Psychological and social factors associated with breastfeeding success: A meta-analytic path analysis
Breastfeeding and depression are both issues of critical importance in the postpartum period, but in opposite ways. While breastfeeding has numerous benefits for mother and child, maternal depression places children at risk of developmental problems and puts undue burden on mothers. Both of these issues are best understood in the context of multiple factors. In addition, these two issues are intertwined with one another. While there are many variables that contribute to the context around breastfeeding and depression, there is a limited set upon which an intervention could have an influence. Therefore, the variables selected as the context around breastfeeding and depression were: stress, social support, and relationship quality/presence. Considering that there is considerable literature, though spread out and rarely brought together, surrounding the issues of breastfeeding and depression, a Meta-Analytic approach was used to aggregate the data across the studies, then the meta-analytic correlation matrices were provided to structural equation modeling software to fit a path analysis. Overall, fixed and random effects meta-analyses provided similar results, with some notable differences. Fixed effects results typically had smaller confidence intervals, were more likely to be significant, and were more likely to be significantly heterogeneous than the random effects results. Contrasts were used to follow-up significant heterogeneity. The only study feature which resulted in significant follow-up contrasts for the multiple bivariate relationships was year of publication, which was possibly related to relationships changing over time, potentially due to effective interventions. The data were then re-analyzed, splitting the studies into two groups by year of publication. These four new meta-analytic matrices (one per group fixed effects and one per group random effects) were then used as the data for additional path analysis models. Multiple Group Modeling was used with the path analyses to test for significant differences by year of publication in the overall model. In spite of the significant contrasts in several individual bivariate relationships, year of publication was not a significant difference when fitting the path model. The hypotheses were partially supported. Relationship quality/presence and social support were significant predictors of stress and depression in the hypothesized directions. In the fixed effects constrained multiple group modeling results, stress was a significant predictor of breastfeeding controlling for depression, but no other models resulted in significant prediction of breastfeeding after controlling for other factors.