Claudia Mastrogiacomo , Institute for Health Metrics and Evaluation
Haidong Wang, University of Washington, Seattle
Demographic estimates are central to epidemiological, sociological, and economic analysis. Fertility rates are important drivers of population growth and can be key indicators of social and economic development. Thus, it is vital that we have accurate estimates of age-specific fertility. We present a comparison of fertility methods used and estimates produced as part of the 2019 Global Burden of Disease (GBD) study with those published in the United Nation’s 2019 World Population Prospects (WPP). For the GBD study, we produce single calendar year estimates of age-specific fertility for five-year age groups using a two-stage spatiotemporal Gaussian process regression model. This model incorporates splitting of non-age-specific data, adjustment of biased data, and smoothing over space and time to produce estimates in 204 countries and territories. The WPP estimates are generated through a decentralized process using country-level data, which is selected by analysts and adjusted when needed in order to ensure consistency with population data. We show that the GBD estimates differ mainly from the WPP estimates in that they are able to more closely follow abrupt temporal changes in fertility in locations with reliable vital registration data and follow adjusted versions of data based on estimated biases rather than raw data in locations with sparser data and varied data sources.