The implementation of community-based cohort studies during pregnancy, with the objective of following the population longitudinally, is essential to prospective studies and achieving answers to crucial questions about a variety of exposures that influence the health outcomes of the mother‒child binomial (1–3). However, these studies face multiple challenges, primarily the progressive decrease in the size of the initial sample during the follow-up waves, mainly due to loss to follow-up by study participants. A greater number of follow-up waves in studies tends to increase dropout rates (4–7). This methodological challenge may impact the results as a potential source of bias (8, 9).

A wide range of individual, family, social and demographic factors influence the decision to drop out of a birth cohort study. The literature shows that the implementation of this type of study in low- and middle-income countries, such as those in Latin America, faces greater challenges related to the complex political, economic and social context in which these studies are developed (1–3). In these countries, the percentages of disadvantaged and vulnerable populations with less education, lower income, lack of employment, and lack of social support, and of individuals who are ethnic minorities are high. These populations tend to be less likely to enroll in research studies and, in turn, have a greater probability of dropping out over time (1, 4, 10, 11). Furthermore, in these social contexts, women have a lower level of education, a lower socioeconomic status, and limited access to support networks; moreover, they are likely to be young mothers and have less autonomy. These characteristics reduce the chances that they will continue to participate in prenatal longitudinal research studies (12, 13).

In addition to the characteristics of the participants, there are factors corresponding to the study design that may increase dropout. These include repeated evaluations, long study durations, and short intervals between each follow-up wave, creating time, emotional, and social burdens for the participants (14, 15). Considering the aforementioned population and methodological characteristics, birth cohort studies often occur with complex structural determinants and a series of challenges to balance the technical design, methodological rigor and, not least, financial investment. In some cases, it becomes necessary to modify protocols to effectively and successfully complete the study (14, 15). This approach may include adjustments to the sample size, the number of waves for data collection and the total duration between the first and final waves (7). However, in the literature, information on attrition in these contexts, in which challenges involving the study design and the sociodemographic characteristics of the participants or in which identifying differences between those who remain in the study vs. those who decide to abandon it are taken into account, is limited (14–16). Notably, such characteristics are useful predictors of possible abandonment (6, 7, 9, 17, 18). On the basis of this knowledge deficit, this paper compares attrition between participants who remained and who dropped out—analyzing reasons, sociodemographic characteristics and predictive factors—in two final waves of birth cohort study, SEMILLA.

The findings from this analysis highlight important reasons for attrition for participants in a community-based birth cohort study, SEMILLA. During Final Waves 1 and 2, the main criteria and reasons for attrition were as follows: (1) failure to comply with protocol requirements, mainly due to moving to another city with a smaller proportion noting apprehension about blood collection from their baby; and (2) voluntarily deciding to withdraw from the study, mainly due to time constraints. With respect to the first reason, in Ecuador, post COVID-19 pandemic, internal migration affected the labor market, increasing uncertainty and family economic conflicts, which prompted migration in search of other employment opportunities (24). According to the World Bank (2021), 77% of families in low- and middle-income countries lost their jobs during this time (7, 25). In Ecuador, an estimated 1.5 million people fell into poverty, forcing many to migrate in search of work in other cities and regions (24). To a lesser degree, in this study, apprehension regarding the collection of the baby's blood sample, was noted as a reason for failure to comply to study protocol in 25% of those who dropped out. This is an aspect that is mentioned in the literature because the most vulnerable research participants tend to experience greater anxiety around procedures for obtaining biological samples from their children. This fear may be linked to widespread mistrust in contexts of vulnerability and social insecurity. For example, a prospective pediatric cohort study on typhoid surveillance in Vellore, South India, revealed that mothers who perceive their children to be at risk in research studies are more likely to drop out. For these women, it may be easier to participate during pregnancy (26). Similarly, Daniels et al. (2006) reported that women in poverty are often particularly concerned with the collection of physical or biological data related to their offspring, especially during the neonatal period. Finally, Brumatti et al. (2013) reported that the participants in the Phime, Italy cohort study were concerned with the possible risks associated with the collection, long-term storage and future use of biological samples from their children, as well as exposure to long and stressful neurodevelopmental tests (25). This discomfort with the study procedures and activities contributes to the increase in loss to follow-up, especially among the most vulnerable women (27).

With respect to the decision to leave the study voluntarily, Kingston and Jagger (2017) have suggested that the follow-up methodology used in a cohort study plays an important role. For example, participation in studies that require a high volume of data collection with surveys that last more than an hour and with short intervals between each wave of follow-up can be an overwhelming time load for women with limited resources and multiple responsibilities (15, 33). Other authors, such as Goldstein et al. (2021), agree that the high burden of participation makes it difficult for women to balance their involvement in the study with their family and work activities. Additionally, the authors highlight that privacy concerns may also influence the women's decisions to withdraw from a research study especially during situations of economic and social vulnerability. Some participants may feel uncomfortable sharing personal information or fear the disclosure of sensitive data (27).

With respect to the sociodemographic characteristics of the participants, a significant difference was observed between the women who remained in the study and those who dropped out. In FW1, those who dropped out were younger; whereas in FW2, they were predominantly Mestiza and had a lower monthly per capita income. When considering the finding on age, in this context, retention in birth cohort studies has been found to increase among older participants (24). In fact, in India, the participation rates were 90% among older women and 68% among younger women (26). Additionally, Sindhu, Srinivasan et al. (2019) noted that the factors contributing to high attrition rates among young mothers are low literacy levels, low socioeconomic strata, and ethnic minority status (10). Additionally, the literature indicates that in Latin American contexts, young women tend to lack autonomy, which may significantly affect their participation in birth cohort studies. This lack of autonomy can be understood from various perspectives, such as the control exercised by the partner, or the family or the social norms that limit their ability to make personal decisions. In various sociocultural contexts, the interruption of women's participation in birth cohort studies can be influenced by the opinions of the couple, especially if they consider such studies to be uncomfortable or waste time. This is particularly evident in traditional family structures, where women must assume domestic or caring responsibilities, making it difficult to balance those obligations with the demands of study (25, 27, 28).

With respect to ethnicity, in the present study, most women self-identified as Mestiza, and this may have contributed to the higher percentage of dropouts in this category for FW2. Our findings noted more attrition in FW2 for those with lower monthly per capita income. Young, Powers and Bell (29) note that in community cohort studies conducted in countries with characteristics similar to those of Ecuador, women with better economic conditions are more likely to continue in a prospective study because it may be easier for them to cover the costs associated with the time invested. They may also face fewer social, cultural and educational barriers that would prevent them from fully understanding the benefits of participating in the research (1, 23).

Concerning the findings on the predictors of attrition, significant disparities were observed between the women who remained in the study and those who dropped out, the latter being those who seem to be in worse socioeconomic conditions, which may have increased their low autonomy and greater vulnerability. Participants with lower incomes face greater economic difficulties, making them prioritize the need to work or attend to financial problems. In addition, they tend to have limited access to resources, so they do not perceive immediate direct benefits from participating in a prospective study, such as financial compensation (30–32). On the other hand, women with low financial resources also face barriers to support domestic and caregiving responsibilities, which makes it difficult for them to dedicate time to participate in these types of cohort studies (25, 27).

Among the main limitations of the study are additional factors that could have contributed to the observed attrition that were not considered, such as having timely information on the perceptions of the participants about the benefits of the study. Similarly, having more participants in FW2 than in FW1 could have generated an unbalanced representation in the analysis of the Final Waves, which could have made comparisons related to the time of participation difficult. No less important are the contextual factors that affected the development of the study, which may have influenced attrition. Despite these limitations, the present study included various sociodemographic variables, which allowed for a detailed analysis of the factors that could influence study attrition. Additionally, the fact that the study was completed through two Final Waves provided the opportunity to compare different follow-up periods and analyze how the dropout factors may vary over time and between groups.

In sum, factors such as younger age and having a lower income seem to be relevant in both high-income countries and low- and middle- income countries, although studies in which the characteristics of dropouts and the reasons for them in low- and middle-income countries remain limited (30–32). The sociodemographic characteristics of the women who dropped out of the SEMILLA study may have potential implications for the study's results and should be considered in future analyses, as existing literature suggests that factors such as maternal age and income can influence the neurobehavioral development of children (30–32).

Attrition analysis in the SEMILLA study showed that dropouts increased with longer follow-up duration. The main reason was protocol non-compliance, often linked to time constraints and competing demands faced by women in vulnerable socioeconomic conditions. Age, income, and ethnicity emerged as important predictors of continued participation, highlighting structural inequities that affect retention in longitudinal studies.

These findings emphasize that attrition is not only a procedural challenge but also a reflection of broader social determinants. Future cohort studies in similar contexts should incorporate strategies to reduce participant burden, such as flexible scheduling, simplified follow-up protocols, and hybrid data collection modalities, and strengthen engagement mechanisms to minimize loss to follow-up. Providing targeted support for younger, lower-income, and ethnically marginalized women could help ensure greater inclusivity and validity in long-term maternal child health.