Background
Heat exposure is associated with substantial occupational mortality and morbidity, including from heat-related illness (HRI), traumatic injuries, and acute kidney injury [
1‐
5]. In 2015, exposure to heat caused 2830 occupational injuries and illnesses resulting in days away from work and 37 work-related deaths in the United States (US), 89% of which occurred during the summer months (June–September) [
6]. Agricultural workers have high rates of HRI and heat-related deaths. From 2000 to 2010, agricultural workers had more than 35 times the risk of heat-related death compared to other industry sectors, with a yearly average fatality rate of 3.1 per 1 million workers [
1]. In the agriculturally intensive State of Washington (WA), there were a total of 918 workers’ compensation HRI claims during 2006–2017, with the agriculture, forestry, fishing, and hunting sector having the second highest third quarter (July–September) rate (102.6 claims per 100,000 full-time employees [FTE]) and the highest annual HRI claims rate (13.0 per 100,000 FTE) [
7]. HRIs are likely more prevalent than data indicate [
7,
8], as less severe injuries and illnesses may be self-treated and not reported to supervisors, and agricultural workers may prioritize work over taking time off for treatment and recuperation [
9]. The risk of HRI is unlikely to diminish in the future, as the frequency and intensity of heat events is projected to increase [
10].
Field evaluations of the effectiveness of interventions to reduce farmworker HRI risk are needed to support prioritization of the most promising approaches. Though there is growing evidence that farmworker education that is participatory, culturally and linguistically appropriate, and tailored to agriculture is effective in improving heat knowledge and behavioral intentions [
11,
12], few studies have investigated the effectiveness of these interventions on objective measures of heat strain. Pilot evaluations of the effectiveness of different cooling strategies and hydration on core body temperature and kidney function among agricultural workers have been performed [
13,
14]. Formative work suggests that supervisor mobile applications that provide local weather conditions and recommendations for protecting workers from heat may be acceptable to agricultural supervisors [
15,
16]. A mobile application that provides users with information about predicted heat stress based on environmental conditions, activity level, clothing, and acclimatization has also been developed and evaluated [
17]. Interventions that include an emphasis on water, rest, and shade at work have shown promise, including in preventing adverse heat health effects among sugarcane workers in Central America [
18]. California, WA, and Oregon are the only three US states that have developed emergency or permanent occupational heat rules intended to prevent outdoor HRI [
19‐
22]. However, research in California suggests an increased risk of HRI even when farms follow California/Occupational Safety Administration heat regulations [
23], suggesting that the way in which rules and practices are implemented and the effectiveness of specific provisions needs further evaluation. Risk factors for adverse heat health effects exist at multiple levels (e.g., individual, co-worker, employer, community, and policy levels), yet few studies have developed interventions using a multi-level framework tailored to agricultural settings [
3].
Heat stress is defined within the American Conference of Governmental Industrial Hygienists (ACGIH) Threshold Limit Value (TLV)® as the net heat load to which a worker may be exposed from the combined contributions of metabolic heat (e.g., from physical work), environmental factors, and clothing [
24]. Heat strain refers to the overall physiological response to heat stress aimed at dissipating excess heat from the body, and the TLV aims to maintain the core body temperature within 1 °C of normal (37 °C) [
24]. HRIs include heat rash, heat exhaustion, heat syncope (fainting), and heat stroke, which is associated with an elevated core body temperature (> 40 °C, 104 °F) and can be fatal. Different HRIs manifest clinically with different groups of symptoms. Though occupational health guidelines and rules incorporate recognition and reporting of HRI symptoms [
19‐
22,
24‐
26], HRI symptoms may be non-specific (e.g., headache, fatigue), there is little consensus on how best to categorize HRI symptoms [
27] or how reporting of symptoms relates to physiological heat strain, and different factors may affect reporting of HRI symptoms. Physiological monitoring of heat strain does not rely on self-report and captures individual responses to heat load, which depend on several factors, including personal factors (e.g., age, sex, fitness level, acclimatization status, health conditions, medications, hydration level), environmental conditions, workload, and clothing [
26].
Agricultural workers are integral to the US food supply, and there are opportunities to improve agricultural worker safety and health. In this study, our primary objective was to evaluate the effectiveness of a multi-level HRI prevention approach that addresses individual, community, and employer level factors through worker education and a supervisor decision support mobile application among agricultural workers in WA. We hypothesize that this multi-level Heat Education and Awareness Tools (HEAT) intervention can improve HRI awareness and prevention practices and therefore reduce physiological heat strain among agricultural workers. Our secondary objective was to describe the relationship between objectively measured physiological heat strain and self-reported symptoms and to describe factors associated with HRI symptoms reporting.
Discussion
We conducted a parallel, comparison group intervention study of the HEAT intervention, consisting of culturally- and agriculture-tailored participatory farmworker heat education and a supervisor decision-support mobile application, in Washington State, US. We found larger decreases in physiological heat strain across a summer season in the intervention compared to comparison group for higher levels of work exertion, but this was not statistically significant. Prior studies suggest that participatory education that is culturally tailored is associated with improved farmworker heat knowledge and behavioral intentions [
11,
12] and that mobile heat safety and decision support applications are well-received by agricultural supervisors [
16]. However, knowledge and behavioral intentions alone may not lead to change. According to the Health Belief Model, behavior change is achieved through targeting perceived barriers, benefits, susceptibility, and threats [
53]. While tailored participatory education and supervisor decision support could influence these factors, findings from our study support the principle that reductions in heat strain and the risk of HRI require additional elements of heat stress management at the workplace level and systemic change to address barriers to reporting symptoms, pay structure, and access to healthcare.
The strongest predictor of physiological strain in this study was work exertion (effort level). Results from our fully adjusted model indicate that workers performing tasks requiring high effort had PSI
max levels almost two points higher, on average, than those performing low/medium-low effort tasks. In our study, high effort corresponds roughly to moderate-high metabolic rate activities (300–415 W) while low/medium-low effort corresponds roughly to light-moderate tasks (180–300 W), depending on the specifics of the task [
24]. This is consistent with other studies among California and Florida agricultural workers that have identified work rate and physical activity level to be associated with elevated core body temperatures [
49,
50]. Contributors to heat stress include metabolic heat (e.g., from physical work), environmental factors, and clothing [
24]. Participants in this study generally wore long-sleeved shirts underneath hooded sweatshirts or button-down shirts and long pants. PSI is a function of heart rate and core body temperature, and the PSI
max for participants performing low effort work was largely in the low range (PSI 3–4), as defined by Moran et al. [
44], compared to the medium PSI range (PSI 5–6) or higher for participants performing high effort work. It is possible that in WA during the study period, when the mean HI
max were in the 80s, metabolic heat was a key driver of heat stress and subsequent heat strain. This is consistent with 2006–2017 WA workers’ compensation HRI claims observations that indicate that the maximum daily temperatures on illness days were below the current WA heat rule temperature threshold of 89 °F (32 °C) for 45% of claims [
7]. Overall, given the strong effect of exertion on PSI, further emphasis is needed to ensure adequate rest breaks, job rotation, and/or work pace reduction in the heat, along with payment for breaks and other mechanisms to reduce financial barriers to cool down and rest.
There are several potential reasons for the observed lack of a statistically significant HEAT intervention effect. First, the sample size of this study may not have been adequate to detect effects in fully adjusted models, including interactions. Second, our estimates of the intervention effect may have been an underestimate. It is possible there was sharing of information between intervention and comparison workers, which may have led to more conservative estimates of the effectiveness of the intervention, as the analysis was intention-to-treat. For example, at one of the large companies (Large-1), crews sometimes lived and worked together. Finally, though HEAT education has been shown to result in significant improvement in worker knowledge compared to comparison crew workers [
12], worker education alone may not lead to actions to reduce heat strain and the risk of HRI. Unlike the existing US Occupational Safety & Health Administration (OSHA)/National Institute for Occupational Safety & Health (NIOSH) Heat Index-based mobile heat application [
54,
55], our HEAT App provides messages tailored to agriculture, local environmental data from nearby agricultural weather stations, and longer forecasting for work planning, in response to early advisor and expert working group feedback. However, in contrast to NIOSH and ACGIH heat safety guidelines, the OSHA/NIOSH and our HEAT applications are simpler tools that don’t explicitly include work pace, clothing, and acclimatization status into risk calculations and may not fully represent risk, particularly at high workloads [
26]. Education and decision support should be combined with other factors in heat stress management plans and policies, including behavioral thermoregulation and sufficient rest breaks, acclimatization procedures, adequate hydration, clothing, emergency response procedures, and consideration of mechanization and work pace.
We did not find a clear relationship between PSI and the number of symptoms reported. These findings are consistent with previous reports of heat stroke occurring in the absence of reported symptoms, and symptoms occurring in the absence of other signs of HRI [
26]. We were unable to distinguish whether the number of symptoms reported was the number of actual symptoms experienced or was influenced by factors affecting willingness to report. Factors associated with PSI and factors associated with the number of symptoms reported were not the same. Participants were more likely to report symptoms if they were older in age and worked in agriculture in the US for 10+ years, which may reflect actual increased symptoms, awareness and perception of symptoms, or comfort with reporting. This is different to findings from our 2013 survey study of 97 WA tree fruit harvest workers, which indicated a lower odds of reported HRI symptoms with increasing age [
40]. However, the present study sample included 37% H-2A workers, who, based on our observations, tended to be younger, male, and perform work requiring higher exertion levels than our 2013 study, which may have been reflected in higher PSI
max values. Our fully adjusted models of heat strain indicated that PSI
max scores decreased almost ½ point per decade of age, on average, in contrast with the increase in symptom reporting with older age. H-2A workers in our study were less likely to report HRI symptoms, consistent with prior studies [
56]. H-2A workers often do not have adequate access to healthcare, and barriers to reporting may include fear of reporting affecting current and future employment and well-being [
57]. Additional work is needed to ensure adequate healthcare for H-2A and other agricultural workers and to address concerns about retaliation for reporting.
Several other factors were associated with symptoms reporting. First, in contrast to a previous study of Florida agricultural workers, women in our study were not significantly more likely to report a higher number of HRI symptoms [
27]. In WA, between 2006 and 2017, there was a higher percentage of males with accepted HRI State Fund workers’ compensation claims, compared to all State Fund claims (82% versus 68%, respectively,
P < 0.001) [
7]. Though this may reflect employment predominance in agriculture by males, our study raises the question of whether differential reporting by men and women may also influence these results. Second, participants were more likely to report symptoms if they reported cooling opportunities outside of work or air conditioning at home. It is difficult to determine whether these participants seek cooling opportunities because they are more likely to experience symptoms or vice versa. There was no significant relationship between prior HRI training and symptoms reporting. Participants were more likely to report symptoms when exposed to hotter work conditions and having to walk more than three minutes to get to the toilet at work. A longer distance to the toilet has been previously reported to be associated with HRI symptoms [
40]. Farmworkers, particularly those paid by the amount harvested (piece rate), may be less likely to fully hydrate and urinate regularly if it requires a longer time away from work to walk to the toilet. Piece-rate pay has been previously reported to be associated with reported HRI symptoms and with acute kidney injury among agricultural workers [
40,
58]. Well-maintained toilets that are mobile, for example, toilets attached to trucks that move with workers, and consideration of payment schemes that do not incentivize skipping breaks, but still allow workers to maintain their well-being, may support optimal hydration and HRI risk reduction.
Strengths & limitations
Strengths of this study include: its controlled, comparison design; recruitment and monitoring of agricultural workers and supervisors, including H-2A workers, over a harvest season; participatory design of the research and intervention; and collection of objective physiological data on heat strain, in addition to subjective symptom reporting. In addition to the small sample size and potential for cross-over previously discussed, this study has several other important limitations. First, we used convenience sampling in the selection of companies and workers. It is possible that companies and workers who chose not to participate may have had a greater potential for improvement in heat strain with the intervention, rendering our results more conservative. Second, weekly symptoms reporting may have been subject to recall bias. Third, due to time constraints and work pressures, the small farm crews were not randomized, and the crew that arrived first was allocated to the intervention. However, we do not have reason to believe that there are systematic differences in crews by their arrival time. Fourth, some participants completed weekly symptoms surveys by mobile application and some by telephone. Those that completed the surveys by mobile application versus telephone may have been more or less likely to report symptoms, but we do not expect there would be any systematic differences by intervention versus comparison group. Fifth, not all companies’ workers participated at the same time during the study period. Though it would be ideal to have participation on the exact same dates by all companies and workers, we were able to account for time-varying variability in factors such as heat exposure, and also accounted for company, in our analyses. Fifth, we used HI rather than more complex environmental metrics that also account for solar radiation and wind speed and that correlate better with heat strain [
59]. However, we were able to assess physiologic heat strain directly, using estimated core body temperature, in our analyses. Though solar radiation has been reported to affect certain heat-related body responses [
60], we do not expect that there was differential variability in solar radiation in the intervention and comparison groups. Sixth, we did not use a published compendium of physical activities to categorize task [
61]. However, we did use expert review of crop and task combinations by study team members with training in occupational safety and health and who observed tasks in the field. Seventh, we used an approach to estimate core body temperature rather than assessing gastrointestinal temperature. However, this approach has been evaluated among WA agricultural workers and demonstrated to correlate well with gastrointestinal temperature [
43]. Finally, our study may not be generalizable to agricultural populations in other states, as it was conducted in a US Pacific Northwest State with an occupational heat rule [
20].
Conclusion
In this study of the HEAT intervention among Washington State, US farmworkers, we found larger decreases in physiological heat strain across a summer season in the intervention compared to comparison group for higher levels of work exertion, but this was not statistically significant. The strongest predictor of physiological strain in this study was work exertion (effort level). In addition to education and administrative controls, other factors that affect heat stress, including effort level, clothing, hydration, acclimatization, and emergency response plans, must be directly addressed in workplace heat management plans to prevent excessive physiological heat strain and its effects. Effort level can be addressed through work/rest cycles, job rotation, and adjustment of work pace.
In our study, work and worker characteristics associated with heat strain and HRI symptoms reporting did not fully overlap. Physiological strain and reported HRI symptoms should not be assumed to be overlapping outcomes for the purposes of evaluating heat prevention interventions for farmworkers. Additional work is needed to understand factors that affect farmworker HRI symptoms reporting and to establish a consensus on specific HRI symptoms for monitoring purposes. During periods of high heat stress, symptoms and indicators of physiological heat strain should both be monitored, if possible, to increase the sensitivity of early HRI detection and prevention for farmworkers. Related structural issues relevant to barriers to reporting symptoms, pay structure, restrooms, and access to healthcare are also critical to consider in reducing risk of HRI for farmworkers.
These multi-level issues should be addressed using a multi-level approach at workplace, community, and individual levels, including through policy development and through the implementation of policies and best practices tools in a manner tailored for agricultural workers. The need for further action is becoming increasingly urgent as the frequency and severity of heat events are projected to increase in the future [
10,
62].
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.