Child mortality, and, in particular under-five mortality (U5MR), is an important indicator of the overall health of a population. Subnational estimation of U5MR is relatively new endeavor
and is of great interest as it allows countries to illuminate areas in need of intervention. In low- and middle-income countries, vital registration is often incomplete, so information on U5MR comes from household surveys or censuses. Data either come in the form of full birth histories (FBH) in which mothers report the birth and death dates of each of their children or summary birth histories (SBH) in which mothers report the number of children they have had and the number that have died. Previous work has performed Bayesian space-time smoothing of direct (survey-weighted) estimates of FBH data in small areas. We extend this method to make estimates at a finer small area level. Constructing naive estimates at a finer geography leads to unstable survey estimates due to low sample sizes. We propose an adjustment for areas where we have small samples sizes and observe no deaths. The method accounts for the complex survey design and is computationally efficient. Additionally, we extend previous work to incorporate data measured under different time schedules by modeling at a yearly time schedule and then aggregating to the required level. We also discuss ongoing work on estimating associations of known determinants of child and infant mortality across countries, examining the effect of adjusting for residual spatial dependence. Of particular interest is whether the associations between known determinants, such as being part of a multiple birth, mother's education or malaria, vary by country. Finally, we briefly outline future work for subnational estimation of child mortality for ages 5-15 and 15-19.