Developing measurement model using Bayesian confirmatory factor analysis in suppressing maternal mortality

The greatest challenge for Indonesia at this moment based on Millennium Development Goals (MDGs) is to reduce maternal mortality or AKI (angka kematian ibu). The number had been decreasing so far from 390 deaths per 100000 births in 1994 to 307 in 2002 - 2003. However, due to complications there were 20000 women died every year. By the end of 2015, Indonesia Bureau of Statistics has projected that there will be 163 deaths per 100000 births, while the target is 102. According to that matter, the government already established safe motherhood as four pillars: planned-family service, antenatal service, safe and clean birth-delivery, and essential obstetric service. This study was conducted to measure variables that expected to affect the number of AKI. The four pillars were viewed as latent variables that expected to affect AKI. AKI was represented by safe motherhood index, also a latent variable. The proposed method was by constructing model using confirmatory factor analysis with Bayesian approach. This approach was due to violation of multivariate normal assumption on the indicators. Variables made up of loading factor values more than 0.5 were planned-family service, antenatal service (0.743), safe and clean birth-delivery (0.810), and essential obstetric service (0.721).