Applying Quantitative Bias Analysis to Epidemiologic Data

ISBN-13: 9783030826727 / Angielski / Twarda / 2022 / 484 str.

Timothy Lash; Matthew Fox; Richard Maclehose

Spis treści

Part I: Introduction

1 Introduction and Objectives

1 Introduction 

1.2 Nonrandomized Epidemiologic Research 

1.3 The Treatment of Uncertainty in Nonrandomized Research 

1.4 Objective 

1.5 Conclusion 

2 A Guide to Implementing Quantitative Bias Analysis 

2.1 Introduction 

2.2 Reducing Error 

2.3 Reducing Error by Design 

2.4 Reducing Error in the Analysis 

2.5 Quantifying Error 

2.6 Evaluating the Potential Value of Quantitative Bias Analysis

2.7 Planning for Bias Analysis 

2.8 Creating a Data Collection Plan for Bias Analysis 

2.9 Creating an Analytic Plan for a Bias Analysis 

2.10 Bias Analysis Techniques 

2.11 Introduction to Inference 

2.12 Conclusion 

3 Data Sources for Bias Analysis 

3.1 Bias Parameters 

3.2 Internal Data Sources 

3.3 Selection Bias 

3.4 Uncontrolled Confounder 

3.5 Information Bias 

3.6 Limitations of Internal Validation Studies 

3.7 External Data Sources 

3.8 Selection Bias 

3.9 Uncontrolled Confounder 

3.10 Information Bias 

3.11 Summary

Part II: Preliminary Methods to Adjust for Systematic Errors 

4 Selection Bias 

4.1 Introduction 

4.2 Definitions and Terms

4.3 Motivation for Bias Analysis 

4.4 Sources of Data 

4.5 Simple Correction for Differential Initial Participation 

4.6 Simple Correction for Differential Loss-to-Follow-up

4.7 Sensitivity Analysis of the Bias Analysis 

4.7 Signed Directed Acyclic Graphs to Estimate the Direction of Bias 

5 Uncontrolled Confounders 

5.1 Introduction 

5.2 Definitions and Terms

5.3 Motivation for Bias Analysis 

5.4 Sources of Data

5.5 Introduction to Simple Bias Analysis 

5.6 Implementation of Simple Bias Analysis

5.7 Sensitivity Analysis of the Bias Analysis 

5.8 Uncontrolled Confounder in the Presence of Effect Modification 

5.9 Polytomous Confounders 

5.10 Bounding the Bias Limits of Uncontrolled Confounding

5.10 Signed Directed Acyclic Graphs to Estimate the Direction of Bias

5.11 Uncontrolled Confounding with Continuous Outcome, Exposure, or Confounder 

6 Misclassification 

6.1 Introduction 

6.2 Definitions and Terms

6.3 Motivation for Bias Analysis

6.4 Sources of Data

6.5 Calculating Classification Bias Parameters from Validation Data

6.6 Exposure Misclassification for Dichotomous Exposures

6.7 Exposure Misclassification for Polytomous Exposures

6.8 Disease Misclassification 

6.9 Covariate Misclassification 

6.10 Dependent Misclassification

6.11 Sensitivity Analysis of the Bias Analysis

6.12 Adjusting Standard Errors for Corrections 

7 Measurement Error for Continuous Variables

7.1 Introduction

7.2 Definition and Terms

7.3 Motivation for Bias Analysis

7.4 Exposure Measurement error

7.5 Outcome Measurement error

7.6 Covariate Measurement Error

7.7 Correlated errors 

8 Multiple Bias Modeling 

8.1 Introduction 

8.2 Order of Bias Analyses

8.3 Multiple Bias Analysis, Simple Methods

Part III: Methods to Incorporate Systematic and Random Errors 

9 Bias Analysis by Simulation for Summary Level Data

9.1 Introduction 

9.2 Probability Distributions 

9.3 Correlated Distributions 

9.4 Analytic Approach 

9.5 Exposure Misclassification Implementation

9.6 Exposure Measurement Error Implementation 

9.7 Uncontrolled Confounding Implementation 

9.8 Selection Bias Implementation 

10 Bias Analysis by Simulation for Record Level Data

10.1 Introduction 

10.2 Analytic Approach 

10.3 Exposure Misclassification Implementation

10.4 Exposure Measurement Error Implementation 

10.5 Uncontrolled Confounding Implementation 

10.6 Selection Bias Implementation 

11 Combining Systematic and Random Error

11.1 Analytic approximation

11.2 Resampling approximation

11.3 Bootstrapping 

12 Bias Analysis by Missing Data Methods

12.1 Introduction 

12.2 Analytic Approach 

12.3 Exposure Misclassification Implementation

12.4 Exposure Measurement Error Implementation 

12.5 Uncontrolled Confounding Implementation 

12.6 Selection Bias Implementation 

12.7 Combining Systematic and Random Error 

13 Bias Analysis by Empirical Methods

13.1 Introduction 

13.2 Analytic Approach 

13.3 Exposure Misclassification Implementation 

13.4 Exposure Measurement Error Implementation

13.5 Uncontrolled Confounding Implementation 

13.6 Selection Bias Implementation 

13.7 Combining Systematic and Random Error 

14 Bias Analysis by Bayesian Methods

14.1 Introduction 

14.2 Analytic Approach 

14.3 Exposure Misclassification Implementation 

14.4 Exposure Measurement Error Implementation 

14.5 Uncontrolled Confounding Implementation 

14.6 Selection Bias Implementation 

14.7 Combining Systematic and Random Error 

15 Multiple Bias Modeling

15.1 Multiple Bias Analysis, Probabilistic Methods

15.2 Multiple Bias Analysis, Missing Data Methods

15.3 Multiple Bias Analysis, Empirical Methods

15.4 Multiple Bias Analysis, Bayesian Methods 

Part IV: Good Practices

16 Good Practices for Quantitative Bias Analysis

16.1 Selection of bias sources

16.2 Selection of analytic strategies

16.3 Selection of values to assign to bias parameters

17 Presentation and Inference 

17.1 Presentation of simple and multidimensional bias analyses

17.2 Presentation of advanced bias analyses 

17.3 Inference 

17.4 Caveats and Cautions 

18 References 

19 Index

O autorze

Timothy Lash, D.Sc., M.P.H., is professor in the Department of Epidemiology at the Rollins School of Public Health and honorary professor of cancer epidemiology in the Department of Clinical Epidemiology at Aarhus University in Aarhus, Denmark. Dr. Lash is also past-President of the Society for Epidemiologic Research (SER) for the 2014-2015 term. His research focuses on predictors of cancer recurrence, including molecular predictors of treatment effectiveness and late recurrence, and he also researches methods and applications of quantitative bias analysis. 

Matthew Fox, D.Sc., M.P.H, is associate professor in the Center for Global Health & Development and in the Department of Epidemiology at Boston University. Before joining Boston University, he was a Peace Corps volunteer in the former Soviet Republic of Turkmenistan. Dr. Fox is currently funded through a K award from the National Institutes of Allergy and Infectious Diseases to work on ways to improve retention in HIV-care programs in South Africa from time of testing HIV-positive through long-term treatment. His research interests include treatment outcomes in HIV-treatment programs, infectious disease epidemiology, and epidemiological methods, including quantitative bias analysis.

Richard MacLehose, Ph.D., is associate professor in the Division of Epidemiology and Community Health at the University of Minnesota. Dr. MacLehose received his M.S. in epidemiology from the University of Washington and his Ph.D. in epidemiology from the University of North Carolina. His research interests include Bayesian statistics (including bias analysis), epidemiologic methods, applied biostatistics, and reproductive and environmental health.

Blurb

This textbook and guide focuses on methodologies for bias analysis in epidemiology and public health, not only providing updates to the first edition but also further developing methods and adding new advanced methods.

As computational power available to analysts has improved and epidemiologic problems have become more advanced, missing data, Bayes, and empirical methods have become more commonly used. This new edition features updated examples throughout and adds coverage addressing:

• Measurement error pertaining to continuous and polytomous variables
• Methods surrounding person-time (rate) data
• Bias analysis using missing data, empirical (likelihood), and Bayes methods

A unique feature of this revision is its section on best practices for implementing, presenting, and interpreting bias analyses. Pedagogically, the text guides students and professionals through the planning stages of bias analysis, including the design of validation studies and the collection of validity data from other sources. Three chapters present methods for corrections to address selection bias, uncontrolled confounding, and measurement errors, and subsequent sections extend these methods to probabilistic bias analysis, missing data methods, likelihood-based approaches, Bayesian methods, and best practices.