Sociodemographic characteristics and predictive factors of attrition: Comparison in two final waves of a birth cohort study in Ecuador

Nataly Cadena^1,2*Alexis J. Handal³Fabián Muñoz⁴Fadya Orozco²

• ¹Universidad San Francisco de Quito, Quito, Ecuador
• ²Centro de Transferencia y Desarrollo de Tecnologías CTT-USFQ, Quito, Ecuador
• ³Department of Epidemiology, School of Public Health, University of Michigan, Ann Arbor, Michigan, United States
• ⁴Visor Análisis Estadístico, Quito, Ecuador

Background. Birth cohort studies are essential to investigate maternal and child health outcomes, yet they face persistent methodological challenges. A major concern is attrition, as participant loss over successive waves can compromise validity and introduce bias. These challenges are particularly acute in low-and middle-income countries, where socioeconomic inequalities and structural barriers further exacerbate participant loss and complicate long-term follow-up. Objective. This paper compares attrition between participants who remained and those who dropped out of the birth cohort study, SEMILLA. We analyze reasons for drop out, and the sociodemographic characteristics and predictive factors associated with attrition. Material and Methods. Recruitment occurred over 30 months. Events such as the COVID-19 pandemic and social conflicts between 2019–2022 affected the final follow-up. The baseline sample included 409 pregnant women, divided into two Final Waves (FW): FW1 completed participation up to the baby's 12 months (n=115), and FW2 up to 18 months (n=294). Dropouts were identified by miscarriage, loss to follow-up, voluntary withdrawal, or protocol noncompliance. Baseline variables included ethnicity, years of schooling, maternal occupational activity, and per capita income. Attrition was calculated for each criterion overall and by Final Wave. Fisher's Exact Test, Pearson's chi-square, and Wilcoxon rank-sum tested differences between participants and dropouts. Logistic regression identified predictors of attrition in each Final Wave. All analyses were conducted with 95% confidence. Results. Of 409 participants, 94 dropped out: 19 in FW1 and 75 in FW2. The main reasons were protocol non-compliance (54%), voluntary withdrawal (21%), miscarriage (13%), and loss to follow-up (12%). In FW1, younger age was associated with attrition (p=0.031), while in FW2, Mestiza ethnicity (p=0.037) and lower income (p=0.014) were significant. Logistic regression showed that older maternal age (OR=0.87, p=0.026) and higher income (OR=0.99, p=0.034) predicted lower attrition. Conclusion. Dropouts increased with longer follow-up, mainly due to time constraints. Age and income disparities significantly predicted continued participation. In contexts with socioeconomic challenges, these factors also affected protocol compliance. Findings underscore the importance of addressing socioeconomic determinants to strengthen the validity and sustainability of longitudinal studies in similar settings.