Parametric Models for First Birth Interval in North-East India: Identifying A Suitable Model

Abstract

Background: Birth intervals are often modelled to understand the health implications of the mother as well as the newborn. Shorter birth intervals are linked with higher risks of maternal and infant mortality. Short birth intervals in North-East India are linked to higher maternal and infant mortality risks, necessitating accurate model for targeted intervention. On the other hand, a longer birth interval has shown substantial reduction in the risk of maternal health issues and a better health outcome of the babies.

Methods: Time-to-event data are often modelled by implementing the popular Cox proportional hazards model. However, the popularity of the Cox model can’t overrule the use of parametric models if the distribution of the survival time has a known parametric form that is derived from past experience in previous research studies. Choosing an appropriate model from amongst various competing models is a topic of interest where different characteristics, such as the nature of censoring and the shape of hazards, are present in different dimensions for different events. We use information criteria and model fit measures to help select the best-fitted model among the five competing models for the data. Further graphical comparisons are made to conclude for a final which best fit the first birth interval.

Conclusion: The three-parameter generalized gamma model shows one of the most appropriate models for modelling first birth interval after marriage data with low proportion of censored data with a mix of hazards. Statistical tests such as the Anderson-Darling and Kolmogorov-Smirnov tests are significantly affected by the presence of extreme values of the time variable at the later observed times. The generalized gamma model can inform policies to extend first birth intervals, reducing risks of adverse maternal and child health outcomes in such datasets which are typical of demographic surveys.