Atnaujinkite slapukų nuostatas

Epidemiology: Study Design and Data Analysis, Third Edition 3rd edition [Kietas viršelis]

3.77/5 (14 ratings by Goodreads)
(University of Oxford, UK; University of Sydney, Australia; and Johns Hopkins University, Baltimore, Maryland, USA)
  • Formatas: Hardback, 898 pages, aukštis x plotis: 254x178 mm, weight: 2220 g, 233 Tables, black and white; 157 Illustrations, black and white
  • Serija: Chapman & Hall/CRC Texts in Statistical Science
  • Išleidimo metai: 19-Dec-2013
  • Leidėjas: Chapman & Hall/CRC
  • ISBN-10: 1439839700
  • ISBN-13: 9781439839706
Kitos knygos pagal šią temą:
Epidemiology: Study Design and Data Analysis, Third Edition 3rd editionDidesnis paveikslėlis
  • Formatas: Hardback, 898 pages, aukštis x plotis: 254x178 mm, weight: 2220 g, 233 Tables, black and white; 157 Illustrations, black and white
  • Serija: Chapman & Hall/CRC Texts in Statistical Science
  • Išleidimo metai: 19-Dec-2013
  • Leidėjas: Chapman & Hall/CRC
  • ISBN-10: 1439839700
  • ISBN-13: 9781439839706
Kitos knygos pagal šią temą:
Pastovi nuoroda: https://www.kriso.lt/db/9781439839706.html
Highly praised for its broad, practical coverage, the second edition of this popular text incorporated the major statistical models and issues relevant to epidemiological studies. Epidemiology: Study Design and Data Analysis, Third Edition continues to focus on the quantitative aspects of epidemiological research. Updated and expanded, this edition shows students how statistical principles and techniques can help solve epidemiological problems.

New to the Third Edition











New chapter on risk scores and clinical decision rules New chapter on computer-intensive methods, including the bootstrap, permutation tests, and missing value imputation New sections on binomial regression models, competing risk, information criteria, propensity scoring, and splines Many more exercises and examples using both Stata and SAS More than 60 new figures

After introducing study design and reviewing all the standard methods, this self-contained book takes students through analytical methods for both general and specific epidemiological study designs, including cohort, case-control, and intervention studies. In addition to classical methods, it now covers modern methods that exploit the enormous power of contemporary computers. The book also addresses the problem of determining the appropriate size for a study, discusses statistical modeling in epidemiology, covers methods for comparing and summarizing the evidence from several studies, and explains how to use statistical models in risk forecasting and assessing new biomarkers. The author illustrates the techniques with numerous real-world examples and interprets results in a practical way. He also includes an extensive list of references for further reading along with exercises to reinforce understanding.

Web Resource

A wealth of supporting material can be downloaded from the books CRC Press web page, including:











Real-life data sets used in the text SAS and Stata programs used for examples in the text SAS and Stata programs for special techniques covered Sample size spreadsheet

Recenzijos

"This text, like its predecessors, hits the mark. The author writes extremely well and the text is resplendent with exercises. It would be a crime if Epidemiology: Study Design and Data Analysis were never used as a text! I wish a text like this had been available for my coursework. Enhancing its value as a text, it will be extremely useful as a reference book for its intended audienceresearchers and applied statisticians. the only excuse for an epidemiologist or applied statistician not to have it on his or her bookshelf is that he or she has not seen or heard of it. Make this book your next purchase!" Gregory E. Gilbert, The American Statistician, November 2014

Praise for Previous Editions:"As a text in quantitative epidemiology, this book also works nicely as a text in biostatisticsThe presentation style is relaxed, the examples are helpful, and the level of technical difficulty makes the material approachable without oversimplificationIt is sufficiently broad and deep in coverage to compete with standard texts in the field and has the added bonus of emphasizing study design. Methods and issues related to designs commonly used in a wide variety of health sciences are included" -Ken Hess, Department of Biomathematics and Biostatistics, Anderson Cancer Center

"The second edition of this epidemiology text is strengthened to cater to the two audiences the author has in mind: applied statisticians wishing to learn how their statistical expertise can be used in the epidemiology field and statistic-curious researchers who want to understand how statistical techniques can be used to solve epidemiological problems. The result is a book that will invariably appeal to the intended audience, one with practical applications of techniques and interpretations of results in an epidemiological context. The book is most certainly an ambitious attempt at covering a broad array of the most important epidemiologic study designs and analytical methods. This is further enforced by the addition of the meta-analysis chapter. This book will be valuable to statisticians in applying their discipline to epidemiology. Mark Woodward's excellent second edition will effectively serve post-graduate or advanced undergraduate students studying epidemiology, as well as statisticians or researchers who are regularly confronted with epidemiological questions." -Journal of the American Statistical Association "This book provides very good coverage of major issues in the design of epidemiological studies, and a decent, but very quick, tour of commonly used statistical models for such studies." -Short Book Reviews Publication of the International Statistical Institute, K.S. Brown, University of Waterloo, Canada

"Amazingly, Woodward manages to describe quite sophisticated models and analysis with nothing more complicated than summation signs. I highly recommend it." -Statistics in Medicine, 2006

"The second edition of this concisely written book covers all statistical methods being of relevance for the planning and analysis of epidemiological studies where the author avoids unnecessary mathematical details for the sake of comprehensibility. The presented statistical principles are always carefully discussed in the context of epidemiological concepts, for instance depending on the different study designs. Detailed practical examples coming from real studies as far as possible illustrate their application. The book can be highly recommended to researchers in epidemiology who want to understand better the statistical principles being typically applied in this field and to statisticians who want to understand more about statistics in epidemiology, but also to graduate students in epidemiology, public health, medical research and statistics." -Biometrics, Sept. 2005

"I think anyone with an interest in both biostatistics and epidemiology will want a copy this book on their bookshelf it is a first-rate reference book."

"I find Professor Woodward's text the most complete and practical introduction to the design and analysis of epidemiological studies I've encountered an excellent text for either a course introducing epidemiologists to statistical thought and methods or a course introducing statisticians to epidemiological thought and methods students appreciate having a readable textbook replete with understandable examples and worked exercisesoffers a complete introduction to statistical and epidemiological methods in the study of disease in human populations. All of the standard topics are included, and the second edition even has a chapter on meta-analysis. This book can be used as a text to introduce epidemiological methods to graduate students in statistics who have no background in epidemiology, or vice versaProfessor Woodward is to be congratulated on a job well done." -Dan McGee, Dept of Statistics, Florida State University

1 Fundamental issues 1(22)
1.1 What is epidemiology?
1(1)
1.2 Case studies: the work of Doll and Hill
2(4)
1.3 Populations and samples
6(1)
1.3.1 Populations
6(1)
1.3.2 Samples
7(1)
1.4 Measuring disease
7(3)
1.4.1 Incidence and prevalence
9(1)
1.5 Measuring the risk factor
10(1)
1.6 Causality
11(3)
1.6.1 Association
11(2)
1.6.2 Problems with establishing causality
13(1)
1.6.3 Principles of causality
14(1)
1.7 Studies using routine data
14(3)
1.7.1 Ecological data
15(1)
1.7.2 National sources of data on disease
16(1)
1.7.3 National sources of data on risk factors
17(1)
1.7.4 International data
17(1)
1.8 Study design
17(3)
1.8.1 Intervention studies
18(1)
1.8.2 Observational studies
19(1)
1.9 Data analysis
20(1)
Exercises
21(2)
2 Basic analytical procedures 23(66)
2.1 Introduction
23(1)
2.1.1 Inferential procedures
23(1)
2.2 Case study
24(1)
2.2.1 The Scottish Heart Health Study
24(1)
2.3 Types of variables
25(2)
2.3.1 Qualitative variables
26(1)
2.3.2 Quantitative variables
26(1)
2.3.3 The hierarchy of type
26(1)
2.4 Tables and charts
27(6)
2.4.1 Tables in reports
29(4)
2.4.2 Diagrams in reports
33(1)
2.5 Inferential techniques for categorical variables
33(8)
2.5.1 Contingency tables
33(3)
2.5.2 Binary variables: proportions and percentages
36(4)
2.5.3 Comparing two proportions or percentages
40(1)
2.6 Descriptive techniques for quantitative variables
41(16)
2.6.1 The five-number summary
43(3)
2.6.2 Quantiles
46(2)
2.6.3 The two-number summary
48(2)
2.6.4 Other summary statistics of spread
50(1)
2.6.5 Assessing symmetry
50(3)
2.6.6 Investigating shape
53(4)
2.7 Inferences about means
57(9)
2.7.1 Checking normality
58(2)
2.7.2 Inferences for a single mean
60(1)
2.7.3 Comparing two means
61(3)
2.7.4 Paired data
64(2)
2.8 Inferential techniques for non-normal data
66(6)
2.8.1 Transformations
66(3)
2.8.2 Nonparametric tests
69(3)
2.8.3 Confidence intervals for medians
72(1)
2.9 Measuring agreement
72(7)
2.9.1 Quantitative variables
72(2)
2.9.2 Categorical variables
74(3)
2.9.3 Ordered categorical variables
77(1)
2.9.4 Internal consistency
78(1)
2.10 Assessing diagnostic tests
79(6)
2.10.1 Accounting for sensitivity and specificity
81(4)
Exercises
85(4)
3 Assessing risk factors 89(36)
3.1 Risk and relative risk
89(3)
3.2 Odds and odds ratio
92(2)
3.3 Relative risk or odds ratio?
94(3)
3.4 Prevalence studies
97(1)
3.5 Testing association
98(7)
3.5.1 Equivalent tests
99(1)
3.5.2 One-sided tests
100(1)
3.5.3 Continuity corrections
101(1)
3.5.4 Fisher's exact test
102(2)
3.5.5 Limitations of tests
104(1)
3.6 Risk factors measured at several levels
105(6)
3.6.1 Continuous risk factors
107(1)
3.6.2 A test for linear trend
108(3)
3.6.3 A test for nonlinearity
111(1)
3.7 Attributable risk r s
111(5)
3.8 Rate and relative rate
116(3)
3.8.1 The general epidemiological rate
119(1)
3.9 Measures of difference
119(1)
3.10 EPITAB commands in Stata
120(1)
Exercises
121(4)
4 Confounding and interaction 125(40)
4.1 Introduction
125(1)
4.2 The concept of confounding
126(3)
4.3 Identification of confounders
129(2)
4.3.1 A strategy for selection
130(1)
4.4 Assessing confounding
131(3)
4.4.1 Using estimation
131(1)
4.4.2 Using hypothesis tests
132(1)
4.4.3 Dealing with several confounding variables
133(1)
4.5 Standardisation
134(9)
4.5.1 Direct standardisation of event rates
135(3)
4.5.2 Indirect standardisation of event rates
138(3)
4.5.3 Standardisation of risks
141(2)
4.6 Mantel-Haenszel methods
143(6)
4.6.1 The Mantel-Haenszel relative risk
146(1)
4.6.2 The Cochran-Mantel-Haenszel test
147(1)
4.6.3 Further comments
148(1)
4.7 The concept of interaction
149(2)
4.8 Testing for interaction
151(9)
4.8.1 Using the relative risk
151(5)
4.8.2 Using the odds ratio
156(2)
4.8.3 Using the risk difference
158(1)
4.8.4 Which type of interaction to use?
159(1)
4.8.5 Which interactions to test?
159(1)
4.9 Dealing with interaction
160(1)
4.10 EPITAB commands in Stata
161(1)
Exercises
161(4)
5 Cohort studies 165(46)
5.1 Design considerations
165(4)
5.1.1 Advantages
165(1)
5.1.2 Disadvantages
165(2)
5.1.3 Alternative designs with economic advantages
167(1)
5.1.4 Studies with a single baseline sample
168(1)
5.2 Analytical considerations
169(4)
5.2.1 Concurrent follow-up
169(1)
5.2.2 Moving baseline dates
170(1)
5.2.3 Varying follow-up durations
170(2)
5.2.4 Withdrawals
172(1)
5.3 Cohort life tables
173(8)
5.3.1 Allowing for sampling variation
175(1)
5.3.2 Allowing for censoring
176(1)
5.3.3 Comparison of two life tables
177(3)
5.3.4 Limitations
180(1)
5.4 Kaplan-Meier estimation
181(3)
5.4.1 An empirical comparison
182(2)
5.5 Comparison of two sets of survival probabilities
184(6)
5.5.1 Mantel-Haenszel methods
184(2)
5.5.2 The log-rank test
186(2)
5.5.3 Weighted log-rank tests
188(2)
5.5.4 Allowing for confounding variables
190(1)
5.5.5 Comparing three or more groups
190(1)
5.6 Competing risk
190(3)
5.7 The person-years method
193(10)
5.7.1 Age-specific rates
194(2)
5.7.2 Summarisation of rates
196(1)
5.7.3 Comparison of two SERs
197(2)
5.7.4 Mantel-Haenszel methods
199(3)
5.7.5 Further comments
202(1)
5.8 Period-cohort analysis
203(3)
5.8.1 Period-specific rates
204(2)
Exercises
206(5)
6 Case-control studies 211(46)
6.1 Basic design concepts
211(3)
6.1.1 Advantages
211(1)
6.1.2 Disadvantages
212(2)
6.2 Basic methods of analysis
214(6)
6.2.1 Dichotomous exposure
214(3)
6.2.2 Polytomous exposure
217(1)
6.2.3 Confounding and interaction
218(1)
6.2.4 Attributable risk
218(2)
6.3 Selection of cases
220(2)
6.3.1 Definition
220(1)
6.3.2 Inclusion and exclusion criteria
220(1)
6.3.3 Incident or prevalent?
221(1)
6.3.4 Source
221(1)
6.3.5 Consideration of bias
221(1)
6.4 Selection of controls
222(7)
6.4.1 General principles
222(2)
6.4.2 Hospital controls
224(2)
6.4.3 Community controls
226(1)
6.4.4 Other sources
227(1)
6.4.5 How many?
228(1)
6.5 Matching
229(2)
6.5.1 Advantages
229(1)
6.5.2 Disadvantages
230(1)
6.5.3 One-to-many matching
231(1)
6.5.4 Matching in other study designs
231(1)
6.6 The analysis of matched studies
231(14)
6.6.1 1 : 1 Matching
232(2)
6.6.2 1 : c Matching
234(6)
6.6.3 1 : Variable matching
240(2)
6.6.4 Many: many matching
242(3)
6.6.5 A modelling approach
245(1)
6.7 Nested case-control studies
245(3)
6.7.1 Matched studies
247(1)
6.7.2 Counter-matched studies
248(1)
6.8 Case-cohort studies
248(2)
6.9 Case-crossover studies
250(1)
Exercises
251(6)
7 Intervention studies 257(38)
7.1 Introduction
257(2)
7.1.1 Advantages
259(1)
7.1.2 Disadvantages
259(1)
7.2 Ethical considerations
259(2)
7.2.1 The protocol
260(1)
7.3 Avoidance of bias
261(4)
7.3.1 Use of a control group
261(1)
7.3.2 Blindness
262(1)
7.3.3 Randomisation
263(1)
7.3.4 Consent before randomisation
264(1)
7.3.5 Analysis by intention-to-treat
265(1)
7.4 Parallel group studies
265(8)
7.4.1 Number needed to treat
268(2)
7.4.2 Cluster randomised trials
270(1)
7.4.3 Stepped wedge trials
270(1)
7.4.4 Non-inferiority trials
271(2)
7.5 Cross-over studies
273(11)
7.5.1 Graphical analysis
275(2)
7.5.2 Comparing means
277(5)
7.5.3 Analysing preferences
282(1)
7.5.4 Analysing binary data
283(1)
7.6 Sequential studies
284(2)
7.6.1 The Haybittle-Peto stopping rule
285(1)
7.6.2 Adaptive designs
286(1)
7.7 Allocation to treatment group
286(5)
7.7.1 Global randomisation
286(2)
7.7.2 Stratified randomization
288(3)
7.7.3 Implementation
291(1)
7.8 Trials as cohorts
291(1)
Exercises
291(4)
8 Sample size determination 295(36)
8.1 Introduction
295(1)
8.2 Power
296(7)
8.2.1 Choice of alternative hypothesis
300(3)
8.3 Testing a mean value
303(4)
8.3.1 Common choices for power and significance level
305(1)
8.3.2 Using a table of sample sizes
305(1)
8.3.3 The minimum detectable difference
306(1)
8.3.4 The assumption of known standard deviation
307(1)
8.4 Testing a difference between means
307(4)
8.4.1 Using a table of sample sizes
308(2)
8.4.2 Power and minimum detectable difference
310(1)
8.4.3 Optimum distribution of the sample
310(1)
8.4.4 Paired data
311(1)
8.5 Testing a proportion
311(2)
8.5.1 Using a table of sample sizes
312(1)
8.6 Testing a relative risk
313(4)
8.6.1 Using a table of sample sizes
315(1)
8.6.2 Power and minimum detectable relative risk
316(1)
8.7 Case-control studies
317(7)
8.7.1 Using a table of sample sizes
319(1)
8.7.2 Power and minimum detectable relative risk
319(2)
8.7.3 Comparison with cohort studies
321(1)
8.7.4 Matched studies
321(3)
8.8 Complex sampling designs
324(1)
8.9 Concluding remarks
325(1)
Exercises
326(5)
9 Modelling quantitative outcome variables 331(78)
9.1 Statistical models
331(1)
9.2 One categorical explanatory variable
332(12)
9.2.1 The hypotheses to be tested
332(1)
9.2.2 Construction of the ANOVA table
333(3)
9.2.3 How the ANOVA table is used
336(1)
9.2.4 Estimation of group means
336(1)
9.2.5 Comparison of group means
337(1)
9.2.6 Fitted values
338(3)
9.2.7 Using computer packages
341(3)
9.3 One quantitative explanatory variable
344(14)
9.3.1 Simple linear regression
344(8)
9.3.2 Correlation
352(3)
9.3.3 Nonlinear regression
355(3)
9.4 Two categorical explanatory variables
358(7)
9.4.1 Model specification
358(1)
9.4.2 Model fitting
359(1)
9.4.3 Balanced data
359(1)
9.4.4 Unbalanced data
359(3)
9.4.5 Fitted values
362(1)
9.4.6 Least squares means
363(1)
9.4.7 Interaction
364(1)
9.5 Model building
365(6)
9.6 General linear models
371(6)
9.7 Several explanatory variables
377(6)
9.7.1 Information criteria
381(2)
9.7.2 Boosted regression
383(1)
9.8 Model checking
383(4)
9.9 Confounding
387(5)
9.9.1 Adjustment using residuals
391(1)
9.10 Splines
392(6)
9.10.1 Choice of knots
395(1)
9.10.2 Other types of splines
396(2)
9.11 Panel data
398(4)
9.12 Non-normal alternatives
402(2)
Exercises
404(5)
10 Modelling binary outcome data 409(98)
10.1 Introduction
409(3)
10.2 Problems with standard regression models
412(1)
10.2.1 The r-x relationship may well not be linear
412(1)
10.2.2 Predicted values of the risk may be outside the valid range
412(1)
10.2.3 The error distribution is not normal
412(1)
10.3 Logistic regression
413(2)
10.4 Interpretation of logistic regression coefficients
415(12)
10.4.1 Binary risk factors
415(2)
10.4.2 Quantitative risk factors
417(2)
10.4.3 Categorical risk factors
419(5)
10.4.4 Ordinal risk factors
424(1)
10.4.5 Floating absolute risks
425(2)
10.5 Generic data
427(1)
10.6 Multiple logistic regression models
428(4)
10.7 Tests of hypotheses
432(12)
10.7.1 Goodness of fit for grouped data
433(2)
10.7.2 Goodness of fit for generic data
435(1)
10.7.3 Effect of a risk factor
435(3)
10.7.4 Information criteria
438(2)
10.7.5 Tests for linearity and nonlinearity
440(3)
10.7.6 Tests based upon estimates and their standard errors
443(1)
10.7.7 Problems with missing values
444(1)
10.8 Confounding
444(1)
10.9 Interaction
445(7)
10.9.1 Between two categorical variables
445(4)
10.9.2 Between a quantitative and a categorical variable
449(3)
10.9.3 Between two quantitative variables
452(1)
10.10 Dealing with a quantitative explanatory variable
452(7)
10.10.1 Linear form
453(1)
10.10.2 Categorical form
453(2)
10.10.3 Linear spline form
455(4)
10.10.4 Generalisations
459(1)
10.11 Model checking
459(3)
10.11.1 Residuals
459(3)
10.11.2 Influential observations
462(1)
10.12 Measurement error
462(5)
10.12.1 Regression to the mean
463(2)
10.12.2 Correcting for regression dilution
465(2)
10.13 Case-control studies
467(2)
10.13.1 Unmatched studies
467(1)
10.13.2 Matched studies
468(1)
10.14 Outcomes with several levels
469(6)
10.14.1 The proportional odds assumption
471(2)
10.14.2 The proportional odds model
473(2)
10.14.3 Multinomial regression
475(1)
10.15 Longitudinal data
475(1)
10.16 Binomial regression
476(12)
10.16.1 Adjusted risks
479(4)
10.16.2 Risk differences
483(1)
10.16.3 Problems with binomial models
484(4)
10.17 Propensity scoring
488(13)
10.17.1 Pair-matched propensity scores
488(1)
10.17.2 Stratified propensity scores
489(1)
10.17.3 Weighting by the inverse propensity score
490(1)
10.17.4 Adjusting for the propensity score
491(1)
10.17.5 Deriving the propensity score
492(1)
10.17.6 Propensity score outliers
493(1)
10.17.7 Conduct of the matched design
493(1)
10.17.8 Analysis of the matched design
494(1)
10.17.9 Case studies
495(3)
10.17.10 Interpretation of effects
498(1)
10.17.11 Problems with estimating uncertainty
499(1)
10.17.12 Propensity scores in practice
499(2)
Exercises
501(6)
11 Modelling follow-up data 507(58)
11.1 Introduction
507(1)
11.1.1 Models for survival data
507(1)
11.2 Basic functions of survival time
507(1)
11.2.1 The survival function
507(1)
11.2.2 The hazard function
507(1)
11.3 Estimating the hazard function
508(4)
11.3.1 Kaplan-Meier estimation
508(2)
11.3.2 Person-time estimation
510(1)
11.3.3 Actuarial estimation
511(1)
11.3.4 The cumulative hazard
512(1)
11.4 Probability models
512(9)
11.4.1 The probability density and cumulative distribution functions
512(2)
11.4.2 Choosing a model
514(1)
11.4.3 The exponential distribution
514(3)
11.4.4 The Weibull distribution
517(3)
11.4.5 Other probability models
520(1)
11.5 Proportional hazards regression models
521(5)
11.5.1 Comparing two groups
521(1)
11.5.2 Comparing several groups
521(2)
11.5.3 Modelling with a quantitative variable
523(1)
11.5.4 Modelling with several variables
524(1)
11.5.5 Left-censoring
525(1)
11.6 The Cox proportional hazards model
526(10)
11.6.1 Time-dependent covariates
535(1)
11.6.2 Recurrent events
536(1)
11.7 The Weibull proportional hazards model
536(5)
11.8 Model checking
541(5)
11.8.1 Log cumulative hazard plots
541(4)
11.8.2 An objective test of proportional hazards for the Cox model
545(1)
11.8.3 An objective test of proportional hazards for the Weibull model
545(1)
11.8.4 Residuals and influence
546(1)
11.8.5 Nonproportional hazards
546(1)
11.9 Competing risk
546(3)
11.9.1 Joint modeling of longitudinal and survival data
548(1)
11.10 Poisson regression
549(10)
11.10.1 Simple regression
550(3)
11.10.2 Multiple regression
553(2)
11.10.3 Comparison of standardised event ratios
555(1)
11.10.4 Routine or registration data
556(2)
11.10.5 Generic data
558(1)
11.10.6 Model checking
559(1)
11.11 Pooled logistic regression
559(2)
Exercises
561(4)
12 Meta-analysis 565(40)
12.1 Reviewing evidence
565(2)
12.1.1 The Cochrane Collaboration
567(1)
12.2 Systematic review
567(5)
12.2.1 Designing a systematic review
567(4)
12.2.2 Study quality
571(1)
12.3 A general approach to pooling
572(12)
12.3.1 Inverse variance weighting
573(1)
12.3.2 Fixed effect and random effects
573(1)
12.3.3 Quantifying heterogeneity
574(2)
12.3.4 Estimating the between-study variance
576(1)
12.3.5 Calculating inverse variance weights
577(1)
12.3.6 Calculating standard errors from confidence intervals
577(1)
12.3.7 Case studies
578(4)
12.3.8 Pooling risk differences
582(1)
12.3.9 Pooling differences in mean values
583(1)
12.3.10 Other quantities
583(1)
12.3.11 Pooling mixed quantities
583(1)
12.3.12 Dose-response meta-analysis
584(1)
12.4 Investigating heterogeneity
584(7)
12.4.1 Forest plots
585(1)
12.4.2 Influence plots
586(2)
12.4.3 Sensitivity analyses
588(1)
12.4.4 Meta-regression
588(3)
12.5 Pooling tabular data
591(2)
12.5.1 Inverse variance weighting
591(1)
12.5.2 Mantel-Haenszel methods
591(1)
12.5.3 The Peto method
592(1)
12.5.4 Dealing with zeros
592(1)
12.5.5 Advantages and disadvantages of using tabular data
593(1)
12.6 Individual participant data
593(1)
12.7 Dealing with aspects of study quality
594(1)
12.8 Publication bias
595(5)
12.8.1 The funnel plot
596(1)
12.8.2 Consequences of publication bias
597(1)
12.8.3 Correcting for publication bias
597(2)
12.8.4 Other causes of asymmetry in funnel plots
599(1)
12.9 Advantages and limitations of meta-analysis
600(1)
Exercises
600(5)
13 Risk scores and clinical decision rules 605(74)
13.1 Introduction
605(3)
13.1.1 Individual and population level interventions
605(2)
13.1.2 Scope of this chapter
607(1)
13.2 Association and prognosis
608(10)
13.2.1 The concept of discrimination
610(1)
13.2.2 Risk factor thresholds
611(4)
13.2.3 Risk thresholds
615(1)
13.2.4 Odds ratios and discrimination
616(2)
13.3 Risk scores from statistical models
618(7)
13.3.1 Logistic regression
618(2)
13.3.2 Multiple variable risk scores
620(1)
13.3.3 Cox regression
621(2)
13.3.4 Risk thresholds
623(1)
13.3.5 Multiple thresholds
624(1)
13.4 Quantifying discrimination
625(12)
13.4.1 The area under the curve
626(3)
13.4.2 Comparing AUCs
629(2)
13.4.3 Survival data
631(1)
13.4.4 The standardised mean effect size
632(5)
13.4.5 Other measures of discrimination
637(1)
13.5 Calibration
637(6)
13.5.1 Overall calibration
638(1)
13.5.2 Mean calibration
638(1)
13.5.3 Grouped calibration
639(2)
13.5.4 Calibration plots
641(2)
13.6 Recalibration
643(5)
13.6.1 Recalibration of the mean
643(1)
13.6.2 Recalibration of scores in a fixed cohort
643(3)
13.6.3 Recalibration of parameters from a Cox model
646(1)
13.6.4 Recalibration and discrimination
647(1)
13.7 The accuracy of predictions
648(3)
13.7.1 The Brier score
648(2)
13.7.2 Comparison of Brier scores
650(1)
13.8 Assessing an extraneous prognostic variable
651(1)
13.9 Reclassification
652(10)
13.9.1 The integrated discrimination improvement from a fixed cohort
653(3)
13.9.2 The net reclassification improvement from a fixed cohort
656(3)
13.9.3 The integrated discrimination improvement from a variable cohort
659(1)
13.9.4 The net reclassification improvement from a variable cohort
660(2)
13.9.5 Software
662(1)
13.10 Validation
662(1)
13.11 Presentation of risk scores
663(11)
13.11.1 Point scoring
664(10)
13.12 Impact studies
674(1)
Exercises
675(4)
14 Computer-intensive methods 679(76)
14.1 Rationale
679(1)
14.2 The bootstrap
679(5)
14.2.1 Bootstrap distributions
681(3)
14.3 Bootstrap confidence intervals
684(8)
14.3.1 Bootstrap normal intervals
685(1)
14.3.2 Bootstrap percentile intervals
686(2)
14.3.3 Bootstrap bias-corrected intervals
688(2)
14.3.4 Bootstrap bias-corrected and accelerated intervals
690(1)
14.3.5 Overview of the worked example
691(1)
14.3.6 Choice of bootstrap interval
692(1)
14.4 Practical issues when bootstrapping
692(4)
14.4.1 Software
692(1)
14.4.2 How many replications should be used?
693(3)
14.4.3 Sensible strategies
696(1)
14.5 Further examples of bootstrapping
696(7)
14.5.1 Complex bootstrap samples
701(2)
14.6 Bootstrap hypothesis testing
703(2)
14.7 Limitations of bootstrapping
705(1)
14.8 Permutation tests
706(3)
14.8.1 Monte Carlo permutation tests
707(2)
14.8.2 Limitations
709(1)
14.9 Missing values
709(7)
14.9.1 Dealing with missing values
711(2)
14.9.2 Types of missingness
713(1)
14.9.3 Complete case analyses
714(2)
14.10 Naive imputation methods
716(4)
14.10.1 Mean imputation
716(1)
14.10.2 Conditional mean and regression imputation
716(2)
14.10.3 Hot deck imputation and predictive mean matching
718(1)
14.10.4 Longitudinal data
719(1)
14.11 Univariate multiple imputation
720(13)
14.11.1 Multiple imputation by regression
720(1)
14.11.2 The three-step process in MI \
721(1)
14.11.3 Imputer's and analyst's models
722(1)
14.11.4 Rubin's equations
723(5)
14.11.5 Imputation diagnostics
728(1)
14.11.6 Skewed continuous data
729(2)
14.11.7 Other types of variables
731(1)
14.11.8 How many imputations?
731(2)
14.12 Multivariate multiple imputation
733(14)
14.12.1 Monotone imputation
733(1)
14.12.2 Data augmentation
734(8)
14.12.3 Categorical variables
742(1)
14.12.4 What to do when DA fails
742(1)
14.12.5 Chained equations
743(4)
14.12.6 Longitudinal data
747(1)
14.13 When is it worth imputing?
747(1)
Exercises
748(7)
Appendix A Materials available on the website for this book 755(4)
Appendix B Statistical tables 759(26)
Appendix C Additional datasets for exercises 785(14)
References 799(22)
Index 821
Mark Woodward is a professor of statistics and epidemiology at the University of Oxford, a professor of biostatistics in the George Institute at the University of Sydney, and an adjunct professor of epidemiology at Johns Hopkins University.