In the summer of 2011, we recruited 260 farmworker women ages 18 to 40 who had been working in Central Florida nurseries and ferneries for at least six months and who worked at least 20 h a week in agriculture. We relied on snowball sampling for recruitment at farmworker housing, local churches, schools, area clinics, local ethnic businesses, community events, FWAF programs and other locations frequented by female nursery and fernery workers. Snowball sampling has often been used alone or in combination with other methods to recruit from hard to reach populations, especially in community-based research [18,19,20]. The majority of interviews were conducted at the FWAF offices in Pierson and Apopka, though some were conducted in the workers’ homes, in the late afternoons after work or on rare occasions during the weekend when workers were off. The participants provided information on current and past work conditions and their pregnancy histories. Participants received compensation of $25 and a “Your Work and Your Health” informational pamphlet developed by the FWAF to help farmworkers recognize signs and symptoms of pesticide exposure and to engage in protective measures. All study procedures were reviewed and approved by the IRB of Emory University, and consents and questionnaires were administered in English, Spanish or Haitian Creole, depending on the interviewee’s stated preference.

Farmworker participants were recruited from two geographic locations in Central Florida: (1) nurseries in Apopka, FL (Orange and Lake counties), an area frequently referred to as the “Indoor Foliage Capital of the World”; and (2) Pierson, FL (Volusia and Putnam counties), an area commonly referred to as “The Fern Capital of the World.” The FWAF has offices in both Pierson and Apopka, where community members come for services, referrals, assistance, membership meetings and events.

For this study, “nurseries” were defined as those companies producing a variety of ornamental plants, including cut flowers, flowering potted plants, hanging baskets, potted foliage, bedding and garden plants and woody ornamentals. Most nursery workers labor inside greenhouse structures enclosed with non-porous heavy plastic. Work tasks at nurseries are varied and may include: planting at conveyor belts; loading pots of plants onto trays; and loading and carrying trays, boxes or bags of soil. According to community-based estimates, there are 10,000 to 12,000 workers in the Central Florida nursery/foliage industry. Nursery workers are primarily of Mexican origin, but there are also indigenous people from Central America, Hispanics from other countries, Haitians and African-Americans. A previous study with Florida nursery worker households showed that a majority of workers (71%) were married or cohabiting and about half (50.2%) had children in the household [21].

“Ferneries” were defined as companies producing cut ornamental foliage, such as that used as floral greenery. Fernery workers labor in fields enclosed on all sides and across the top by porous black saran shade cloth or under tree canopy. To harvest ferns, workers bend over, thrust their arms into masses of ferns, cut fronds at their base, and secure them into bunches of 20 to 25 fronds. They leave the fern bunches on the ground until they have a particular quantity, then they gather up all of the bunches in their arms and quickly carry them to a trailer waiting at the edge of the field. The most experienced fern cutters can harvest up to 300 bunches of leatherleaf ferns a day. According to community-based estimates, there are an estimated 13,000 workers in the Central Florida fern industry. The majority of fernery workers are of Mexican origin, but there are also indigenous people from Central America. A previous study found that the majority of fernery workers surveyed (76.2%) were married or cohabiting and a large proportion (64.9%) had children in the household [21].

A comprehensive cross-sectional survey was developed by the project team partners with input from a community advisory team and administered by trained community interviewers. The questions were developed using data from focus groups with women working in agriculture completed in the initial phases of this CBPR project [22,23]. Survey items were included to capture workplace conditions, as well as perceptions of workplace risks, health symptoms experienced by the women (relating to heat, ergonomic and pesticide workplace stressors) and pregnancy histories [21]. The survey obtained information on demographics (e.g., age, country of birth, native language and pregnancy status), work practices (e.g., work tasks, health symptoms), work-related hygiene (e.g., access to hand-washing facilities, availability of clean water, use of bathroom), health beliefs (e.g., perceptions of pesticide exposures and health risks), a brief lifetime pregnancy history and a detailed history of the most recent pregnancy (see the Supplemental Material for the survey). Study questionnaires were translated into Spanish and Creole by bilingual FWAF staff members who understood the meanings of the original questions and back translated by independent parties. Survey questions were pilot tested on a sample of 15 FWAF community members (i.e., farmworker members) who previously worked in nursery or fernery operations. This article focuses on findings related to a woman’s most recent pregnancy history.

Pregnancy health-related categories included in this retrospective, cross-sectional analysis were: (1) the duration of work in agriculture while pregnant; (2) access to prenatal care and physician diagnosed health conditions (mother and child); (3) health symptoms during pregnancy; (4) pregnancy outcomes and infant health for the first year of life; (5) breastfeeding practices; (6) fathers work in agriculture; and (7) body pain discomfort. Paternal work in agriculture was determined by the mother reporting whether or not the father worked in agriculture during her last pregnancy. Women who reported experiencing musculoskeletal discomfort while working in agriculture during their most recent pregnancy were asked to complete a body pain discomfort chart. Musculoskeletal pain was categorized into the following nine areas: (1) neck; (2) chest and shoulder; (3) upper back; (4) mid-lower back; (5) upper abdomen; (6) lower abdomen; (7) upper arms; (8) lower arms; and (9) lower extremity.

Pregnancy Outcomes. For this analysis, we defined a pregnancy complication as one of the following events: miscarriage, abortion, stillbirth (>20 weeks) or ectopic or molar pregnancy. The two primary birth outcomes of interest were: (1) preterm birth (birth at a gestational age greater than 20 weeks, but less than 37 weeks); and (2) low birth weight (LBW) (infant delivered weighing less than 2,500 grams).

Infant Outcomes. Women were asked if their child had ever been diagnosed by a healthcare provider with any birth defects or health complications from the time of delivery until 1 year of age; responses were recorded as infant health complications (yes/no). Women were also asked if a healthcare provider had ever diagnosed their child with any health problems, including respiratory or breathing problems, psychological or learning problems and/or other chronic health problems in the first year of life. For this question, women were also asked to specify the diagnosis. Given that the majority of self-reported child health problems were respiratory-related (23 out of 30 responses), we created a bivariate respiratory condition outcome (yes/no). The two infant outcomes considered in the analysis were: (1) child diagnosis of chronic condition (yes/no); and (2) child diagnosis of respiratory condition (yes/no).

Descriptive statistics examined differences in demographic, work-related changes while pregnant, musculoskeletal discomfort and maternal and infant health outcome characteristics between nursery and fernery workers using a Pearson chi-square or Fisher exact test for categorical variables and Student’s t-tests for continuous outcomes. A significance level of p = 0.05 was used, and all bivariate statistical analyses were performed using SAS 9.3 (Cary, NC, USA).

Compared to nursery workers (88.7%), fernery workers (65.7%) were less likely to report being pregnant to their supervisors (p < 0.001), and when female fernery workers did tell their supervisors, they waited almost a month longer than nursery workers (p = 0.02). More than 97% of the women reported receiving prenatal care. Nursery workers reported entering prenatal care at 11.5 weeks and fernery workers at 11.6 weeks (p = 0.96). Fernery workers were more likely to be asked by a healthcare provider (HCP) to alter their work routine (p < 0.001) and were more likely than nursery workers to adhere to physician-suggested work alterations by decreasing their work hours (p = 0.02) and taking more frequent breaks (p < 0.001). Nursery workers reported sitting more than standing at the workplace (p = 0.02) following health provider suggested behavior changes at work (Table 2).

Women working in ferneries while pregnant reported significantly more body pain discomfort in seven out of nine body sites, including the neck (34.3% vs. 10.5%) (p < 0.001), upper back (31.5% vs. 3.5%) (p < 0.0001) and mid-lower back (53.7% vs. 19.3%) (p < 0.0001), upper abdomen (40.0% vs. 15.8%) (p <0.01), lower abdomen (53.7% vs. 24.6%) (p < 0.001), lower arms (22.2% vs. 5.3%) (p < 0.01) and lower extremity (56.5% vs. 15.8%) (p < 0.0001) compared to nursery workers (Table 3).

Maternal health. There were no significant differences between nursery and fernery workers in report of maternal diagnosis of high blood pressure (p = 0.38), diabetes or high blood sugar (p = 0.28) or premature labor (p = 0.76) experienced during their most recent pregnancy. Women were also asked to report on health symptoms experienced during their last pregnancy. Compared to nursery workers, a larger proportion of women working in ferneries reported having eye irritations (17% vs. 3%) (p < 0.01) and throat irritations (15% vs. 3%) (p = 0.02) during their most recent pregnancy. Nursery workers reported significantly more nausea-related symptoms than fernery workers (46% vs. 31%) (p < 0.05) (Table 4).

Birth outcomes. We did not observe a statistically significant difference between occupational groups for pregnancy complications (p = 0.36), low birth weight delivery (p = 0.29) or preterm delivery (p = 0.88) (Table 5).

Infant health. There was no difference in self-report of infant health complications in the first year of life between mothers from the two agricultural groups (10.0% vs. 10.4%, p = 0.94). More than half of the women in both groups reported they breastfed their infants and continued to breastfeed to almost six months of age of their infants. Among fernery workers, 24.7% reported their child had been diagnosed by a healthcare provider with any (chronic) health problems, including respiratory or breathing problems and psychological or learning problems (p = 0.01), compared to only 8.5% of nursery workers. Women were also asked to provide details about the child’s diagnosed health problem. Results showed that significantly more mothers working in ferneries (24.7%) reported a child diagnosis of a chronic health problem compared to nursery workers (8.5%) (p = 0.01). Fernery workers were also 2.7-times more likely to report a child diagnosis of a respiratory condition compared to women working in nurseries (17.6% vs. 6.4%, p = 0.04) (Table 5).

Haitian versus Mexican women. Because 1/3 of nursery workers were Haitian immigrants, we decided to conduct a sub-analysis examining differences between the two agricultural industries based on nationality. Haitians told their supervisors almost two months later than their Mexican coworkers that they were pregnant (p < 0.001). Haitians also reported entering prenatal care at least eight weeks later than their Mexican counterparts (p < 0.001). In terms of work practices, Mexican workers were told to alter their work routine more often than Haitians (p < 0.001). However, Haitians were more likely to stop work due to a reported inability to make changes at work (p < 0.001). In terms of pesticide exposure symptoms, Mexican women reported more nausea/vomiting (p = 0.003), more dizziness/fainting (p < 0.001) and eye irritations (p = 0.04). Haitian women were more likely to report tachycardia/rapid heartbeat (p = 0.02). There were no differences between Mexican and Haitian mothers in pregnancy complications (p = 0.53), low birth weight (p = 0.25) or preterm deliveries (p = 0.17). Mexican mothers were more likely to self-report that their child was ever diagnosed with a chronic medical problem compared to Haitian mothers, who reported no chronic medical conditions among their children (p < 0.01).

One limitation in this retrospective, cross-sectional study is the potential for maternal recall bias resulting in significant over- or under-reporting of pregnancy and infant health outcomes (i.e., outcome misclassification). Less than 10% of our sample reported on their most recent pregnancy occurring within one year of the survey completion date in the summer of 2011, leaving more than 90% reporting on their last pregnancy dating back several years. Research with Norwegian female farmers has shown improved recall of adverse birth outcomes, such as small for gestational age and late pregnancy loss, with a previous pregnancy loss; however, birth records proved more reliable among mothers without a previous pregnancy loss [50]. Additionally, interviewer and low-response bias are important study limitations to consider in cross-sectional studies. To reduce interviewer bias, all community interviewers (four interviewers in Apopka and three interviewers in Pierson) had at least a high school education and experience working with the FWAF on various issues important to farmworker health and safety and were trusted by community members. Each interviewer received 8 h of training over a two-day period on how to conduct the interviews. Moreover, participation rates were high—a total of three women approached refused participation—largely due to the strong FWAF presence in recruitment areas, further reducing the potential for self-selection bias in our study.

We used a snowball sampling approach to ensure adequate sampling and recruitment of female farmworkers in nurseries and ferneries, a vulnerable and hard-to-reach population. One limitation of this sampling approach is the likelihood of selection bias; however, a larger sample size, much like the one employed in our study, may help to reduce selection bias in studies using snowball sampling methodologies [45].

While the literature supports that work in agriculture is a risk factor for adverse birth outcomes, few studies have been able to capture the lived experience of women working in agriculture during pregnancy. One key advantage is that the current community-academic partnership was able to successfully obtain self-reported pregnancy health data for a hard-to-reach female farmworker population. The purpose of this paper was to compare the pregnancy histories of two types of agricultural workers, those employed in nursery operations and those in ferneries. We have previously described the work conditions and differences in these two types of agricultural settings [51]. We did not ask participants the name of their employers, since that could inhibit participation in the study; therefore, we can only describe differences between the two employment settings and not among different employers within the two employment settings. However, in our formative research with this population, we did not find evidence that the work conditions vary greatly from one employer to another within each setting [22]. Our future work will extend beyond these two employment settings and document the working conditions and pregnancy outcomes of women working in other types of agricultural industries, including citrus and vegetable crops. Future studies with this population might include a prospective birth cohort to quantify and compare multiple environmental exposures for pregnant women working in distinct types of agriculture. Using a birth cohort study design, researchers could also investigate the potential effects of these maternal environmental exposures pre- and post-natally on an array of early child health outcomes, including adverse pregnancy events, respiratory illness and neurodevelopmental performance.