Markov Analysis of Contraceptive Use Behavior: Implications for the Family Planning Program in India

Deepali Godha, Independent Conultant

With continued family planning program effort, a continuous re-evaluation is needed on “where” to improve further. This study utilizes contraceptive calendar data from India DHS 2005-06. A semi-Markov modeling technique, using TreeAge software 2012, has been employed to simulate the dynamics of contraceptive use behavior among rural poor women. Monte Carlo simulations and sensitivity analysis have been conducted to offer programmatic insight. Monthly transition probabilities have been obtained by survival analysis techniques using Stata 10. The study model provides comprehensive estimates of contraceptive behavior in study population and measure of subsequent pregnancies/ births/ terminations, differentiating the latter based on previous state of contraceptive behavior. The study demonstrates variation in relative importance and magnitude of impact of various aspects of contraceptive behavior on mean number of pregnancies in an atmosphere of constantly changing contraceptive behavior mix. The predictive model further gives insight into future contraceptive behavior mix and resulting demographic effects.