Statistics for Epidemiology

Authors: Nicholas P. Jewell 公卫考场
Publisher: Chapman & Hall / CRC
Copyright: 2004
ISBN-10: 1-58488-433-9
ISBN-13: 978-1-58488-433-0
Pages: 333; hardcover
Price: $67.00
Statistics for Epidemiology is the latest in a long line of texts that can be used to provide the basis for a course in statistical epidemiology aimed at graduate students in the medical professions. Given the target audience, such texts must strike a delicate balance so as to not be too theoretical while also providing enough statistical background to avoid producing a cookbook for a "plug-and-chug" course. This text very much succeeds in this regard. 公卫家园

Covered topics include some basic probability (including discussion of conditional probability and Berkson's bias), measures of risk, study designs, analysis of tables, interaction, regression models for binary outcomes, advanced logistic regression, matching, and Cox regression.

1 Introduction
1.1 Disease processes
1.2 Statistical approaches to epidemiological data

1.2.1 Study design
1.2.2 Binary outcome data

1.3 Causality
1.4 Overview

1.4.1 Caution: what is not covered

1.5 Comments and further reading

2 Measures of Disease Occurrence
2.1 Prevalence and incidence
2.2 Disease rates

2.2.1 The hazard function

2.3 Comments and further reading
2.4 Problems

3 The Role of Probability in Observational Studies 公卫人
3.1 Simple random samples
3.2 Probability and the incidence proportion
3.3 Inference based on an estimated probability
3.4 Conditional probabilities

3.4.1 Independence of two events

3.5 Example of conditional probabilities—Berkson's bias 公卫考场
3.6 Comments and further reading
3.7 Problems

4 Measure of Disease–Exposure Association
4.1 Relative risk
4.2 Odds ratio
4.3 The odds ratio as an approximation to the relative risk
4.4 Symmetry of roles of disease and exposure in the odds ratio
4.5 Relative hazard
4.6 Excess risk
4.7 Attributable risk
4.8 Comments and further reading
4.9 Problems

5 Study Designs
5.1 Population-based studies

5.1.1 Example—mother's marital status and infant birthweight

5.2 Exposure-based sampling—cohort studies 公卫论坛
5.3 Disease-based sampling —case–control studies
5.4 Key variants of the case–control design

5.4.1 Risk-set sampling of controls
5.4.2 Case-cohort studies

5.5 Comments and further reading
5.6 Problems

6 Assessing Significance in a 2 x 2 Table 公卫论坛
6.1 Population-based designs

6.1.1 Role of hypothesis tests and interpretation of p-values

6.2 Cohort designs
6.3 Case–control designs

6.3.1 Comparison of the study designs

6.4 Comments and further reading

6.4.1 Alternative formulations of the χ2 test statistic

公卫考场

6.4.2 When is the sample size too small to do a χ2 test?

6.5 Problems

7 Estimation and Inference for Measures of Association
7.1 The odds ratio

7.1.1 Sampling distribution of the odds ratio
7.1.2 Confidence interval for the odds ratio 公卫家园
7.1.3 Example—coffee drinking and pancreatic cancer
7.1.4 Small sample adjustments for estimators of the odds ratio

7.2 The relative risk

7.2.1 Example—coronary heart disease in the Western Collaborative Group Study

7.3 The excess risk
7.4 The attributable risk
7.5 Comments and further reading

7.5.1 Measurement error or misclassification

7.6 Problems

8 Causal Inference and Extraneous Factors: Confounding and Interaction
8.1 Causal inference

8.1.1 Counterfactuals

公卫家园

8.1.2 Confounding variables
8.1.3 Control of confounding by stratification

8.2 Causal graphs

8.2.1 Assumptions in causal graphs
8.2.2 Causal graph associating childhood vaccination to subsequent health condition
8.2.3 Using causal graphs to infer the presence of confounding 公卫人

8.3 Controlling confounding in causal graphs

8.3.1 Danger: controlling for colliders
8.3.2 Simple rules for using a causal graph to choose the crucial confounders

8.4 Collapsibility over strata
8.5 Comments and further reading 公卫人
8.6 Problems

9 Control of Extraneous Factors
9.1 Summary test of association in a series of 2 x 2 tables

9.1.1 The Cochran–Mantel–Haenszel test
9.1.2 Sample size issues and a historical note

9.2 Summary estimates and confidence intervals for the odds ratio, adjusting for confounding factors

9.2.1 Woolf's method on the logarithm scale
9.2.2 The Mantel–Haenszel method
9.2.3 Example—the Western Collaborative Group Study: part 2
9.2.4 Example—coffee drinking and pancreatic cancer: part 2

9.3 Summary estimates and confidence intervals for the relative risk, adjusting for confounding factors

公卫家园

9.3.1 Example—the Western Collaborative Group Study: part3

9.4 Summary estimates and confidence intervals for the excess risk, adjusting for confounding factors

9.4.1 Example—the Western Collaborative Group Study: part4

9.5 Further discussion of confounding

9.5.1 How do adjustments for confounding affect precision?
9.5.2 An empirical approach to confounding

9.6 Comments and further reading
9.7 Problems

10 Interaction
10.1 Multiplicative and additive interaction

10.1.1 Multiplicative interaction

公卫论坛

10.1.2 Additive interaction

10.2 Interaction and counterfactuals
10.3 Test of consistency of association across strata

10.3.1 The Woolf method
10.3.2 Alternative tests of homogeneity
10.3.3 Example—the Western Collaborative Group Study: part 5
10.3.4 The power of the test for homogeneity

10.4 Example of extreme interaction
10.5 Comments and further reading
10.6 Problems

11 Exposures at Several Discrete Levels
11.1 Overall test of association
11.2 Example—coffee drinking and pancreatic cancer: part 3 公卫考场
11.3 A test for trend in risk

11.3.1 Qualitatively ordered exposure variables
11.3.2 Goodness of fit and nonlinear trends in risk

11.4 Example—the Western Collaborative Group Study: part 6
11.5 Example—coffee drinking and pancreatic cancer: part 4 公卫人
11.6 Adjustment for confounding, exact tests, and interaction
11.7 Comments and further reading
11.8 Problems

12 Regression Models Relating Exposure to Disease
12.1 Some introductory regression models

12.1.1 The linear model

公卫百科

12.1.2 Pros and cons of the linear model

12.2 The log linear model
12.3 The probit model
12.4 The simple logistic regression model

12.4.1 Interpretation of logistic regression parameters

12.5 Simple examples of the models with a binary exposure 公卫百科
12.6 Multiple logistic regression model

12.6.1 The use of indicator variables for discrete exposures

12.7 Comments and further reading
12.8 Problems

13 Estimation of Logistic Regression Model Parameters
13.1 The likelihood function

13.1.1 The likelihood function based on a logistic regression model
13.1.2 Properties of the log likelihood function and the maximum likelihood estimate
13.1.3 Null hypotheses that specify more than one regression coefficient

13.2 Example—the Western Collaborative Group Study: part 7

公卫百科

13.3 Logistic regression with case–control data
13.4 Example—coffee drinking and pancreatic cancer: part 5
13.5 Comments and further reading
13.6 Problems

14 Confounding and Interaction within Logistic Regression Models
14.1 Assessment of confounding using logistic regression models

公卫论坛

14.1.1 Example—the Western Collaborative Group Study: part 8

14.2 Introducing interaction into the multiple logistic regression model
14.3 Example—coffee drinking and pancreatic cancer: part 6
14.4 Example—the Western Collaborative Group Study: part 9 公卫家园
14.5 Collinearity and centering variables

14.5.1 Centering independent variables
14.5.2 Fitting quadratic models

14.6 Restrictions on effective use of maximum likelihood techniques
14.7 Comments and further reading

14.7.1 Measurement error

公卫家园

14.7.2 Missing data

14.8 Problems

15 Goodness of Fit Tests for Logistic Regression Models and Model Building
15.1 Choosing the scale of an exposure variable

15.1.1 Using ordered categories to select exposure scale
15.1.2 Alternative strategies 公卫家园

15.2 Model building
15.3 Goodness of fit

15.3.1 The Hosmer–Lemeshow test

15.4 Comments and further reading
15.5 Problems

16 Matched Studies
16.1 Frequency matching
16.2 Pair matching
16.2.1 Mantel–Haenszel techniques applied to pair-matched data

公卫考场

16.2.2 Small sample adjustment for odds ratio estimator

16.3 Example—pregnancy and spontaneous abortion in relation to coronary heart disease in women
16.4 Confounding and interaction effects

16.4.1 Assessing interaction effects of matching variables 公卫论坛
16.4.2 Possible confounding and interactive efforts due to nonmatching variables

16.5 The logisitic regression model for matched data

16.5.1 Example—pregnancy and spontaneous abortion in relation to coronary heart disease in women: part 2 公卫考场

16.6 Example—the effect of birth order on respiratory distress syndrome in twins
16.7 Comments and further reading

16.7.1 When can we break the match?
16.7.2 Final thoughts on matching

16.8 Problems

17 Alternatives and Extensions to the Logistic Regression Model 公卫论坛
17.1 Flexible regression model
17.2 Beyond binary outcomes and independent observations
17.3 Introducing general risk factors into formulation of the relative hazard—the Cox model
17.4 Fitting the Cox regression model
17.5 When does time at risk confound an exposure–disease relationship?

17.5.1 Time-dependent exposures
17.5.2 Differential loss to follow-up

17.6 Comments and further reading
17.7 Problems

18 Epilogue: The Examples
References
Glossary of Common Terms and Abbreviations
Index