RESUMEN
Estimation and prediction of subnational mortality rates for small areas are essential planning tools for studying health inequalities. Standard methods do not perform well when data are noisy, a typical behavior of subnational datasets. Thus, reliable estimates are difficult to obtain. I present a Bayesian hierarchical model framework for prediction of mortality rates at a small or subnational level. By combining ideas from demography and epidemiology, the classical mortality modeling framework is extended to include an additional spatial component capturing regional heterogeneity. Information is pooled across neighboring regions and smoothed over time and age. To make predictions more robust and address the issue of model selection, a Bayesian version of stacking is considered using leave-future-out validation. I apply this method to forecast mortality rates for 96 regions in Bavaria, Germany, disaggregated by age and sex. Uncertainty surrounding the forecasts is provided in terms of prediction intervals. Using posterior predictive checks, I show that the models capture the essential features and are suitable to forecast the data at hand. On held-out data, my predictions outperform those of standard models lacking a regional component.
Asunto(s)
Teorema de Bayes , Humanos , Predicción , Alemania/epidemiologíaRESUMEN
BACKGROUND: Household survey data are frequently used to measure reproductive, maternal, newborn, child and adolescent health (RMNCAH) service utilisation in low and middle income countries. However, these surveys are typically only undertaken every 5 years and tend to be representative of larger geographical administrative units. Investments in district health management information systems (DHMIS) have increased the capability of countries to collect continuous information on the provision of RMNCAH services at health facilities. However, reliable and recent data on population distributions and demographics at subnational levels necessary to construct RMNCAH coverage indicators are often missing. One solution is to use spatially disaggregated gridded datasets containing modelled estimates of population counts. Here, we provide an overview of various approaches to the production of gridded demographic datasets and outline their potential and their limitations. Further, we show how gridded population estimates can be used as alternative denominators to produce RMNCAH coverage metrics in combination with data from DHMIS, using childhood vaccination as examples. METHODS: We constructed indicators on the percentage of children one year old for diphtheria, pertussis and tetanus vaccine dose 3 (DTP3) and measles vaccine dose (MCV1) in Zambia and Nigeria at district levels. For the numerators, information on vaccines doses was obtained from each country's respective DHMIS. For the denominators, the number of children was obtained from 3 different sources including national population projections and aggregated gridded estimates derived using top-down and bottom-up geospatial methods. RESULTS: In Zambia, vaccination estimates utilising the bottom-up approach to population estimation substantially reduced the number of districts with > 100% coverage of DTP3 and MCV1 compared to estimates using population projection and the top-down method. In Nigeria, results were mixed with bottom-up estimates having a higher number of districts > 100% and estimates using population projections performing better particularly in the South. CONCLUSIONS: Gridded demographic data utilising traditional and novel data sources obtained from remote sensing offer new potential in the absence of up to date census information in the estimation of RMNCAH indicators. However, the usefulness of gridded demographic data is dependent on several factors including the availability and detail of input data.
Asunto(s)
Servicios de Salud del Adolescente , Adolescente , Niño , Familia , Humanos , Renta , Lactante , Recién Nacido , Vacuna Antisarampión , VacunaciónRESUMEN
Reliable subnational mortality estimates are essential in the study of health inequalities within a country. One of the difficulties in producing such estimates is the presence of small populations among which the stochastic variation in death counts is relatively high, and thus the underlying mortality levels are unclear. We present a Bayesian hierarchical model to estimate mortality at the subnational level. The model builds on characteristic age patterns in mortality curves, which are constructed using principal components from a set of reference mortality curves. Information on mortality rates are pooled across geographic space and are smoothed over time. Testing of the model shows reasonable estimates and uncertainty levels when it is applied both to simulated data that mimic U.S. counties and to real data for French départements. The model estimates have direct applications to the study of subregional health patterns and disparities.