Determinants of Birth Intervals in Tamil Nadu in India: Developing Cox Hazard Models with Validations and Predictions

Abstract

The present study uses data from National Family Health Survey (NFHS-1) 1992-93 (International Institute for Population Sciences 1995) conducted in the state of Tamil Nadu, India. Cox models were developed to analyze the effect of breastfeeding as time varying and time dependent factor on birth intervals. Breastfeeding alone improved the log likelihood up to a higher level in each birth interval. Other factors that entered into the models were: at first birth interval, womens education (high school & above) and working status of women; at second birth interval, survival status of index child alive and husbands education (high school & above), and at third birth interval, breastfeeding more than 22 month were found to be protective factors for next births. Validation of the developed models was done through bootstrapping to predict birth intervals.

Keywords

• Cox model
• Multivariate analysis
• Validation
• Predictions