Modeling of Years of Cohabitation, Children Ever Born, and Number of Living Children of Women of Reproductive Age Using Trivariate Poisson Regression

Abstract

Background and Aims: To have the right policies and treatments, population and public health research must comprehend the interrelated dynamics of different cohabitation and reproduction patterns. The effects of cohabitation duration and reproductive experiences on child survival among women of reproductive age are examined in this study. Methods: Employing a trivariate Poisson regression (TPR) model, the analysis captures the interrelated nature of births within the last 5 years, pregnancy losses, and birth registration. Parameter estimation was conducted using the Expectation-Maximization algorithm, which is well-suited for handling complex statistical frameworks. Data were drawn from the 2022 Demographic and Health Survey. Results: The study indicates significant constant covariance among the three outcomes, confirming their interconnectedness. The findings show that the socio-demographic factors exert a strong influence on reproductive histories and highlight the pathways through which reproductive experiences affect fertility behavior and child survival. The methodological framework demonstrates improved statistical efficiency and model fit, as evidenced by the likelihood ratio test, Akaike information criterion (AIC), and Bayesian information criterion (BIC). Conclusion: Importantly, the analysis reveals that cohabitation duration and fertility patterns are strongly shaped by recent pregnancies, pregnancy losses, and birth registration, while child survival is most contingent upon pregnancy losses and registered births. These results emphasize the need for reproductive health policies that provide targeted support for women experiencing recent pregnancies, pregnancy loss, and unregistered births to strengthen child survival outcomes.