Modelling the Risk Factors for Birth Weight in Twin Gestations: A Quantile Regression Approach

Abstract

Birth weight is used as a proxy for the general health condition of newborns. Low birth weight leads to adverse events and its effects on child growth are both short- and long-term. Low birth weight babies are more common in twin gestations. The aim of this study was to assess the effects of maternal and socio-demographic risk factors at various quantiles of the birth weight distribution for twin gestations using quantile regression, a robust semi-parametric technique. Birth records of multiple pregnancies from between 1991 and 2005 were identified retrospectively from the birth registry of the Christian Medical College and hospitals in Vellore, India. A total of 1304 twin pregnancies were included in the analysis. Demographic and clinical characteristics of the mothers were analysed. The mean gestational age of the twins was 36 weeks with 51% having preterm labour. As expected, the examined risk factors showed different effects at different parts of the birth weight distribution. Gestational age, chroniocity, gravida and child's sex had significant effects in all quantiles. Interestingly, mother's age had no significant effect at any part of the birth weight distribution, but both maternal and paternal education had huge impacts in the lower quantiles (10th and 25th), which were underestimated by the ordinary least squares (OLS) estimates. The study shows that quantile regression is a useful method for risk factor analysis and the exploration of the differential effects of covariates on an outcome, and exposes how OLS estimates underestimate and overestimate the effects of risk factors at different parts of the birth weight distribution.

Prevalence of Total and Domain-Specific Physical Activity and Associated Factors Among Nepalese Adults: A Quantile Regression Analysis

“…Furthermore, due to the lack of data, we were not able to examine the association with other important correlates of PA, such as psychosocial and environmental variables. 42…”