| |
| |
Preface | |
| |
| |
Acknowledgments | |
| |
| |
| |
Introduction | |
| |
| |
| |
Why Spatial Data in Public Health? | |
| |
| |
| |
Why Statistical Methods for Spatial Data? | |
| |
| |
| |
Intersection of Three Fields of Study | |
| |
| |
| |
Organization of the Book | |
| |
| |
| |
Analyzing Public Health Data | |
| |
| |
| |
Observational vs. Experimental Data | |
| |
| |
| |
Risk and Rates | |
| |
| |
| |
Incidence and Prevalence | |
| |
| |
| |
Risk | |
| |
| |
| |
Estimating Risk: Rates and Proportions | |
| |
| |
| |
Relative and Attributable Risks | |
| |
| |
| |
Making Rates Comparable: Standardized Rates | |
| |
| |
| |
Direct Standardization | |
| |
| |
| |
Indirect Standardization | |
| |
| |
| |
Direct or Indirect? | |
| |
| |
| |
Standardizing to What Standard? | |
| |
| |
| |
Cautions with Standardized Rates | |
| |
| |
| |
Basic Epidemiological Study Designs | |
| |
| |
| |
Prospective Cohort Studies | |
| |
| |
| |
Retrospective Case-Control Studies | |
| |
| |
| |
Other Types of Epidemiological Studies | |
| |
| |
| |
Basic Analytic Tool: The Odds Ratio | |
| |
| |
| |
Modeling Counts and Rates | |
| |
| |
| |
Generalized Linear Models | |
| |
| |
| |
Logistic Regression | |
| |
| |
| |
Poisson Regression | |
| |
| |
| |
Challenges in the Analysis of Observational Data | |
| |
| |
| |
Bias | |
| |
| |
| |
Confounding | |
| |
| |
| |
Effect Modification | |
| |
| |
| |
Ecological Inference and the Ecological Fallacy | |
| |
| |
| |
Additional Topics and Further Reading | |
| |
| |
| |
Exercises | |
| |
| |
| |
Spatial Data | |
| |
| |
| |
Components of Spatial Data | |
| |
| |
| |
An Odyssey into Geodesy | |
| |
| |
| |
Measuring Location: Geographical Coordinates | |
| |
| |
| |
Flattening the Globe: Map Projections and Coordinate Systems | |
| |
| |
| |
Mathematics of Location: Vector and Polygon Geometry | |
| |
| |
| |
Sources of Spatial Data | |
| |
| |
| |
Health Data | |
| |
| |
| |
Census-Related Data | |
| |
| |
| |
Geocoding | |
| |
| |
| |
Digital Cartographic Data | |
| |
| |
| |
Environmental and Natural Resource Data | |
| |
| |
| |
Remotely Sensed Data | |
| |
| |
| |
Digitizing | |
| |
| |
| |
Collect Your Own! | |
| |
| |
| |
Geographic Information Systems | |
| |
| |
| |
Vector and Raster GISs | |
| |
| |
| |
Basic GIS Operations | |
| |
| |
| |
Spatial Analysis within GIS | |
| |
| |
| |
Problems with Spatial Data and GIS | |
| |
| |
| |
Inaccurate and Incomplete Databases | |
| |
| |
| |
Confidentiality | |
| |
| |
| |
Use of ZIP Codes | |
| |
| |
| |
Geocoding Issues | |
| |
| |
| |
Location Uncertainty | |
| |
| |
| |
Visualizing Spatial Data | |
| |
| |
| |
Cartography: The Art and Science of Mapmaking | |
| |
| |
| |
Types of Statistical Maps | |
| |
| |
Map Study: Very Low Birth Weights in Georgia Health Care District 9 | |
| |
| |
| |
Maps for Point Features | |
| |
| |
| |
Maps for Areal Features | |
| |
| |
| |
Symbolization | |
| |
| |
| |
Map Generalization | |
| |
| |
| |
Visual Variables | |
| |
| |
| |
Color | |
| |
| |
| |
Mapping Smoothed Rates and Probabilities | |
| |
| |
| |
Locally Weighted Averages | |
| |
| |
| |
Nonparametric Regression | |
| |
| |
| |
Empirical Bayes Smoothing | |
| |
| |
| |
Probability Mapping | |
| |
| |
| |
Practical Notes and Recommendations | |
| |
| |
Case Study: Smoothing New York Leukemia Data | |
| |
| |
| |
Modifiable Areal Unit Problem | |
| |
| |
| |
Additional Topics and Further Reading | |
| |
| |
| |
Visualization | |
| |
| |
| |
Additional Types of Maps | |
| |
| |
| |
Exploratory Spatial Data Analysis | |
| |
| |
| |
Other Smoothing Approaches | |
| |
| |
| |
Edge Effects | |
| |
| |
| |
Exercises | |
| |
| |
| |
Analysis of Spatial Point Patterns | |
| |
| |
| |
Types of Patterns | |
| |
| |
| |
Spatial Point Processes | |
| |
| |
| |
Stationarity and Isotropy | |
| |
| |
| |
Spatial Poisson Processes and CSR | |
| |
| |
| |
Hypothesis Tests of CSR via Monte Carlo Methods | |
| |
| |
| |
Heterogeneous Poisson Processes | |
| |
| |
| |
Estimating Intensity Functions | |
| |
| |
Data Break: Early Medieval Grave Sites | |
| |
| |
| |
K Function | |
| |
| |
| |
Estimating the K Function | |
| |
| |
| |
Diagnostic Plots Based on the K Function | |
| |
| |
| |
Monte Carlo Assessments of CSR Based on the K Function | |
| |
| |
Data Break: Early Medieval Grave Sites | |
| |
| |
| |
Roles of First- and Second-Order Properties | |
| |
| |
| |
Other Spatial Point Processes | |
| |
| |
| |
Poisson Cluster Processes | |
| |
| |
| |
Contagion/Inhibition Processes | |
| |
| |
| |
Cox Processes | |
| |
| |
| |
Distinguishing Processes | |
| |
| |
| |
Additional Topics and Further Reading | |
| |
| |
| |
Exercises | |
| |
| |
| |
Spatial Clusters of Health Events: Point Data for Cases and Controls | |
| |
| |
| |
What Do We Have? Data Types and Related Issues | |
| |
| |
| |
What Do We Want? Null and Alternative Hypotheses | |
| |
| |
| |
Categorization of Methods | |
| |
| |
| |
Comparing Point Process Summaries | |
| |
| |
| |
Goals | |
| |
| |
| |
Assumptions and Typical Output | |
| |
| |
| |
Method: Ratio of Kernel Intensity Estimates | |
| |
| |
Data Break: Early Medieval Grave Sites | |
| |
| |
| |
Method: Difference between K Functions | |
| |
| |
Data Break: Early Medieval Grave Sites | |
| |
| |
| |
Scanning Local Rates | |
| |
| |
| |
Goals | |
| |
| |
| |
Assumptions and Typical Output | |
| |
| |
| |
Method: Geographical Analysis Machine | |
| |
| |
| |
Method: Overlapping Local Case Proportions | |
| |
| |
Data Break: Early Medieval Grave Sites | |
| |
| |
| |
Method: Spatial Scan Statistics | |
| |
| |
Data Break: Early Medieval Grave Sites | |
| |
| |
| |
Nearest-Neighbor Statistics | |
| |
| |
| |
Goals | |
| |
| |
| |
Assumptions and Typical Output | |
| |
| |
| |
Method: q Nearest Neighbors of Cases | |
| |
| |
Case Study: San Diego Asthma | |
| |
| |
| |
Further Reading | |
| |
| |
| |
Exercises | |
| |
| |
| |
Spatial Clustering of Health Events: Regional Count Data | |
| |
| |
| |
What Do We Have and What Do We Want? | |
| |
| |
| |
Data Structure | |
| |
| |
| |
Null Hypotheses | |
| |
| |
| |
Alternative Hypotheses | |
| |
| |
| |
Categorization of Methods | |
| |
| |
| |
Scanning Local Rates | |
| |
| |
| |
Goals | |
| |
| |
| |
Assumptions | |
| |
| |
| |
Method: Overlapping Local Rates | |
| |
| |
Data Break: New York Leukemia Data | |
| |
| |
| |
Method: Turnbull et al.'s CEPP | |
| |
| |
| |
Method: Besag and Newell Approach | |
| |
| |
| |
Method: Spatial Scan Statistics | |
| |
| |
| |
Global Indexes of Spatial Autocorrelation | |
| |
| |
| |
Goals | |
| |
| |
| |
Assumptions and Typical Output | |
| |
| |
| |
Method: Moran's I | |
| |
| |
| |
Method: Geary's c | |
| |
| |
| |
Local Indicators of Spatial Association | |
| |
| |
| |
Goals | |
| |
| |
| |
Assumptions and Typical Output | |
| |
| |
| |
Method: Local Moran's I | |
| |
| |
| |
Goodness-of-Fit Statistics | |
| |
| |
| |
Goals | |
| |
| |
| |
Assumptions and Typical Output | |
| |
| |
| |
Method: Pearson's x[superscript 2] | |
| |
| |
| |
Method: Tango's Index | |
| |
| |
| |
Method: Focused Score Tests of Trend | |
| |
| |
| |
Statistical Power and Related Considerations | |
| |
| |
| |
Power Depends on the Alternative Hypothesis | |
| |
| |
| |
Power Depends on the Data Structure | |
| |
| |
| |
Theoretical Assessment of Power | |
| |
| |
| |
Monte Carlo Assessment of Power | |
| |
| |
| |
Benchmark Data and Conditional Power Assessments | |
| |
| |
| |
Additional Topics and Further Reading | |
| |
| |
| |
Related Research Regarding Indexes of Spatial Association | |
| |
| |
| |
Additional Approaches for Detecting Clusters and/or Clustering | |
| |
| |
| |
Space-Time Clustering and Disease Surveillance | |
| |
| |
| |
Exercises | |
| |
| |
| |
Spatial Exposure Data | |
| |
| |
| |
Random Fields and Stationarity | |
| |
| |
| |
Semivariograms | |
| |
| |
| |
Relationship to Covariance Function and Correlogram | |
| |
| |
| |
Parametric Isotropic Semivariogram Models | |
| |
| |
| |
Estimating the Semivariogram | |
| |
| |
Data Break: Smoky Mountain pH Data | |
| |
| |
| |
Fitting Semivariogram Models | |
| |
| |
| |
Anisotropic Semivariogram Modeling | |
| |
| |
| |
Interpolation and Spatial Prediction | |
| |
| |
| |
Inverse-Distance Interpolation | |
| |
| |
| |
Kriging | |
| |
| |
Case Study: Hazardous Waste Site Remediation | |
| |
| |
| |
Additional Topics and Further Reading | |
| |
| |
| |
Erratic Experimental Semivariograms | |
| |
| |
| |
Sampling Distribution of the Classical Semivariogram Estimator | |
| |
| |
| |
Nonparametric Semivariogram Models | |
| |
| |
| |
Kriging Non-Gaussian Data | |
| |
| |
| |
Geostatistical Simulation | |
| |
| |
| |
Use of Non-Euclidean Distances in Geostatistics | |
| |
| |
| |
Spatial Sampling and Network Design | |
| |
| |
| |
Exercises | |
| |
| |
| |
Linking Spatial Exposure Data to Health Events | |
| |
| |
| |
Linear Regression Models for Independent Data | |
| |
| |
| |
Estimation and Inference | |
| |
| |
| |
Interpretation and Use with Spatial Data | |
| |
| |
Data Break: Raccoon Rabies in Connecticut | |
| |
| |
| |
Linear Regression Models for Spatially Autocorrelated Data | |
| |
| |
| |
Estimation and Inference | |
| |
| |
| |
Interpretation and Use with Spatial Data | |
| |
| |
| |
Predicting New Observations: Universal Kriging | |
| |
| |
Data Break: New York Leukemia Data | |
| |
| |
| |
Spatial Autoregressive Models | |
| |
| |
| |
Simultaneous Autoregressive Models | |
| |
| |
| |
Conditional Autoregressive Models | |
| |
| |
| |
Concluding Remarks on Conditional Autoregressions | |
| |
| |
| |
Concluding Remarks on Spatial Autoregressions | |
| |
| |
| |
Generalized Linear Models | |
| |
| |
| |
Fixed Effects and the Marginal Specification | |
| |
| |
| |
Mixed Models and Conditional Specification | |
| |
| |
| |
Estimation in Spatial GLMs and GLMMs | |
| |
| |
Data Break: Modeling Lip Cancer Morbidity in Scotland | |
| |
| |
| |
Additional Considerations in Spatial GLMs | |
| |
| |
Case Study: Very Low Birth Weights in Georgia Health Care District 9 | |
| |
| |
| |
Bayesian Models for Disease Mapping | |
| |
| |
| |
Hierarchical Structure | |
| |
| |
| |
Estimation and Inference | |
| |
| |
| |
Interpretation and Use with Spatial Data | |
| |
| |
| |
Parting Thoughts | |
| |
| |
| |
Additional Topics and Further Reading | |
| |
| |
| |
General References | |
| |
| |
| |
Restricted Maximum Likelihood Estimation | |
| |
| |
| |
Residual Analysis with Spatially Correlated Error Terms | |
| |
| |
| |
Two-Parameter Autoregressive Models | |
| |
| |
| |
Non-Gaussian Spatial Autoregressive Models | |
| |
| |
| |
Classical/Bayesian GLMMs | |
| |
| |
| |
Prediction with GLMs | |
| |
| |
| |
Bayesian Hierarchical Models for Spatial Data | |
| |
| |
| |
Exercises | |
| |
| |
References | |
| |
| |
Author Index | |
| |
| |
Subject Index | |