The data for this analysis are from a demonstration project evaluating the feasibility of using the La Familia Sana program as a public health tool to promote pesticide safety knowledge among migrant and seasonal farmworker families.^3,4 The demonstration project involved a partnership of the North Carolina Farmworkers Project (NCFP), which was responsible for implementing the La Familia Sana program, and Wake Forest School of Medicine (WFSM), which was responsible for evaluation. The NCFP implemented the La Familia Sana program through 6 agencies serving a total of 13 counties in east central North Carolina. The project goal was to enroll and educate participants from 600 families over 18 months (April 2009 – September 2010). To be eligible to participate, the participant had to be at least 18 years of age or married, self-identify as Latino or Hispanic, live in a household containing at least one farmworker employed in agriculture during the past 12 months, and plan to stay in the dwelling for the next 2 months. One adult in the household had to be pregnant or have at least one child 12 years of age or younger living in the dwelling. All study procedures were approved by the Wake Forest University Health Sciences Institutional Review Board.

La Familia Sana has been described elsewhere,⁵ but is summarized here to provide context for the study. The program consists of 6 pesticide safety education lessons taught by a promotora through a minimum of 5 home visits. Topics for the 6 lessons are (1) protection of the family by knowing pesticide exposure symptoms and long-term consequences, (2) pesticide residues and paraoccupational exposure pathways, (3) exposure reduction to agricultural pesticides at home, (4) integrated pest management (IPM) at home, (5) pesticide risks for pregnant women and young children, and (6) persuasion of others to change behavior. The Health Belief Model (HBM)⁶ provided the theoretical foundation for the overall project. The content of each La Familia Sana lesson focused on increasing participant’s knowledge about pesticides and providing and practicing concrete strategies for reducing pesticide exposure.

The “knowledge” component of the lessons targeted several of the HBM’s main concepts (italicized). For example, perceived severity was targeted in the first lesson by conveying that research has linked pesticide exposure with a variety of long-term health outcomes like cancer, memory and learning impairment, and sterility.^7,8 Similarly, the fifth lesson targeted perceived severity by highlighting the link between pesticide exposure and congenital anomalies in children⁹ and explaining that the health effects of pesticide exposure are compounded in fetuses and small children due to their small size and rapid development.¹⁰ Perceived susceptibility was conveyed in several lessons through descriptions of pesticide residues and activities that showed participants everyday places children come into contact with pesticides. Similarly material in lesson 6 addressed a common perceived barrier confronted by women in farmworker households, facilitating behavior change in the men living with them. The concrete strategies introduced in each lesson targeted perceived benefits that have strong face validity (if one deep cleans a house, one can reduce pesticide exposure) or are advocated in WPS training (eg, washing work clothes separately from nonwork clothes). Each lesson targeted self-efficacy through the liberal use of reflection points, role-plays and activities designed to help participants integrate advocated pesticide safety practices into daily life. The intervention itself was a cue to action to promote pesticide safety behavior, as were several of the concrete strategies for reducing exposure. For example, one suggestion in lesson 2 was to place a mat outside the door as a visual reminder (ie, cue to action) that work boots should be removed before entering the home.

Pre- and posttest interviewer-administered survey questionnaires were constructed with a total of 68 items designed to elicit knowledge or self-reported practices related to the lesson objectives. Both the pre- and posttest questionnaires were piloted on a small group of women in Latino farmworker families and revised based on feedback from the pilot. Because the vast majority of the questions are about concrete behaviors with discrete answers as opposed to abstract latent constructs, pilot testing revealed little difficulty understanding the intended meaning of the questions or issues articulating a response. Pretest interviewer-administered survey questionnaires were conducted with program participants prior to their receiving any lessons from the promotoras. Posttests interviews were conducted after all promotora lessons had been completed. Participants received $10 after completing the pretest and $20 after the posttest.

Seven interviewers identified with the help of liaisons collected all evaluation data. All interviewers were native Spanish-speakers, several had prior experience conducting survey interviews, and all were trained for this study. The interviewers field-edited the completed interviews before leaving the participant’s homes. WFSM staff reviewed the interview with the interviewer when collecting it in the field. Two Spanish-speaking staff prior to data entry also reviewed interviews, and any inconsistencies or queries were returned to the interviewer to clarify.

Descriptive statistics were calculated for each measure (pesticide-related safety behavior, conceptual targets, participant involvement in the program, promotora attributes, and adherence to training activities) at the pre- and posttest times as well as the difference between time points (post-pre). A series of linear regression models were fit to examine the association between individual predictor variables and the post-pre change scores for each set of pesticides safety behavior outcomes. Finally, an additional set of incremental models was fit for each pesticide safety behavior outcome. Each additional model’s overall R², the amount of additional variance that was explained (ΔR²) by each model compared to the pretest only model, and the corresponding p-value were calculated to test whether the additional variables improved outcome prediction. All analyses were done using SAS 9.2 (SAS Institute, Cary, NC).

Complete pre- and posttest data were obtained from 610 individuals. Over half of the sample was 30 years of age or older. Over a quarter of the sample reported having fewer than 5 years of formal education, whereas only 11.6% reported completing 12 or more years of schooling (Table 1). Nearly all participants were born in Mexico, and most (54%) had been living in the United States for 9 or fewer years, although 46% had been in the United States for 10 or more years. Over 80% of participants were currently married or living as married, just under one-third currently had one child or was pregnant at the time of enrollment, whereas a comparable percentage had 2 children, and the remainder had 3 or more children. Most participants expressed the greatest comfort speaking Spanish.

Each type of pesticide-related safety behavior improved from pre- to posttest (Table 2). Increases were also noted for each of the conceptual targets: the score of pesticide knowledge increased from an average of 4.2 (SD = 2.7) to 12.5 (SD = 2.9); the average response for perceived dangerousness of pesticides increased from 3.45 (SD = 0.9) to 3.75 (0.5); and the pesticide safety self-efficacy average increased from 0.95 (SD = 0.6) to 1.37 (SD = 0.6). Over 60% of participants reported engaging with the written materials left behind by the promotora, whereas 87.2% reported watching the provided videos.

In terms of promotora attributes, participants reported having an average of 5 lessons (SD = 0.73), over 85% of participants indicated that the promotora demonstrated high professionalism, and nearly 95% of participants believed the promotora was effective. Virtually all participants had the same promotora throughout the educational phase of the program, and 91.5% of participants (N = 558) received both the residue demonstration and the pesticide mapping, believed to be key training activities. A small number of participants (N = 38, 6.2%) received either the residue demonstration or the pesticide mapping activities (but not both). Only 14 (2.3%) individuals did not receive either of these activities.

Change in pesticide knowledge was the only conceptual target associated with greater use of IPM practices (Table 3, column 1 and 2): change in both perceived dangerousness of pesticides and pesticide safety self-efficacy were unassociated with this outcome. Several participant attributes were associated with increased use of IPM practices from pre- to posttest. Participant educational attainment, language preference, years in the United States, and self-reported engagement with written materials were all associated with changes in participants’ use of IPM activities. None of the promotora attributes were associated with change in IPM.

Increased use of WPS recommendations from pre-to posttest was associated with changes in both pesticide knowledge and perceived dangerousness of pesticides. Change in adherence to WPS recommendations did not differ by any of the participant attributes, except self-reported extra review of video material that was associated with less change in this outcome. None of the promotora attributes were associated with change in WPS recommendations. Change in the safe storage of pesticides in the home was not predicted by any factor.

Pre- to posttest change in the conceptual targets explained the most variance in participants’ use of IPM (Table 4). Collectively, changes in the conceptual targets of the intervention accounted for 6% of participants’ posttest use of IPM, controlling for pretest values of IPM. By contrast, personal attributes and promotora attributes each accounted for approximately 3% of the variance in this behavioral domain. Changes in adherence to WPS recommendations were most influenced by promotora attributes (ΔR² = 0.02) followed by personal attributes (ΔR² = 0.01). Changes in the targeted conceptual levers did not explain any additional variance beyond the pretest-only model. Change in the conceptual targets added significantly to model fit of safe storage of pesticides in the home (ΔR² = 0.02), whereas both personal and promotora attributes did not explain any change in this outcome.