A team based at the University of Washington has developed a model to help the United Nations identify local areas in sub-Saharan Africa and Southern Asia where additional resources may be needed to mitigate high under five mortality.
Around the world, under-five deaths are at the lowest point on record. However, falling national and global averages mask areas within a country where child mortality rates have remained steady or, in some cases, increased.
Jon Wakefield, Professor of Statistics and Biostatistics at UW, and Jessica Godwin, a PhD candidate in the UW Department of Statistics and a fellow of the Center for Studies in Demography & Ecology, led the work. Wakefield explains the model’s methodology:
“Small area estimation (SAE) is an important endeavor in global health, epidemiology, and demography. SAE is often based on data obtained from complex surveys, and one must acknowledge the survey design when statistical analysis is performed so that measures of uncertainty correctly incorporate sampling variability and avoid bias. Often data in particular areas are sparse (perhaps non-existent) and so spatial smoothing is advantageous to ‘borrow strength’ from neighboring areas, with in addition, smoothing over time. Informative graphical summaries are key for dissemination and for using the estimates to inform interventions and assess whether targets are being met.”
Students made major contributions to the model and computational methods. The team included many students from statistics and biostatistics, with representation from other departments as well as collaborators in a number of universities around the world.
“The students all took different countries and ran a variety of models. We then collectively examined results and gave countries the opportunity to comment, which sometimes led to changes in the analyses,” said Wakefield. Wakefield has been on UN missions to Quito (Ecuador), Johannesburg (South Africa) and Blantyre (Malawi), where he has taken part in workshops in which participants have learned how to implement the subnational estimation methods using software developed in the group, led by former UW Statistics PhD student Zehang Richard Li.
Students continue to work on methodological improvements and extensions in order to improve subnational estimation of child mortality and to increase the number of countries for which estimates are made. For example, work by current Statistics PhD students includes:
- Jessica Godwin — combining full and summary birth history data and demographic methods for child mortality modeling;
- Johnny Paige — adjusting for the complex survey design and spatial aggregation;
- Peter Gao — model-assisted small area estimation methods;
- Ziyu Jiang — interpretation of random effects intervals, and model validation.
“I am proud and feel lucky to work on statistical improvements in estimation of child mortality, one of the most relied-upon indicators of population health, with a multi-disciplinary team of collaborators and with the support of UN IGME who allow our word to be immediately impactful. I’m also grateful to have had the experience of watching this project grow from a chapter in one student’s dissertation into a vibrant research group under Jon Wakefield’s mentorship,” said Godwin.
Full details on the project are available here and updates are reported on Twitter by Wakefield (@jonwakeblue) and Godwin (@j_l_godwin).