University of Washington School of Public Health
Winter Course 2018: BIOST 555/EPI 555/GH 534 - Statistical Methods for Spatial Epidemiology
Winter 2018 Course Offering: BIOST 555/EPI 555/GH 534 Statistical Methods for Spatial Epidemiology
Quarter: Winter 2018
Time and Location: T Th 9:00-10:20am, Health Sciences T739
Grading: Graded, 3 credits
Instructor: Jon Wakefield, Professor of Statistics and Biostatistics (email@example.com)
Prerequisites: Previous exposure to regression modeling, some familiarity with log-linear or logistic modeling is desirable.
This course motivates the need for, and describes methods for, the analysis of spatially indexed epidemiological data. Major topics to be covered include disease mapping, clustering and cluster detection, spatial regression, methods for infectious disease data, small area estimation and an introduction to geographical information systems.
Learning Objectives: At the end of the course the student will be able to describe the need for specialized methods for the analysis of spatial data, distinguish between different types of spatial data, and choose an appropriate analysis method.
- Disease mapping area-level cancer incidence and mortality data
- Cluster investigation close to a pollution source
- Small area estimation in a developing world context
- Analysis of space-time measles data
- Geostatistical models for exposure mapping
Both point-references and spatially aggregated data will be considered. The use of R packages for analysis will be described.
The availability of geographically indexed health, population and exposure data, and advances in computing, geographic information systems, and statistical methodology, have enabled the realistic investigation of spatial variation in disease risk. Each of the population, exposure and health data may have associated exact spatial and temporal information (point data), or be available as aggregated summaries (count data). The following specific topics, with analysis methods listed for each, will be covered:
- Geostatistical smoothing models for point-level data, including prevalence data
- Kernel density estimation
- Models for area-level (aggregate) data
- Extensions to space-time modeling
Clustering and cluster detection:
- Autocorrelation statistics, including Moran’s I
- Scan statistics including SatScan
- Problems with conventional analyses
- Methods for acknowledgement of residual spatial dependence
- The ecological fallacy
Methods for infectious disease data:
- Discrete-time Susceptible-Infectious-Recovered (SIR) and related models
Small area estimation (domain estimation):
- Bayesian model-based approaches
- Incorporation of design weights
- Background to GIS
- How to visualize spatial data in R
Evaluation: Via biweekly homework's (40%) and a course project (60%). The final version of the latter is to be completed by the end of the quarter, with an outline to be handed in approximately halfway through the quarter.
Elliott, P., Wakefield, J., Best, N. and Briggs, D. (2000). Spatial Epidemiology: Methods and Applications, Oxford University Press.
Waller, L.A. and Gotway, C.A. (2004). Applied Spatial Statistics for Public Health Data, Wiley, New York.