A major public health aim is to provide accurate estimates and predictions of disease burden at the area level;this endeavor is vital for prevention strategies. This proposal describes methods for modeling spatially and temporally indexed health data. The first two aims concern infectious disease data and the third aim small-area estimation (SAE) based on complex survey data. Despite the widespread collection of infectious disease data in time and space there are important gaps in statistical methodology for the analysis of such data. Specifically, there are both a limited number of modeling options and a lack of software implementations for those that are available. Hence, the emphasis in this project is on practically applicable methods that will be made available within freely-available software, based on modern Bayesian smoothing models. With respect to infectious disease data, a flexible spline- based model is proposed. The methods will be developed in the context of a number of developing world infectious disease applications including hand, foot and mouth disease and tuberculosis. For SAE the proposed approach combines design-based estimation techniques with spatial smoothing priors, to produce estimates with both low bias and low variance. These models will be applied to data from the Behavioral Risk Factor Surveillance System (BRFSS) and to infant mortality and HIV data from the Tanzania demographic and health survey (DHS), in order to answer important public health questions.
The modeling of area-level disease counts in space and time has a number of important public health uses. In the developing world, vital registration systems often do not have full coverage, and estimates of infant deaths or HIV prevalence by area must be based on a variety of data sources;in general, the prediction of areas in which disease burden is likely to be high is an important public health endeavor since it allows appropriate action to be taken, such as the concentration of resources in those areas. The methods to be developed in this proposal focus on the modeling of infectious disease counts and on small area estimation, with smoothing over space and time being a key ingredient in both ventures, and with an emphasis on implementable methods.
|Wakefield, Jonathan; Kim, Albert (2013) A Bayesian model for cluster detection. Biostatistics 14:752-65|
|Ross, Michelle; Wakefield, Jon (2013) Bayesian inference for two-phase studies with categorical covariates. Biometrics 69:469-77|
|Wakefield, Jon; Haneuse, Sebastien; Dobra, Adrian et al. (2011) Bayes computation for ecological inference. Stat Med 30:1381-96|
|Wang, Yu; Feng, Zijian; Yang, Yang et al. (2011) Hand, foot, and mouth disease in China: patterns of spread and transmissibility. Epidemiology 22:781-92|
|Fong, Youyi; Rue, Havard; Wakefield, Jon (2010) Bayesian inference for generalized linear mixed models. Biostatistics 11:397-412|
|Glynn, Adam; Wakefield, Jon (2010) Ecological Inference in the Social Sciences. Stat Methodol 7:307-322|
|Ross, Michelle E; Wakefield, Jon; Davis, Scott et al. (2010) Spatial clustering of myelodysplastic syndromes (MDS) in the Seattle-Puget Sound region of Washington State. Cancer Causes Control 21:829-38|
|Glynn, Adam; Wakefield, Jon; Handcock, Mark S et al. (2008) Alleviating Linear Ecological Bias and Optimal Design with Sub-sample Data. J R Stat Soc Ser A Stat Soc 171:179-202|
|Haneuse, Sebastien J-P A; Wakefield, Jonathan C (2008) The Combination of Ecological and Case-Control Data. J R Stat Soc Series B Stat Methodol 70:73-93|
|Wakefield, Jon; Haneuse, Sebastien J-P A (2008) Overcoming ecologic bias using the two-phase study design. Am J Epidemiol 167:908-16|
Showing the most recent 10 out of 16 publications