentioned during the interviews. It also included questions about soil contact (including incidental ingestion) and potential methods of increasing or decreasing soil contact (e.g., wearing personal protective equipment, typical work attire, and hand hygiene facilities onsite). Finally, the guide contained questions to solicit information about health and safety concerns experienced by growers while working onsite.

All interviews were audio-recorded and ranged from 21 to 92 min (median = 55 min; mean = 50 min). Recordings were transcribed verbatim using the NVivo transcription service and verified by one author who listened to the recordings while reading the automatically generated transcripts to verify transcription accuracy and correct typos as needed. We used the NVivo software program to facilitate the organize, coding and analysis of the qualitative data.

Qualitative data analysis

We used a framework approach for analysis comprised of the following steps: 1) transcription; 2) familiarization with the data; 3) coding; 4) developing and working with an analytical framework; 5) applying the analytical framework [30]. We coded each transcript using a combination of inductive and deductive coding methods [31]. In the first round of coding, we developed a set of deductive codes designed to capture key concepts targeted directly in the IDI guide (e.g., farm description, farmer background, task). After coding each transcript with the set of deductive codes, we re-read all transcripts for emergent themes related to specific farming tasks mentioned. For example, we asked growers to describe a typical workday at their farm/garden. While the overall sentiment was that no day is “typical” and every day is different, every grower mentioned at least two different work activities or tasks conducted on a typical day, as well as a host of other factors that may impact which tasks were done.

We collected descriptions of what growers do at each level of activity (macro, meso and micro) while engaging in farming (i.e., the given macro-activity) and characterized the human and environmental factors that influence the task and corresponding extent of soil exposure. Figure 1 defines each of these levels and provides examples. For example, a grower may choose to wear long pants or gloves to weed in the winter when it is cold outdoors, but the same grower may choose to wear shorts and no gloves in the summer when it is warm outdoors, resulting in greater direct soil contact in the summer. Furthermore, this pattern in attire may not be consistent across all tasks (i.e., growers may wear long pants for bed preparation tasks in the summer, even when it is warm). For each of the six emergent themes (i.e., meso-activities, or tasks) (Fig. 1), we aggregated all data related to a single task (e.g., planting), and then inductively coded for factors that further describe the nature of the task (e.g., crop, technique, tool use), and the nature of soil interaction (e.g., time of day, attire worn, handwashing practices). Growers also described a variety of micro-activities (i.e., hand or object to face or mouth behaviors) that may occur within meso-activities (Fig. 1).

We mapped the identified factors to existing exposure science concepts (e.g., soil ingestion, time in contact with soil) [32]. For example, factors such as age, sex and attire, specific to each grower were mapped to the term “receptor.” We collapsed these into broad categories of influence (Fig. 2); for example, age and sex were grouped into “biological” factors pertaining to the receptor. Factors specific to the farm such as its location or size were mapped to the term “environment.” Some factors were collapsed into larger categories to inclusively represent the phenomena described by growers. We organized each of these ten factors into a framework comprised of four classes of factors that characterize the extent of soil exposure in the agricultural context – environmental, activity, timing, and receptor (EAT-R) factors for each task growers may engage in (Fig. 2).

Fig. 2: Environmental, Activity, Timing – Receptor (EAT-R) Framework describing factors impacting soil exposure in the agricultural context.figure 2

This figure summarizes the environmental, activity, timing and receptor factors that may impact soil exposure in the agricultural context. The bold text indicates the four broad categories of factors identified via interviews with growers. The plain text identifies the ten sub-factors that impact soil exposures. The arrows indicate the direction of potential influences of factors on each other.

Demonstration of framework

We define each of the qualitative factors from the EAT-R framework and identify the quantitative inputs pertaining to each (Table 1). We also provide a hypothetical example of how the qualitative descriptions (informed by our IDIs) could be translated and incorporated quantitatively into traditional ingestion and dermal exposure models [33]. We also describe potential interactions between each of the factors.

Table 1 Description of qualitative factors in Environment, (meso-)Activity, Timing and Receptor (EAT-R) framework and summary of hypothetical quantitative impacts and interactions.

We developed the following task-based models to demonstrate the improved precision in exposure estimation afforded by consideration of these factors in our framework (Tables S110). We then compare soil exposure estimates for incidental ingestion (Table 2) and dermal exposure (Table 3) pathways derived using the models informed by our framework to those generated using the traditional models for dose estimation.

Table 2 Summary of sensitivity analyses of ingestion exposure incorporating Environment, (meso-Activity) and Timing factors.Table 3 Summary of sensitivity analyses of dermal contact exposure incorporating Environment, (meso-Activity) and Timing factors.

We used an iterative, stepwise approach to integrate EAT-R factors into traditional inhalation and dermal exposure models [33]. All models were designed to yield average daily doses incurred over an averaging time of one year to facilitate comparison across models. Average daily dose via ingestion was calculated using the equation:

$$Average\;daily\;dose\left( \right) = \; (Concentration \ast Intake\;rate \ast Exposure\;factor) \\ /\left( \right)$$

where the exposure factor describes the frequency and duration of exposure for the scenario of interest divided by the averaging time. In this context, exposure factors are a term greater than or equal to 0 (no time in contact with the hazard) and less than or equal to 1 (always in contact with the hazard) that describe the fraction of time a receptor is in contact with the hazard. The average daily dose via dermal contact was calculated using the equation:

$$Average\,daily\,dose_( ) = \; ( } \ast Surface\,area )\,\ast Exposure\,frequency \\ \ast Exposure\,duration \ast Event\,Frequency \\ \div ( ),$$

Where \(\beginAbsorbed\,dose_ = Concentration \ast conversion\,factor \ast Soil\,to\,skin\,adherence\,factor\\ \ast Dermal\,absorption\,fraction\end\)

Models 1 and 2 (Tables 2, 3) (Tables S1, 2) demonstrate the use of default assumptions regarding soil exposure via ingestion and dermal contact recommended in the US EPA Exposure Factors Handbook [7]. The EPA Exposure Factors Handbook contains recommended soil ingestion rates, estimated time in contact with soil, bodyweights, and anthropometric data pertaining to dermal exposure from the scientific literature and national surveys. These recommendations are frequently used to inform exposure assessments for contaminated lands and derive public and occupational health guidance values for contaminants in soil. Though the Exposure Factors Handbook does not have recommendations specific to agricultural scenarios, we chose inputs (Tables S110) for these models to most closely align with an agricultural exposure scenario. Models 3 and 4 (Tables S3, 4) illustrate the integration of meso-activity and yield average daily dose estimates (for ingestion and dermal exposure) for each task. In models 5 and 6 (Tables S5, 6) we incorporate timing factors to demonstrate the seasonal nature of agricultural work and account for the differential conduct of specific tasks across seasons. Models 7 and 8 (Tables S7, 8) show how environmental factors (e.g., differences in soil moisture attributable to different weather conditions) may impact soil exposure. Specifically, in Model 7 we differentially vary the intake rate across seasons for the ingestion model. In Model 8 we vary the skin to soil adherence factor across seasons for the dermal model. In models 9 and 10 (Tables S9, 10) we illustrate the impact of person-specific receptor factors (e.g., sex-specific body weights and differences in seasonal preferences for attire) by modeling exposure for two hypothetical growers of different ages and sex and different behavioral preferences.

留言 (0)

沒有登入
gif