Prevalence & Risk factors of Patent Ductus Arteriosus (PDA) in Preterm Neonates: Evidence from Survival & Shared Frailty Model

Abstract

Background: The ductus arteriosus is a leftover fetal artery connecting the main body artery (aorta) and the main lung artery (pulmonary artery). The ductus allows blood to detour away from the lungs before birth. Every baby is born with a ductus arteriosus. After birth, the opening is no longer needed and it usually narrows and closes within the first few days. Patent ductus arteriosus (PDA), resulted when this artery remains open (patent) after birth, is a heart problem that occurs soon after birth in some babies. Its human and economic loss is too much in general if it is not treated at all or treated early. Objective: The overall objective of this study was to assess prevalence of PDA in preterm neonates; their survival time (time to death) & associate risk factors. Methods: To address the study objective, a secondary data from Health Facilities of Mekelle City was collected from 125 preterm neonates who initiated their follow up between December 2019 and December 2020. The Cox PH model with parametric shared frailty distribution where hospitals (health facilities) preterm neonates treated used as a clustering effect in the models. The gamma and inverse Gaussian shared frailty distributions with Exponential, Weibull and log-logistic baseline models was employed to analyze risk factors associated with age at circumcision using socio-economic and demographic factors. All the fitted models were compared by using AIC and BIC values from actual dataset. Results: A total of 125 children were seen at the Health Facilities of Mekelle City during the study period. The result revealed that about 20% of preterm neonates were exposed to PDA while the remaining were not. The AIC value for the three baseline distributions (Exponential, Weibull, and Gompertz) of PH model was found 173.6275, 174.6895, and 175.6086 and the BIC value of those baseline distributions for the same model was also found 216.0522, 219.9425, and 220.8616 respectively. The AIC value for the three baseline distributions for gamma shared frailty model was found 172.5497, 176.6895, and 176.6896 (the same value with Weibull) and BIC value of those baseline distributions for the same model also found 220.8805, 224.7708, and 224.7708 (the same with Weibull) respectively. Based on AIC and BIC values from simulation experiment and graphical evidences, Gamma shared frailty model, with the exponential baseline preferred when compared with other models. The clustering effect (the hospital effect) was significant for modeling the determinants of time-to-death preterm neonates with PDA. The estimated value of ix | P a g e theta (θ) which is a measure of contribution of a frailty component to the model was 1.1056 and a chi-square value of 0.003722 with one degree of freedom resulted a p-value of 0.0027. Based on the result of Gamma shared frailty model with the exponential baseline, gestational age, birth weight, place of delivery at home, maternal history with HIV, and treatment with paracetamol were found to be the most significant risk factors of the outcome variable, survival time (time to death from PDA). The hazard ratio and 95% Confidence interval of gestational age and birth weight was also 2.0742 with CI (0.5905259, 0.9013781) and 2.7191 with CI (1.000007, 1.000614) that yielded a p-value of 0.003 and 0.045 respectively. The hazard ratio and 95% CI for the covariates place of delivery at home, maternal history with HIV, and treatment with paracetamol were 5.1852 with CI (0.7416363, 5.1268725), 2.139 and (2.417719, 263.6325), and 42.0056 with (-3881.608, 3846.564) which yielded a p-value of 0.0092, and 0.007 respectively. They have also a prevalence rate of a unit (for place of delivery at home and maternal history with HIV) and 0.13 for treatment with paracetamol (No). The overall prevalence rate was also yielded 0.32. Conclusions: The model suggested that there is a strong evidence of heterogeneity among health facilities where the preterm neonates were treated. From the candidate models, Gamma shared frailty model, with the exponential baseline was an appropriate model for predicting the PDA data. There was a frailty effect on the survival of the preterm neonates that arises due to differences in the distribution of hospitals of the neonates. The risk factors place of delivery at home, maternal history with HIV, gestational age (week), Birth weight, and treatment with paracetamol were statistically significant for the survival of preterm neonates whereas the other risk factors were not statistically significant. The frailty component had also a significant effect to the model