Survey topics were identified using information obtained from a literature review, analyses of WA workers’ compensation HRI claims,³ and focus group sessions with WA crop workers.¹³ Survey topics included work history and current work activities; work payment methods; breaks and hours typically worked; work exertion, hydration, cooling methods, and clothing; health and HRI symptoms; medications, alcohol and tobacco use; level of concern about workplace heat exposure; and HRI training.

Survey questions were adapted from existing validated surveys when possible, modeled after questions from a validated A-CASI survey instrument designed to identify risk factors for cholinesterase depression in agricultural pesticide handlers in WA,²² or developed by the research team when previously used, validated survey questions were not available. Assessment of workplace exertion was adapted from the Borg and OMNI Rating of Perceived Exertion scales. ^23–25 Draft questions were developed in English and then translated into Spanish and audio recorded by bilingual and bicultural project staff members. Questions about factors that change over time, such as work tasks and activities, asked about the past week to minimize recall bias. In other contexts, one-week recall questions have yielded reliable and valid results.^26–28

The survey was developed using Open Data Kit (http://opendatakit.org/), a freely available platform for Android devices. The survey included Spanish and English narrations of questions and photographs and illustrations, which were designed to be vivid and realistic, characteristics that have been shown to facilitate understanding in low-literate, Latino farmworkers.²⁹ A group of six crop workers representative of the study population evaluated the survey instrument for content validity and usability. The survey was iteratively revised based on this feedback and suggestions from collaborators at Oregon State University, who adapted the survey for use in a separate study of agricultural workers.¹⁹ The final survey instrument consisted of 64 items.

Seventeen outdoor crop workers from one WA orchard participated in concurrent validity and test-retest reliability evaluation of the survey during the summer of 2013. These workers were observed by trained research staff, who recorded observations on clothing, the type and quantity of beverages consumed, how workers cooled themselves (e.g. sitting in the shade), when workers started and ended their work days, durations of employer-mandated and self-initiated breaks, and descriptions of tasks, during four workdays on standardized forms. Three of the four days occurred within one week, and observational data collected on these days were used for validation analyses. Body mass index (BMI) was calculated from measured height and weight as (weight[kg]/height[m]²). Project staff members assigned work tasks to exertion categories based on the American Conference of Governmental Industrial Hygienists (ACGIH) Heat Stress Threshold Limit Value (TLV) metabolic rate categories ³⁰ and project staff consensus, with exertion ranging from “light” to “very heavy.” Demographic characteristics, work activities, and certain health characteristics that were not expected to vary over time were selected for reliability evaluation (Appendix 1). Questions that asked about activities or behaviors that were not observable at the workplace, such as medication use and chronic health conditions, were not evaluated for validity. Participants who were observed took the survey on the first and last days of observations (spaced 15 days apart).

Concurrent test-retest reliability and validity statistics (percent agreement and kappa coefficients) for survey responses are shown in Appendices 1 and 2, respectively. In general, survey questions covering demographics, health status, health conditions, training, health behaviors, and HRI appeared to be reasonably reliable (% agreement 71–100% or kappa 0.70–1.00, comparing participant responses at each survey administration day). Survey questions assessing work tasks, times, payment schemes, types of beverages consumed, workplace shade, and certain clothing questions demonstrated acceptable validity (% agreement between survey responses from the first survey administration day and field observations 71–100%).

Adults engaged in outdoor, summer crop work in Central or Eastern WA were eligible to participate in the study. During the summer of 2013, bilingual and bicultural project staff members, who reside in Central and Eastern WA, contacted local orchard and farm supervisors and individual crop workers. Sampling was not random; research staff contacted growers and workers whom they felt were likely perform outdoor summer crop work. Research staff asked for permission from employers to recruit workers at their workplaces. Project staff travelled to workplaces or mutually-agreed upon meeting locations, explained the goals of the project, and asked eligible workers if they were interested in participating. Interested participants provided informed consent.

The survey was self-administered on touch screen tablets (Asus Eee Pad Transformer Prime 10.1 inch screen, ASUS Computer International, Fremont, CA, USA) to 100 participants from 9 workplaces (median [range] of 6 [2–28] participants per workplace) in Central and Eastern WA from July 2013 through September 2013. Twenty of these participants were additionally recruited to participate in the previously described reliability and validity studies (two dropped out in the middle of the study, and one did not complete the first survey, leaving 17 for the reliability and validity analyses). Comparisons between the full participant group (N=97) and the observation participants (n=17) are shown in Appendix 3. The University of Washington Institutional Review Board approved all study procedures.

The outcome was defined a priori as self-reported HRI symptoms (dizziness/light-headedness or heavy sweating, versus none of these symptoms, during a hot day at work in the past week). The survey asked about specific symptoms, as participants were not assumed to know which symptoms were associated with heat strain or HRI. This a priori combination of specific symptoms was used as a single outcome variable in the analyses. The outcome definition (light-headedness/dizziness or heavy sweating) focused on symptoms that are both symptoms of HRIs and also reflect underlying physiological mechanisms that, when overwhelmed, can lead to heat stroke. Increased cardiovascular demands and heavy sweating (particularly without adequate fluid replacement) can lead to inadequate delivery of blood to the tissues and associated symptoms of light-headedness/dizziness, less efficient evaporative and convective heat loss, and a rise in core body temperature.⁴⁰ Symptoms of light-headedness/dizziness and heavy sweating are also associated with heat syncope and heat exhaustion. Of note, although fainting was included in the original outcome definition, no workers reported fainting. Heat rash, cramps, headache, fatigue, and nausea/vomiting were reported (Appendix 3) and can also be associated with HRI, but these symptoms were not included in the outcome definition because they are often caused by other illnesses and may not be directly related to underlying physiological mechanisms of interest. Dizziness/light-headedness can occur as a result of hypoglycemia in diabetics, particularly those taking certain diabetes medications. However, none of the participants that reported dizziness/light-headedness during a hot day at work reported being told by a health provider that they had diabetes. Reactive and fasting hypoglycemia is relatively rare in non-diabetics, particularly those that are relatively healthy (41).

Potential HRI risk factors in the following domains were selected a priori for inclusion in the risk factors analysis based on the existing scientific literature: 1) demographic; 2) HRI training in the past year; 3) work factors; 4) hydration; 5) clothing; 6) health; and 7) environmental conditions. Preference was given to potential risk factors for which corresponding survey questions had acceptable performance in reliability and validity evaluations (Appendices 1 and 2). The variables included in the risk factors analysis are shown in Table 1.

Hourly temperature and relative humidity data were obtained from Washington State University’s AgWeatherNet weather station program,³¹ and used to calculate hourly heat indices using standard methods,^32,33 as previously described.³ Maximum daily heat indices for self-reported work hours for each participant were used to compute mean maximum daily heat indices over the past week (HI_max), as the past week was the duration of recall of most survey questions.

Ninety-seven participants’ responses were included in the analyses. Of the 100 participants to whom the survey was administered, three participants’ responses were excluded from the descriptive analyses because they did not complete the survey (n=1) or they indicated that they did not work during the preceding week (n=2), the timeframe asked about in the majority of the survey questions.

Separate mixed effects logistic regression models, with random effects for workplace, were constructed for each domain of risk factors. All variables were coded as categorical variables, as shown in Table 1, except age (years), HI_max (°F), and BMI (kg/m²), which were coded as continuous variables, in regression models. Variables with a P-value < 0.50 in single-domain models were entered together into a multi-domain mixed effects logistic regression model, with a random effect for workplace, of HRI. Statistical analyses were performed using Stata 10 (StataCorp, College Station, TX, USA).

Characteristics of the study population are shown in Table 1, and additional details are shown in Appendix 3. The majority (91%) of participants were born in Mexico, and nearly all identified as Latino/a. The mean (standard deviation) age was 41 (13), 53% of participants were male, and over half of participants reported only a primary school education. Fifty-nine and 11% of participants reported being able to read very well in Spanish and English, respectively. The majority of participants reported working with tree fruit, and common tasks included harvesting and thinning green fruit.

The mean (standard deviation) BMI was 28 (4) kg/m². Thirteen percent of participants reported that a healthcare provider has told them they have diabetes, and 12% reported taking medications for hypertension in the past week. Approximately one third of participants reported experiencing HRI symptoms (light-headedness/dizziness or heavy sweating) during a hot day at work in the past week. Ninety percent of participants reported starting work for the season at least three weeks before the survey, and the mean (standard deviation) number of days worked in the past week was 4.9 (1.5), indicating that most participants were likely acclimatized to the Central/Eastern Washington outdoor summertime environment.

Seventy-four percent of participants reported feeling that they were allowed to take extra breaks if needed to rest or drink water. Approximately one third of participants reported usually having to walk for more than three minutes to get to the toilet. Only about one third of workers reported receiving training about working outdoors in the heat or health effects of working in the heat in the past year. Approximately half of the participants reported being paid by the piece. The mean (standard deviation) HI_max during reported working hours was 84 (2) °F. The temporal and geographical distribution of HI_max during the study period is described in Figure 1. During the study period, the maximum daily temperature ranged from 77°F to 97°F. Mean temperatures in July and August in Central/Eastern Washington area are typically in the 70s °F.³¹

Workers reported drinking water (96%), including water brought from home and provided at work, soda (31%), sports drinks (23%), juice (8%), energy drinks (6%), and coffee or tea (4%) at work. Fifty-seven percent of workers reported usually drinking water every thirty minutes or more in the past week. The majority (92%) of workers reported access to shade from trees at work. Nearly all workers reported wearing some type of head covering, over three quarters of participants reported wearing a light-colored shirt, and 13% reported wearing some type of personal protective equipment (3% Tyvek® or chemical resistant suits; 1% respirator) at work in the past week.

Participants reporting HRI symptoms (light-headedness/dizziness or heavy sweating) in the past week, compared to participants who did not report HRI symptoms, were more likely to report being female, not having HRI training the past year, being paid by the piece, not feeling that they were allowed to take extra breaks to rest or drink water, working harder, having a greater distance to walk to the toilet, drinking caffeine, drinking less frequently, and having good or fair (versus excellent or very good) general health (Table 1). The mean (standard deviation) age was lower in participants reporting HRI (36 [13] years), compared to participants not reporting HRI (43 [13] years), and participants reporting HRI were less likely to report being told by a healthcare provider they had diabetes or using anti-hypertensive medications.

Results from the final multi-domain mixed effects logistic regression model are shown in Table 2. An increase in age was associated with a lower odds of HRI (odds ratio [OR] 0.92; 95% confidence interval [CI] 0.87–0.98). Piece rate compared to hourly pay (OR 6.20; 95% CI 1.11–34.54), and needing to walk for more than three minutes to get to the toilet, compared to less than three minutes (OR 4.86; 95%CI 1.18–20.06), were associated with a higher odds of HRI.

This study has several important limitations. First, outdoor crop workers in WA were not randomly sampled. Participating workplaces may have been more likely to engage in HRI prevention, leading to an underestimate of HRI symptom prevalence. It is also possible that workers that participated are systematically different than all WA outdoor crop workers. Second, the HRI outcome, and personal and workplace risk factors, were self-reported. Risk factor analyses incorporating an outcome of heat strain estimated from core body temperature and heart rate, using established methods such as the Physiological Strain Index,³⁹ could provide further insight into HRI risk. In comparable populations, objective measures could complement survey questions that were determined to be reasonably reliable and valid in this study (Appendix 4). Third, this study is cross sectional and relatively small. There may not have been sufficient power to identify all HRI risk factors. Finally, the results of this study, which was conducted in Latino crop workers in WA, may not be generalizable to all crop workers.

In this study of Washington crop workers, decreased age (and less work experience), piece rate pay, and longer distance to the toilet were associated with self-reported HRI. Modifiable workplace factors should be considered in HRI prevention efforts that are evaluated using validated, objective measures in representative working populations.