Recruitment process for studying occupational and environmental chemical exposures in a birth cohort study in Ecuador: challenges dealing with a stratified sample.

backgroundThe Study of Environmental Exposure of Mothers and Infants Impacted by Large-Scale Agriculture (SEMILLA) is a community-based birth cohort study carried out in the Cayambe, Pedro Moncayo region, Ecuador, with intensive industrial floricultural activity. Maternal occupational and environmental pesticide exposure was assessed among three groups of pregnant women stratified according to "maternal occupational activity" at the time of recruitment. This article describes the recruitment process and strategies used, determines their effectiveness in targeting pregnant women with different levels of exposure, and discusses reasons that influenced recruitment rates.

methodsThe projected sample was stratified into three occupational activity groups: floricultural/agricultural workers, non-agricultural workers, and non-workers (women not working outside the home for pay). Recruitment strategies include community activities, collaboration with health care providers, information from obstetric records (“Obstetric Census”), and non-monetary recognition. Analyses were conducted by occupational activity at the time of recruitment. Recruitment rates were calculated by occupational group. Monitoring indicators included the number and percentage of participants recruited by strategy. Wilcoxon rank-sum test and Fisher's exact test were used to compare characteristics between eligible and participating pregnant women; the Kruskal–Wallis test was used to compare among participant groups.

resultsA total of 409 participants were recruited, 74% of floricultural/agricultural workers, 81% of non-agricultural workers, and 64.5% of non-workers. The strategy that referred the highest number of participants was the “non-monetary recognition” (60.9%), proving its effectiveness for recruiting floricultural/agricultural and non-agricultural workers (74.5% and 67.7%, respectively). The “Obstetric Census” was the second most important strategy (30.6%), especially effective in recruiting non-workers (49.6%). Compared to floricultural/agricultural and non-agricultural workers, non-workers were younger (p < 0.001) and had a significantly higher average number of prenatal check-ups (p = 0.0006). Floricultural/agricultural workers had a significantly higher average number of pregnancies and living children than the other two groups (p > 0.01). Time restrictions among women working outside the home for pay, and lack of autonomy among non-workers, were crucial factors that limited participation in the study.

conclusionsStratifying recruitment by comparability groups requires differentiated strategies. Understanding living conditions and the social context of the study region are critical to effectively targeting recruitment efforts.