A FLEXIBLE BAYESIAN FRAMEWORK TO ESTIMATE AGE- AND CAUSE-SPECIFIC CHILD MORTALITY OVER TIME FROM SAMPLE REGISTRATION DATA

Abstract

In order to implement disease-specific interventions in young age groups, policy makers in low- and middle-income countries require timely and accurate estimates of age- and cause-specific child mortality. High-quality data is not available in settings where these interventions are most needed, but there is a push to create sample registration systems that collect detailed mortality information. current methods that estimate mortality from this data employ multistage frameworks without rigorous statistical justification that separately estimate all-cause and cause-specific mortality and are not sufficiently adaptable to capture important features of the data. We propose a flexible Bayesian modeling framework to estimate age- and cause-specific child mortality from sample registration data. We provide a theoretical justification for the framework, explore its properties via simulation, and use it to estimate mortality trends using data from the Maternal and Child Health Surveillance System in China.

Keywords: Bayesian inference; cause-specific mortality; child mortality; sample registration system.

Fig. 1.

Empirical cause-specific mortality fractions over time by region and age group that were observed in the Maternal and Child Health Surveillance System in China.

Fig. 2.

Relative bias, coverage, and width of 95% intervals for log-mortality rate estimates from multistage and unified models. For (a), data were generated with IID Normal random effects for each observation and three possible exposure values. For (b), data were generated with bivariate IID Normal random effects for each region-age-year strata, defining the diagonal elements of the covariance matrix as σ² and the off diagonal elements as ρσ². Estimates are averaged over all observations and simulations per scenario.

Fig. 3.

Selected results from the MCHSS data analysis showing the observations, estimated posterior medians, and posterior 80% intervals for log-mortality rates. Combinations with no deaths are represented by an open square.

Fig. 4.

Comparisons of estimated CSMFs between our model and the model in He et al. (2017) in the east rural region. Panels A1 and B1 show cause fractions from He et al. (2017) for the zero to one month and one to 59 month age groups, respectively, while panels A2 and B2 show the estimates from our model for the finer age groups, as labeled, that are within the broad age groups in A1 and B1.