Statistical Methods in Spatial Epidemiology

by Lawson, Andrew B.

• ISBN13: 9780470014844
• ISBN10: 0470014849
• Edition: 2nd
• Format: Hardcover
• Copyright: 2006-06-16
• Publisher: WILEY

Summary

Spatial epidemiology is the description and analysis of the geographical distribution of disease. It is more important now than ever, with modern threats such as bio-terrorism making such analysis even more complex. This second edition of Statistical Methods in Spatial Epidemiology is updated and expanded to offer a complete coverage of the analysis and application of spatial statistical methods. The book is divided into two main sections: Part 1 introduces basic definitions and terminology, along with map construction and some basic models. This is expanded upon in Part II by applying this knowledge to the fundamental problems within spatial epidemiology, such as disease mapping, ecological analysis, disease clustering, bio-terrorism, space-time analysis, surveillance and infectious disease modelling. Provides a comprehensive overview of the main statistical methods used in spatial epidemiology. Updated to include a new emphasis on bio-terrorism and disease surveillance. Emphasizes the importance of space-time modelling and outlines the practical application of the method. Discusses the wide range of software available for analyzing spatial data, including WinBUGS, SaTScan and R, and features an accompanying website hosting related software. Contains numerous data sets, each representing a different approach to the analysis, and provides an insight into various modelling techniques.This text is primarily aimed at medical statisticians, researchers and practitioners from public health and epidemiology. It is also suitable for postgraduate students of statistics and epidemiology, as well professionals working in government agencies.

Author Biography

Professor Andrew B. Lawson is a respected and well-known academic. He has published many papers in leading journals, and a number of books on spatial statistics, including five for Wiley.

Table of Contents

Preface and Acknowledgements to Second Edition	xv
Preface and Acknowledgements	xvii
I The Nature of Spatial Epidemiology	1	(108)
1 Definitions, Terminology and Data Sets	3	(22)
1.1 Map Hypotheses and Modelling Approaches	5	(2)
1.2 Definitions and Data Examples	7	(3)
1.2.1 Case event data	7	(1)
1.2.2 Count data	8	(2)
1.3 Further Definitions	10	(1)
1.3.1 Control events and processes	10	(1)
1.3.2 Census tract information	10	(1)
1.3.3 Clustering definitions	10	(1)
1.4 Some Data Examples	11	(14)
1.4.1 Case event examples	11	(8)
1.4.2 Count data examples	19	(6)
2 Scales of Measurement and Data Availability	25	(6)
2.1 Small Scale	26	(1)
2.2 Large Scale	26	(1)
2.3 Rate Dependence	27	(1)
2.4 Data Quality and the Ecological Fallacy	27	(1)
2.5 Edge Effects	28	(3)
3 Geographical Representation and Mapping	31	(10)
3.1 Introduction and Definitions	31	(1)
3.2 Maps and Mapping	32	(5)
3.2.1 Statistical maps and mapping	34	(1)
3.2.2 Object process mapping	34	(2)
3.2.3 Geostatistical mapping	36	(1)
3.3 Statistical Accuracy	37	(1)
3.4 Aggregation	37	(1)
3.5 Mapping Issues Related to Aggregated Data	37	(2)
3.6 Conclusions	39	(2)
4 Basic Models	41	(26)
4.1 Sampling Considerations	41	(1)
4.2 Likelihood-Based and Bayesian Approaches	42	(1)
4.3 Point Event Models	42	(16)
4.3.1 Point process models and applications	43	(1)
4.3.2 The basic Poisson process model	44	(5)
4.3.3 Hybrid models and regionalisation	49	(1)
4.3.4 Bayesian models and random effects	50	(2)
4.3.5 MAP estimation, empirical Bayes and full Bayesian analysis	52	(1)
4.3.6 Bivariate/multivariate models	53	(3)
4.3.7 Hidden structure and mixture models	56	(1)
4.3.8 Space-time extensions	56	(2)
4.4 Count Models	58	(9)
4.4.1 Standard models	60	(3)
4.4.2 Approximations	63	(1)
4.4.3 Random-effect extensions	63	(1)
4.4.4 Hidden structure and mixture models	64	(1)
4.4.5 Space-time extensions	65	(2)
5 Exploratory Approaches, Parametric Estimation and Inference	67	(42)
5.1 Exploratory Methods	68	(12)
5.1.1 Cartographic issues	69	(2)
5.1.2 Case event mapping	71	(4)
5.1.3 Count mapping	75	(5)
5.2 Parameter Estimation	80	(16)
5.2.1 Case event likelihood models	80	(5)
5.2.2 Count event likelihood models	85	(2)
5.2.3 Approximations	87	(1)
5.2.4 Bayesian models	88	(8)
5.3 Residual Diagnostics	96	(2)
5.4 Hypothesis Testing	98	(1)
5.5 Edge Effects	99	(12)
5.5.1 Edge effects in case events	101	(1)
5.5.2 Edge effects in counts	101	(1)
5.5.3 Edge weighting schemes and MCMC methods	102	(2)
5.5.4 Discussion	104	(1)
5.5.5 The Tuscany example	105	(4)
II Important Problems in Spatial Epidemiology	109	(204)
6 Small Scale: Disease Clustering	111	(32)
6.1 Definition of Clusters and Clustering	112	(3)
6.2 Modelling Issues	115	(3)
6.3 Hypothesis Tests for Clustering	118	(5)
6.3.1 General non-specific clustering	118	(3)
6.3.2 Specific clustering	121	(2)
6.4 Space-Time Clustering	123	(4)
6.4.1 Modelling issues	123	(3)
6.4.2 Hypothesis testing	126	(1)
6.5 Clustering Examples	127	(11)
6.5.1 Humberside example	127	(4)
6.5.2 Larynx cancer example	131	(2)
6.5.3 Count data clustering example	133	(3)
6.5.4 Space-time clustering examples	136	(2)
6.6 Other Methods Related to Clustering	138	(5)
6.6.1 Wombling	140	(3)
7 Small Scale: Putative Sources of Hazard	143	(46)
7.1 Introduction	143	(1)
7.2 Study Design	144	(3)
7.2.1 Retrospective and prospective studies	144	(1)
7.2.2 Study region design	145	(1)
7.2.3 Replication and controls	146	(1)
7.3 Problems of Inference	147	(6)
7.3.1 Exploratory techniques	148	(5)
7.4 Modelling the Hazard Exposure Risk	153	(9)
7.5 Models for Case Event Data	162	(5)
7.5.1 Estimation	164	(1)
7.5.2 Hypothesis tests	164	(2)
7.5.3 Diagnostic techniques	166	(1)
7.6 A Case Event Example	167	(2)
7.7 Models for Count Data	169	(3)
7.7.1 Estimation	171	(1)
7.7.2 Hypothesis tests	171	(1)
7.8 A Count Data Example	172	(2)
7.9 Other Directions	174	(15)
7.9.1 Multiple disease analysis	174	(10)
7.9.2 Space-time modelling	184	(1)
7.9.3 Space-time exploratory analysis	184	(1)
7.9.4 Space-time Bayesian analysis	185	(4)
8 Large Scale: Disease Mapping	189	(58)
8.1 Introduction	189	(1)
8.2 Simple Statistical Representation	189	(5)
8.2.1 Crude rates	190	(1)
8.2.2 Standardised mortality/morbidity ratios, standardisation and relative risk surfaces	191	(2)
8.2.3 Interpolation	193	(1)
8.2.4 Exploratory mapping methods	193	(1)
8.3 Basic Models	194	(7)
8.3.1 Likelihood models	194	(3)
8.3.2 Random effects and Bayesian models	197	(4)
8.4 Advanced Methods	201	(8)
8.4.1 Non-parametric methods	202	(1)
8.4.2 Incorporating spatially correlated heterogeneity	203	(3)
8.4.3 Case event modelling	206	(3)
8.5 Model Variants and Extensions	209	(3)
8.5.1 Semiparametric modelling	209	(1)
8.5.2 Geographically weighted regression	210	(1)
8.5.3 Mixture models	211	(1)
8.6 Approximate Methods	212	(1)
8.7 Multivariate Methods	213	(3)
8.8 Evaluation of Model Performance	216	(3)
8.9 Hypothesis Testing in Disease Mapping	219	(3)
8.9.1 First-order effects	219	(2)
8.9.2 Second-order and variance effects	221	(1)
8.10 Space-Time Disease Mapping	222	(7)
8.11 Spatial Survival and Longitudinal Data	229	(3)
8.11.1 Spatial survival analysis	229	(2)
8.11.2 Spatial longitudinal analysis	231	(1)
8.11.3 Spatial multiple event modelling	232	(1)
8.12 Disease Mapping: Case Studies	232	(15)
8.12.1 Eastern Germany	232	(7)
8.12.2 Ohio respiratory cancer	239	(8)
9 Ecological Analysis and Scale Change	247	(22)
9.1 Ecological Analysis: Introduction	247	(5)
9.2 Small-Scale Modelling Issues	252	(3)
9.2.1 Hypothesis tests	253	(1)
9.2.2 Ecological aggregation effects	253	(2)
9.3 Changes of Scale and MAUP	255	(6)
9.3.1 MAUP: the modifiable areal unit problem	255	(5)
9.3.2 Large-scale issues	260	(1)
9.4 A Simple Example: Sudden Infant Death in North Carolina	261	(2)
9.5 A Case Study: Malaria and IDDM	263	(6)
10 Infectious Disease Modelling	269	(24)
10.1 Introduction	269	(1)
10.2 General Model Development	270	(3)
10.3 Spatial Model Development	273	(7)
10.3.1 Count data	273	(5)
10.3.2 Individual-level data	278	(2)
10.4 Modelling Special Cases for Individual-Level Data	280	(3)
10.4.1 Proportional hazards interpretation	280	(1)
10.4.2 Subgroup modifications	281	(1)
10.4.3 Cluster function specification	282	(1)
10.5 Survival Analysis with Spatial Dependence	283	(1)
10.6 Individual-Level Data Example	284	(4)
10.6.1 Distribution of susceptibles S(x,t)	285	(1)
10.6.2 The spatial distance function h	285	(1)
10.6.3 The function g	285	(1)
10.6.4 Fitting the model	286	(1)
10.6.5 Revised model	287	(1)
10.7 Underascertainment and Censoring	288	(1)
10.8 Conclusions	289	(4)
11 Large Scale: Surveillance	293	(20)
11.1 Process Control Methodology	294	(1)
11.2 Spatio-Temporal Modelling	295	(2)
11.3 S-T Monitoring	297	(7)
11.3.1 Fixed spatial and temporal frame	297	(4)
11.3.2 Fixed spatial frame and dynamic temporal frame	301	(3)
11.4 Syndromic Surveillance	304	(1)
11.5 Multivariate–Multifocus Surveillance	305	(3)
11.6 Bayesian Approaches	308	(2)
11.6.1 Bayesian alarm functions, Bayes factors and syndromic analyses	308	(2)
11.7 Computational Considerations	310	(1)
11.8 Infectious Diseases	311	(1)
11.9 Conclusions	312	(1)
Appendix A Monte Carlo Testing, Parametric Bootstrap and Simulation Envelopes	313	(12)
A.1 Nuisance Parameters and Test Statistics	313	(1)
A.2 Monte Carlo Tests	314	(1)
A.3 Null Hypothesis Simulation	315	(4)
A.3.1 Spatial case	316	(2)
A.3.2 Spatio-temporal case	318	(1)
A.4 Parametric Bootstrap	319	(5)
A.4.1 Bayesian spatial models	322	(1)
A.4.2 Spatio-temporal case	323	(1)
A.5 Simulation Envelopes	324	(1)
Appendix B Markov Chain Monte Carlo Methods	325	(6)
B.1 Definitions	325	(1)
B.2 Metropolis and Metropolis–Hastings Algorithms	326	(5)
B.2.1 Metropolis algorithm	326	(1)
B.2.2 Metropolis–Hastings extension	327	(1)
B.2.3 The Gibbs sampler	327	(1)
B.2.4 M–H versus Gibbs algorithms	328	(1)
B.2.5 Examples	328	(3)
Appendix C Algorithms and Code	331	(28)
C.1 Data Exploration	331	(4)
C.2 Likelihood and Bayesian Models	335	(1)
C.3 Likelihood Models	336	(5)
C.3.1 Case event data	336	(4)
C.3.2 Count data	340	(1)
C.4 Bayesian Hierarchical Models	341	(5)
C.4.1 Case event data	341	(3)
C.4.2 Count data	344	(2)
C.5 Space-Time Analysis	346	(13)
C.5.1 Data exploration	346	(3)
C.5.2 Likelihood models	349	(2)
C.5.3 Bayesian models	351	(6)
C.5.4 Infectious disease models	357	(2)
Appendix D Glossary of Estimators	359	(4)
D.1 Case Event Estimators	359	(2)
D.2 Tract Count Estimators	361	(2)
Appendix E Software	363	(4)
E.1 Software	363	(4)
E.1.1 Spatial statistical tools	363	(2)
E.1.2 Geographical information systems	365	(2)
Bibliography	367	(22)
Index	389