Application and demonstration of meso-activity exposure factors to advance estimates of incidental soil ingestion among agricultural workers

Agricultural workers can encounter multiple physical, chemical, and biological hazards, yet soil as a contaminant reservoir remains understudied despite potential ingestion and dermal pathways. Existing occupational pesticide assessments emphasize inhalation and dermal exposure and largely ignore incidental soil ingestion. General-population soil ingestion defaults (e.g., 10–50 mg/day) and worker defaults (e.g., 100–330 mg/day) may not reflect the activities and intensities typical of agricultural work. Because agricultural tasks vary by type, duration, and season, a single daily ingestion rate likely mischaracterizes exposure. This study aims to quantify meso-activity-specific exposure factors (frequency, duration, soil-contact intensity, PPE and hygiene behaviors) for growers and apply these data to improve estimates of soil ingestion exposures using daily, hourly, and task-specific ingestion rate approaches.

Prior research and regulatory databases (PHED, ARTF) emphasize inhalation and dermal routes for pesticide exposures and do not quantify incidental soil ingestion among agricultural workers. Soil and dust are distinct media with differing behaviors and definitions that complicate exposure characterization. Methods to estimate soil ingestion include tracer element and biokinetic approaches and activity-pattern assessments; confidence in general-population soil ingestion rates is low, and data for highly exposed occupations are sparse. EPA recommends defaults (e.g., 100 mg/day for adult residential screening levels, 330 mg/day for construction workers) without validation for agriculture. Occupational epidemiology has used task and crop-based matrices for pesticides, but comparable matrices for soil contaminants are lacking. A prior qualitative framework (EAT-R) identified environmental, activity, timing, and receptor factors influencing soil ingestion, motivating the need for quantitative estimates to inform risk assessment.

Study design: A longitudinal, season-specific soil contact activity questionnaire was administered via telephone to 38 Maryland fruit and vegetable growers beginning spring 2020 and repeated in summer, fall, and winter (April 2020–March 2021). Eligibility included adults engaged in edible plant production within the past 12 months and expected to continue for the next 12 months. Data capture used REDCap. Participants received incremental gift cards; IRB approval was obtained. Questionnaire: Based on the EAT-R framework, growers reported frequency (days/week or days/month) and duration (hours/day) for overall on-site work and six meso-activities: bed preparation, planting seeds, transplanting, irrigation, weeding, and harvesting. They estimated the fraction of time hands contacted soil during each task; described tools, clothing, PPE, and ergonomic positions; and reported events/behaviors influencing ingestion (soil in mouth, soil on face, sampling produce, eating on-site). A spoon-image aid elicited estimated amounts ingested on days with soil-in-mouth events using EPA-referenced benchmarks (e.g., 10, 20, 50, 100, 200, 1000 mg/day; with <10 and >1000 mg options). Demographic and farm characteristics were collected. Data analysis: Nonparametric tests (Kruskal-Wallis) assessed seasonal and task differences in exposure factors. Shapiro-Wilk tested distributional assumptions. Pearson correlations examined associations of farm size with hours worked; Kendall’s tau assessed associations with employment status. Open-ended responses were thematically coded. Exposure assessment: ADDs to a hypothetical soil contaminant at constant concentration C = 400 mg/kg were estimated for each grower empirically over a 30-day period using three approaches and complemented by Monte Carlo simulations (n=5000) per season. Method 1 (daily ingestion): Used a high-contact daily soil+dust ingestion rate IR = 378 mg/day (95th percentile from modeled high-contact scenario) with grower-reported exposure frequency (days/week converted to days/month), exposure duration 1 month, averaging time 30.5 days, and body weight from EPA Exposure Factors Handbook by age/sex. Method 2 (hourly ingestion): Distinguished outdoor working hours and indoor non-working hours. Hourly ingestion rates: 45.25 mg/h for outdoor agricultural work (derived from 362 mg/day over 8 h) and 1.38 mg/h for indoor non-working time (22 mg/day over 16 h). Exposure factors were grower-reported hours/day and days/week converted to hours/month for outdoors; indoor hours were the complement of 730.8 h/month. ADDs for indoor and outdoor were summed, converted to daily basis. Method 3 (hourly task-specific): Adjusted the baseline outdoor hourly ingestion rate (45.25 mg/h) by task using grower-reported fraction of time with direct soil contact as scaling factors. Example adjustments: transplanting and weeding (high contact, mean 87% and 72%) doubled to ~90.5 mg/h; seeding and preparing beds (49% and 41%) retained baseline 45.25 mg/h; harvesting and watering (35% and 8%) halved to ~22.63 mg/h. Grower-reported hours/day and days/month per task yielded hours/month per task; unaccounted hours were treated as indoor at 1.38 mg/h. ADDs across tasks plus indoor were summed. Monte Carlo simulation: For n=5000 simulated growers per season, body weights were sampled from NHANES 2017–March 2020 adult distribution with survey weights. Method 1 sampled days/month at site from empirical season-specific responses. Method 2 sampled hours/month at site similarly. Method 3 sampled task hours/month by season from empirical responses and sampled task-specific ingestion rates from uniform distributions ±20% around empirical method-3 multipliers. Seasonal ADD distributions were generated for each method.

• Participation and demographics: 38 growers completed at least one season; 22 completed all four. Sex: 52% male. Median age 40 years (range 26–69). Most worked ≥35 h/week (71%). Farms spanned 0.4–835 ha; 21% USDA-certified Organic; 95% reported organic practices.
• Time-activity patterns: On-site hours/month were highest in summer (mean 204.6; median 184.9 h) and lowest in winter (mean 76.8; median 54.4 h). Meso-activity frequencies/durations varied seasonally: irrigation occurred on the greatest number of days/month (median 6 fall to 20 summer), while preparing beds had the greatest hours/day (median ≥3 h). Hours/month by activity: preparing beds dominated spring (median 21 h), fall (18 h), and winter (12 h); harvesting dominated summer (30 h). Total hours/month across six tasks ranged 1–360.
• Soil-contact intensity and behaviors: Transplanting and weeding had the highest self-reported hand contact with soil (means 87% and 72%). Ergonomics aligned with contact intensity: kneeling common for transplanting and weeding; standing for bed prep, seeding, watering; bending for harvesting. Glove use among those who wore gloves averaged >65% of task time, highest for transplanting (mean 76%) and weeding (mean 71%). Long pants and closed-toe shoes were universal in fall/winter; sandals reported by some in spring/summer. Over half reported handwashing after most tasks (except watering), especially after transplanting and harvesting.
• Non-dietary ingestion-related events: More growers reported soil on face in spring/fall; frequency per reporter was highest in spring/summer. Among those reporting soil ingestion in past 30 days, mean days with ingestion were 9.5 (summer) and 7.9 (spring). Across seasons, 16 growers estimated <10 mg/day, 11 estimated 10–20 mg/day, and 6 estimated ≥50 mg/day on days with soil ingestion events. Sampling produce while working was common, most frequent in summer (mean 15 events/month); produce washing before sampling was more frequent in winter (reported 100% of the time by those who sampled) than spring (mean 92.5%). Eating meals/snacks on-site was least reported but occurred for about 60% each season.
• Statistical tests: Significant seasonal differences (p < 0.05) in days/month at site, hours/month at site, and hours/month across all activities; significant differences across activities in hours/month. Time on-site positively correlated with farm size (approached significance, p=0.08). Employment status correlated with hours/month at site (full-time > part-time > others).
• Internal consistency: Reported on-site hours typically exceeded summed hours in the six queried tasks (71% of responses), indicating time spent on other activities not captured by the six meso-activities or concurrent tasks.
• Exposure modeling results: Empirical annual exposures across methods were of similar order of magnitude; method 1 generally yielded the highest cumulative annual ADDs and method 3 the lowest. Most exposure accrued in spring and summer across methods. In method 3, no consistent pattern in activity-specific contributions to annual exposure emerged. Simulated seasonal ADD distributions across methods were comparable in magnitude; method 1 distributions were left-skewed, method 2 less skewed with broader ranges. Within method 3, season-task ADDs showed greater variability and distinct distribution shapes (e.g., multimodal for transplanting). Overall, incorporating task-specific time-activity and contact intensity refined exposure estimates without producing large shifts in seasonal ADD magnitudes.

The study addresses a key gap in quantifying incidental soil ingestion among agricultural workers by generating season- and task-specific exposure factors and demonstrating their application in exposure models. Findings confirm substantial variability in time allocation and soil-contact intensity across meso-activities, seasons, and ergonomic contexts. Incorporating these dimensions enables more nuanced estimates than single daily ingestion defaults. Although the three ingestion-rate approaches produced similar magnitudes of seasonal ADDs, task-specific modeling provided insight into when and during which activities exposure may be concentrated and highlighted the role of indoor versus outdoor time. The results suggest that interventions may need to consider both highly frequent tasks (e.g., irrigation) and high-contact-intensity tasks (e.g., transplanting, weeding). The approach forms a foundation for integrating soil ingestion into broader agricultural chemical risk assessments, including pesticides, thereby complementing inhalation and dermal pathways typically emphasized in regulation.

This work develops and applies meso-activity exposure factors for agricultural workers, quantifying frequency, duration, and soil-contact intensities across six key tasks by season, and demonstrates how these data can be used to refine soil ingestion exposure estimates via daily, hourly, and task-specific ingestion-rate models. The approach supports more robust and context-specific soil exposure assessments for agricultural workers, a population often omitted from soil ingestion guidance. Future research should validate task-specific ingestion rates, expand task inventories beyond the six studied, account for concurrent activities and micro-activities (e.g., hand-to-mouth), incorporate behavioral modifiers (e.g., handwashing timing), and use direct measurements or novel tracers to reduce uncertainty and improve generalizability to diverse agricultural contexts, including more mechanized operations.

• Self-reported time-activity and ingestion amounts are subject to recall and perception biases and were not validated against objective measures or tracers.
• Only six meso-activities were queried; time spent on other tasks or concurrent activities likely led to under- or overestimation in task-specific models.
• Potential misclassification between soil and dust in participant responses and ambiguity in environmental media definitions.
• Indoor/outdoor allocation assumptions (e.g., non-task time being indoors) may underestimate outdoor exposures and soil contact.
• Single participant per farm and a Maryland-based sample limit generalizability; most participants were managers/owners, and mechanization levels vary across operations.
• Data were collected during the COVID-19 pandemic, potentially altering hygiene practices, staffing, and PPE availability.
• Task definitions and ergonomics vary within and across growers (e.g., irrigation includes diverse actions), affecting contact intensity estimates.
• Default ingestion rates and task-specific scaling factors are not empirically validated for agricultural workers; uncertainty remains in daily, hourly, and task-specific rates.