Applying quantitative bias analysis to epidemiologic data.

Preface
Acknowledgments
Contents
Chapter 1: Introduction, Objectives, and an Alternative
	Introduction: Biases in Health Research
	Statistical Inference in Public Health Research
		The Treatment of Uncertainty in Nonrandomized Research
		When Bias Analysis Will Be Most Useful
	Judgments Under Uncertainty
		The Dual-Process Model of Cognition
		Anchoring and Adjustment
		Overconfidence
		Failure to Account for the Base-Rate
		Conclusion
	References
Chapter 2: A Guide to Implementing Quantitative Bias Analysis
	Introduction
	Reducing Error
		Reducing Error by Design
		Reducing Error in the Analysis
	Quantifying Error
		Evaluating the Potential Value of Quantitative Bias Analysis?
		Planning for Bias Analysis
		Creating a Data Collection Plan for Bias Analysis
	Creating an Analytic Plan for a Bias Analysis
		Type of Data: Record-Level Versus Summary
		Type of Bias Analysis
		Order of Bias Analysis Adjustments
	Bias Analysis Techniques
		Simple Bias Analysis
		Multidimensional Bias Analysis
		Probabilistic Bias Analysis
		Multiple Bias Modeling
		Direct Bias Modeling and Missing Data Methods
		Bayesian Bias Analysis
	Assigning Values and Distributions to Bias Parameters
	Directed Acyclic Graphs
	Conclusion
	References
Chapter 3: Data Sources for Bias Analysis
	Bias Parameters
	Internal Data Sources
		Selection Bias
		Uncontrolled Confounding
		Information Bias
		Design of Internal Validation Studies
		Limitations of Internal Validation Studies
	External Data Sources
		Selection Bias
		Unmeasured Confounder
		Information Bias
	Expert Opinion
	Summary
	References
Chapter 4: Selection Bias
	Introduction
	Definitions and Terms
		Conceptual
		Depicting Selection Bias Using Causal Graphs
		Design Considerations
	Bias Analysis
		Motivation for Bias Analysis
		Sources of Data
	Simple Bias-Adjustment for Differential Initial Participation
		Example
		Introduction to Bias Analysis
		Bias Analysis by Projecting the Exposed Proportion Among Nonparticipants
		Bias Analysis Using Selection Proportions
		Bias Analysis Using Inverse Probability of Participation Weighting
	Simple Bias-Adjustment for Differential Loss-to-Follow-up
		Example
		Bias Analysis by Modeling Outcomes
		Bias Analysis by Inverse Probability of Attrition Weighting
	Multidimensional Bias Analysis for Selection Bias
		Example
	References
Chapter 5: Uncontrolled Confounders
	Introduction
	Key Concepts
		Definitions
		Motivation for Bias Analysis
		Data Sources
	Introduction to Simple Bias Analysis
		Approach
		Introduction to the Example
		Bias Parameters
	Implementation of Simple Bias Analysis
		Ratio Measures
		Example
		Difference Measures
		Person-time Designs
		Unmeasured Confounder in the Presence of Effect Measure Modification
		Polytomous Confounders
	Multidimensional Bias Analysis for Unmeasured Confounding
		Example
	Bounding the Bias Limits of an Unmeasured Confounding
		Analytic Approach
		The E-Value and G-Value
		Signed Directed Acyclic Graphs to Estimate the Direction of Bias
	References
Chapter 6: Misclassification
	Introduction
	Definitions and Terms
		Differential vs. Nondifferential Misclassification
		Dependent vs. Independent Misclassification
		Directed Acyclic Graphs and Misclassification
		Calculating Classification Bias Parameters from Validation Data
		Sources of Data
	Bias Analysis of Exposure Misclassification
		Bias-Adjusting for Exposure Misclassification Using Sensitivity and Specificity: Nondifferential and Independent Errors
		Bias-Adjusting for Exposure Misclassification Using Predictive Values
		Bias-Adjustment for Nondifferential Outcome Misclassification Using Positive Predictive Values for the Risk Ratio Measure of A...
		Bias-Adjustments Using Sensitivity and Specificity: Differential Independent Errors
		Bias-Adjustments Using Sensitivity and Specificity: Internal Validation Data
		Overreliance on Nondifferential Misclassification Biasing Toward the Null
	Disease Misclassification
		Bias-Adjustments with Sensitivity and Specificity: Nondifferential and Independent Errors
		Disease Misclassification in Case-Control Studies
		Overreliance on Nondifferential Misclassification Biasing Toward the Null
	Covariate Misclassification
		Bias-Adjustments with Sensitivity and Specificity: Nondifferential and Differential Misclassification with Independent Errors
		Overreliance on Nondifferential Misclassification of Covariates Biasing Toward the Null
	Dependent Misclassification
	Matrix Method for Misclassification Adjustment
	Multidimensional Bias Analysis for Misclassification
	Limitations
		Negative Expected Cell Frequencies
		Other Considerations
	References
Chapter 7: Preparing for Probabilistic Bias Analysis
	Introduction
	Preparing for Probabilistic Bias Analysis
	Statistical Software for Probabilistic Bias Analysis
	Summary Level Versus Record Level Probabilistic Bias Analysis
	Describing Uncertainty in the Bias Parameters
	Probability Distributions
		Uniform Distribution
		Generalized Method for Sampling from Distributions
		Trapezoidal Distribution
		Triangular Distribution
		Normal Distribution
		Beta Distribution
		Bernoulli and Binomial Distributions
		Other Probability Distributions
	Sensitivity to Chosen Distributions
	Correlated Distributions
	Conclusions
	References
Chapter 8: Probabilistic Bias Analysis for Simulation of Summary Level Data
	Introduction
	Analytic Approach for Summary Level Probabilistic Bias Analysis
	Exposure Misclassification Implementation
		Step 1: Identify the Source of Bias
		Step 2: Select the Bias Parameters
		Step 3: Assign Probability Distributions to Each Bias Parameter
		Step 4: Use Simple Bias Analysis Methods to Incorporate Uncertainty in the Bias Parameters and Random Error
			Step 4a: Sample from the Bias Parameter Distributions
			Step 4b: Generate Bias-Adjusted Data Using Simple Bias Analysis Methods and the Sampled Bias Parameters
			Step 4c: Incorporate Conventional Random Error by Sampling Summary Statistics
			Step 4c (Alternate): Resample the Prevalence of Misclassification Adjusted Exposure
		Step 5: Save the Bias-Adjusted Estimate and Repeat Steps 4a-c
		Step 6: Summarize the Bias-Adjusted Estimates with a Frequency Distribution that Yields a Central Tendency and Simulation Inte...
	Misclassification Implementation: Predictive Values
	Misclassification Implementation: Predictive Values - Alternative
	Misclassification of Outcomes and Confounders
	Uncontrolled Confounding Implementation
		Step 1: Identify the Source of Bias
		Step 2: Identify the Bias Parameters
		Step 3: Assign Probability Distributions to Each Bias Parameter
		Step 4: Use Simple Bias Analysis Methods to Incorporate Uncertainty in the Bias Parameters and Random Error
			Step 4a: Sample from the Bias Parameter Distributions
			Step 4b: Generate Bias-Adjusted Data Using Simple Bias Analysis Methods and the Sampled Bias Parameters
			Step 4c: Sample the Bias-Adjusted Effect Estimate
		Steps 5 and 6: Resample, Save and Summarize
	Confounding Implementation Alternative: Relative Risk Due to Confounding
		Step 4: Use Simple Bias Analysis Methods to Incorporate Uncertainty in the Bias Parameters and Random Error
			Step 4a: Sample from the Bias Parameter Distributions
			Step 4b: Generate Bias-Adjusted Results Using Simple Bias Analysis Methods and the Sampled Bias Parameters
			Step 4c: Sample the Bias-Adjusted Effect Estimate
		Steps 5 and 6: Resample, Save and Summarize
	Selection Bias Implementation
	An Example of Probabilistic Bias Analysis in the Presence of Substantial Source Population Data
		Step 1: Identify the Source of Bias
		Step 2: Identify the Bias Parameters
		Step 3: Assign Probability Distributions to Each Bias Parameter
		Step 4: Use Simple Bias Analysis Methods to Incorporate Uncertainty in the Bias Parameters and Random Error
			Step 4a: Sample from the Bias Parameter Distributions
			Step 4b: Generate Bias-Adjusted Data Using Simple Bias Analysis Methods and the Sampled Bias Parameters
			Steps 4c: Sample the Bias-Adjusted Effect Estimate
		Steps 5 and 6: Resample, Save and Summarize
	Selection Bias Adjustment Using Selection Probabilities
		Step 1: Identify the Source of Bias
		Step 2: Identify the Bias Parameters
		Step 3: Assign Probability Distributions to Each Bias Parameter
		Step 4: Use Simple Bias Analysis Methods to Incorporate Uncertainty and Random Error
			Step 4a: Sample from the Bias Parameter Distributions
			Step 4b: Generate Bias-Adjusted Data Using Simple Bias Analysis Methods and the Sampled Bias Parameters
			Steps 4c: Sample the Bias-Adjusted Effect Estimate
		Steps 5 and 6: Resample, Save and Summarize
	Computing Issues with Summary Level Probabilistic Bias Analysis
	Bootstrapping
	Impossible Values for Bias Parameters and Model Diagnostic Plots
	Conclusions
	Appendix: Sampling Models for Exposure Misclassification
	References
Chapter 9: Probabilistic Bias Analysis for Simulation of Record-Level Data
	Introduction
	Exposure Misclassification Implementation
		Step 1: Identify the Source of Bias
		Step 2: Select the Bias Parameters
		Step 3: Assign Probability Distributions to Each Bias Parameter
		Step 4: Use Simple Bias Analysis Methods to Incorporate Uncertainty in the Bias Parameters and Random Error
			Step 4a: Randomly Sample from the Bias Parameter Distributions
			Step 4b: Use Simple Bias Analysis Methods and Incorporate Uncertainty and Conventional Random Error
			Step 4c: Sample the Bias-Adjusted Effect Estimate
		Step 5: Save the Bias-Adjusted Estimate and Repeat Steps 4a-c
		Step 6: Summarize the Distribution of Bias-Adjusted Estimates in a Simulation Interval
		Computing Issues with Record-Level Probabilistic Bias Analysis
		Diagnostic Plots
		Misclassification Implementation Alternative: Predictive Values
		Misclassification of Outcomes and Confounders
	Unmeasured Confounding Implementation
		Step 1: Identify the Source of Bias
		Step 2: Identify the Bias Parameters
		Step 3: Assign Probability Distributions to Each of the Bias Parameters
		Step 4: Use Simple Bias Analysis Methods to Incorporate Uncertainty in the Bias Parameters and Random Error
			Step 4a: Randomly Sample from the Bias Parameter Distributions
			Step 4b: Use Simple Bias Analysis Methods and Incorporate Uncertainty and Conventional Random Error
			Step 4c: Sample the Bias-Adjusted Effect Estimate
		Steps 5 and 6: Resample, Save, Summarize
	Selection Bias Implementation
		Step 1: Identify the Source of Bias
		Step 2: Identify the Bias Parameters
		Step 3: Assign Probability Distributions to Each Bias Parameter
		Step 4: Use Simple Bias Analysis Methods to Incorporate Uncertainty in the Bias Parameters and Random Error
			Step 4a: Randomly Sample from the Bias Parameter Distributions
			Step 4b: Use Simple Bias Analysis Methods and Incorporate Uncertainty and Conventional Random Error
			Step 4c: Sample the Bias-Adjusted Effect Estimate
		Steps 5 and 6: Resample, Save, Summarize and Include Random Error
	Selection Bias Implementation: Individual Selection Probabilities
		Step 1: Identify the Source of Bias
		Step 2: Identify the Bias Parameters
		Step 3: Assign Probability Distributions to Each Bias Parameter
		Step 4: Use Simple Bias Analysis Methods to Incorporate Uncertainty in the Bias Parameters and Random Error
			Step 4a: Sample from the Bias Parameter Distributions
			Step 4b: Generate Bias-Adjusted Data Using Simple Bias Analysis Methods and the Sampled Bias Parameters
			Steps 4c: Sample the Bias-Adjusted Effect Estimate
		Steps 5 and 6: Resample, Save and Summarize
	Alternative Methods for Incorporating Random Error in Bias-Adjusted Estimates
	Conclusions
	References
Chapter 10: Direct Bias Modeling and Missing Data Methods for Bias Analysis
	Introduction
	Directly Incorporating Bias into Effect Estimation
		Standard Errors for Misclassification When Sensitivity and Specificity Are Used
		Standard Errors for Misclassification When Predictive Values Are Used
		Predictive Value Weighting
	Classical and Berkson Measurement Error for Continuous Exposures
	Regression Calibration
	Missing Data Methods
		Overview of Missing Data Structures
		Imputation Methods
		Uncontrolled Confounding Example
		Misclassification Example
		Selection Bias Example
		When to Impute, When to Use Bias Analysis
	References
Chapter 11: Bias Analysis Using Bayesian Methods
	Introduction
	An Introduction to Bayesian Inference and Analysis
		Software for Bayesian Analyses
	Bayesian Adjustment for Exposure Misclassification
		Exposure Misclassification Using Sensitivities and Specificities
		Handling Impossible Sensitivity and Specificity Values in Bayesian Bias Analyses
		Exposure Misclassification Using PPV and NPV
	Selection Bias
	Uncontrolled Confounding
	Conclusion
	Appendix: Justification for Probabilistic Bias Analysis for Exposure Misclassification in a Case-Control Study
	References
Chapter 12: Multiple Bias Modeling
	Introduction
	Order of Bias Adjustments
	Multiple Bias Analysis Example
	Serial Multiple Bias Analysis, Simple Methods
		Simple Misclassification Bias Analysis
		Simple Selection Bias Analysis
		Simple Unmeasured Confounder Analysis
	Serial Multiple Bias Analysis, Multidimensional Methods
		Misclassification Scenarios
		Selection Bias Scenarios
		Unmeasured Confounder Scenarios
	Serial Multiple Bias Analysis, Probabilistic Methods
		Probabilistic Misclassification Bias Analysis
		Probabilistic Selection Bias Analysis
		Probabilistic Unmeasured Confounder Bias Analysis
		Interpretation of Probabilistic Multiple Bias Analysis
	References
Chapter 13: Best Practices for Quantitative Bias Analysis
	Introduction
	When Is Bias Analysis Practical and Productive?
		Cases in Which Bias Analysis Is Not Essential
		Cases in Which Bias Analysis Is Advisable
		Cases in Which Bias Analysis Is Arguably Essential
	Which Sources of Bias to Model?
	How Does One Select a Method to Model Biases?
	Implications Regarding Transparency and Credibility
	Using Available Resources Versus Writing a New Model
	Implications of Poor Bias Analyses
	Who Should Be Calling for Bias Analysis?
	Conclusions
	References
Chapter 14: Presentation and Inference
	Introduction
	Presentation
		Methods
		Results
		Availability of Data and Code
	Inference
		Inferential Framework
		Caveats and Cautions
	Utility
	References
Index