Using Mobile Phone Text Messaging to Reduce Maternal and Infant Deaths in Remote Rural Areas in China

1. Sampling method:

Random sampling method will be used. Counties in Huaihua area of Chinese province of Hunan will be randomly allocated to intervention and non-intervention groups. 2. The intervention:

Mobile phone text messages containing maternal and newborn health care education material will be sent to mothers-to-be at the first, second and third trimester, and again postpartum period. 3. Statistical analysis:

We will first compare the baseline characteristics (including the average income, adult educational level, and maternal and perinatal mortality rates in the past two years) between the 2 arms of experiment. For costs data, total costs, costs per participating woman, and costs per maternal/infant death avoided will be reported. For primary outcomes, we will determine and report rate ratios, risk differences, and numbers needed to treat (NNT) from the experimental arm (i.e., mobile phone text messaging), using the usual care arm (no text messaging) as the reference. Relative risks and 95% confidence intervals will be expressed as the effect measures. We will estimate the adjusted relative risks and 95% confidence intervals by regression analysis. Mixed models taking consideration of multi-level measures (e.g., mortality and morbidity measured at individual level and intervention measured at village level) will be used in the regression analysis (Donner and Klar, 2000).