, vol. 47, issue 2, pp. 1-34, 2014.
Using Dynamic Random Effect Models in Studying Determinants of Intrauterine Fetal Loss in Egypt
By
Somaya M. El-Saadani Sara Assem Ibrahim
Abstract:
Fetal death, at any point during gestation, is a traumatic event to the family. Maternal deaths reached 287000 deaths in 2010; 99% of these deaths happen in developing countries. Spontaneous fetal loss is considered one of the major causes of these deaths.
This study aimed to examine the effect of direct and indirect determinants on the probability of having loss and investigate the effect of directly previous pregnancy outcomes on the index pregnancy. The study applied Random Effect Dynamic Probit Model as a proper methodology to take into account the dependence between observations and to test for state-dependence among pregnancies (lag effect) using data of Social Mobility and Fertility Survey (SSMF 1995)
Results provide evidence on the dependence between pregnancies that belong to the same mother and on the significance of the unobserved heterogeneity variable.
Results provide evidence that there is no significant effect of previous loss on the probability of loss in case of adding the pregnancy interval variable to the model because of the relationship between previous loss and the small pregnancy intervals. The following variables; maternal age (greater than 35 years), first pregnancy, pregnancy interval, previous pregnancy loss (if pregnancy interval variable is omitted from the model), pregnancy cohort, and place of residence are statistically significant, while the variables; pregnancy order, mother’s level of education, father’s level of education, marrying relatives are not statistically significant.
Keywords: Dynamic random effect probit models, unobserved heterogeneity, state dependence, transition matrix, maternal age, first pregnancy, previous pregnancy loss, pregnancy interval, pregnancy order, and place of residence.
