Mendelian Randomization Methods for Causal Inference Using Genetic Variants

Cover
Half Title
Series Page
Title Page
Copyright Page
Contents
Preface to the second edition
Abbreviations
Notation
I: Understanding and performing Mendelian randomization
	1. Introduction and motivation
		1.1. Shortcomings of classical epidemiology
		1.2. The rise of genetic epidemiology
		1.3. Motivating example: The inflammation hypothesis
		1.4. Other examples of Mendelian randomization
		1.5. Overview of book
		1.6. Summary
	2. What is Mendelian randomization?
		2.1. What is Mendelian randomization?
		2.2. Why use Mendelian randomization?
		2.3. A brief overview of genetics
		2.4. Classification of Mendelian randomization investigations
		2.5. Summary
	3. Assumptions for causal inference
		3.1. Observational and causal relationships
		3.2. Finding a valid instrumental variable
		3.3. Testing for a causal relationship
		3.4. Example: Lp-PLA2 and coronary heart disease
		3.5. Estimating a causal effect
		3.6. Summary
	4. Estimating a causal effect from individual-level data
		4.1. Ratio of coefficients method
		4.2. Two-stage methods
		4.3. Example: Body mass index and smoking intensity
		4.4. Computer implementation
		4.5. Summary
	5. Estimating a causal effect from summarized data
		5.1. Motivating example: interleukin-1 and cardiovascular diseases
		5.2. Inverse-variance weighted method
		5.3. Heterogeneity and pleiotropy
		5.4. Computer implementation
		5.5. Example: Body mass index and smoking intensity reprised
		5.6. Summary
	6. Interpretation of estimates from Mendelian randomization
		6.1. Internal and external validity
		6.2. Comparison of estimates
		6.3. Example: Lipoprotein(a) and coronary heart disease
		6.4. Discussion
		6.5. Recap of examples considered so far
		6.6. Summary
II: Advanced methods for Mendelian randomization
	7. Robust methods using variants from multiple gene regions
		7.1. Motivating example: LDL- and HDL-cholesterol and coronary heart disease
		7.2. Consensus methods
		7.3. Outlier-robust methods
		7.4. Modelling methods
		7.5. Other methods and comparison
		7.6. Example: LDL- and HDL-cholesterol and coronary heart disease reprised
		7.7. Computer implementation
		7.8. Summary
	8. Other statistical issues for Mendelian randomization
		8.1. Weak instrument bias
		8.2. Allele scores
		8.3. Sample overlap
		8.4. Winner's curse
		8.5. Selection and collider bias
		8.6. Covariate adjustment
		8.7. Non-collapsibility
		8.8. Time and time-varying effects
		8.9. Power to detect a causal effect
		8.10. Choosing variants from a single gene region
		8.11. Binary exposure
		8.12. Alternative estimation methods
		8.13. Summary
	9. Extensions to Mendelian randomization
		9.1. Multivariable Mendelian randomization
		9.2. Network Mendelian randomization
		9.3. Non-linear Mendelian randomization
		9.4. Factorial Mendelian randomization
		9.5. Bidirectional Mendelian randomization
		9.6. Mendelian randomization and meta-analysis
		9.7. Summary
	10. How to perform a Mendelian randomization investigation
		10.1. Motivation and scope
		10.2. Data sources
		10.3. Selection of genetic variants
		10.4. Variant harmonization
		10.5. Primary analysis
		10.6. Robust methods for sensitivity analysis
		10.7. Other approaches for sensitivity analysis
		10.8. Data presentation
		10.9. Interpretation
		10.10. Summary
III: Prospects for Mendelian randomization
	11. Future directions
		11.1. GWAS: large numbers of genetic variants
		11.2. -omics: Large numbers of risk factors
		11.3. Hypothesis-free: Large numbers of outcomes
		11.4. Biobanks: Large numbers of participants
		11.5. Clever designs: The role of epidemiologists
		11.6. Conclusion
Bibliography
Index