Survival Analysis on Age at Teenage Pregnancies Using Parametric Frailty Models
Selamawit Endale Gurmu *
Department of Statistics, College of Natural and Computational Science, Assosa University, Assosa, Ethiopia.
Tashome Fenta Biru
Department of Statistics, College of Natural and Computational Science, Assosa University, Assosa, Ethiopia.
*Author to whom correspondence should be addressed.
Abstract
Teenage pregnancies are common public health problem in the world. It is one of the main issues concerning reproductive health of teenagers. Although prevention of unwanted teenage pregnancy is our main goal, many females continue to become pregnant at early age. The current article goal is identifying the risk factors affecting survival time of teenage pregnancies. The information of this study was obtained from well-prepared questionnaire and focus group discussion. Female between age intervals 15 to 19 was used for assessing age at teenage pregnancies. Semi parametric model (Cox proportional hazard model) and parametric models (parametric shared frailty) were used to age at teenage pregnancies. The study subjects in this article came from clustered community. Parametric shared frailty models were explored by assuming that women with in the same residence shares similar risk factors. Weibull, Log logistics and Log normal distributions were analyzed for teenagers’ data set. All models were compared for their performance based on Akaike information criterion accordingly the log logistic inverse Gaussian shared frailty model was the best model for this data set since it has the minimum Akaike information criterion. This article show that marital status, age at marriage, teenager’s education level, Teenager’s occupation, mass media, family planning and Religion were significant risk factors for age at teenage pregnancies.
Keywords: Teenage pregnancies, accelerated failure time, parametric frailty, survival data, heterogeneity.