Maternal work stress is a major problem for working mothers, but its impact on both the continuation of breastfeeding and on maternal mental health is complex and not yet fully understood. This study used large open-access datasets from Open Science Framework (n=2,010) to examine the interplay of these complicated relationships. Statistical and machine learning methods of analysis were employed to explore the relationship between maternal work stress, sociodemographic variables, breastfeeding duration, and maternal mental health markers. Traditional regression models showed that variables like maternal education were predictive of birth weight; they accounted for relatively little variance (Polynomial R 2 = 0.24), implying that direct, linear associations fall short of representing the complete picture of maternal-child health outcomes. In contrast, a Random Forest classification model correctly predicted cessation of breastfeeding at a high rate (85%), accurately identifying intricate, non-linear patterns that are obscured by means of traditional approaches. Clustering analysis further revealed that work stress and breastfeeding are not directly related, but are moderated by different subgroups that are determined by sociodemographic factors such as access to healthcare, social support, and work culture. A high level of maternal distress was detected across all groups, pointing to a widespread public health concern. These findings emphasize the limitations of oversimplified cause-and-effect models in understanding maternal health. They underscore the need for multi-level interventions, such as workplace policies that support mothers, early mental health screening, and targeted public health campaigns. Addressing these issues is not only instrumental in enhancing breastfeeding outcomes and maternal well-being, but also integrating early work-stress screening and lactation support as a part of standard maternal care will help in identifying at-risk mothers and thereby provide personalized psychological and breastfeeding therapies for better outcomes.
Darin Mansor Mathkor (Fri,) studied this question.