Statistical Methods
Overview
Statistical methods in genetics provide the tools for measuring associations between genetic exposures and health outcomes. This chapter covers odds ratios and relative risk, the two primary measures used to quantify the strength of association in genetic epidemiology studies. While this is a smaller chapter, the concepts are important for interpreting research literature and for understanding how genetic risk factors are identified and communicated.
Calculating an odds ratio from a 2x2 table, interpreting a relative risk value, and knowing when each measure is appropriate are all core skills for critically evaluating genetic research. This is increasingly important as genetic counselors are expected to incorporate evidence from genome-wide association studies (GWAS), clinical trials, and epidemiologic studies into patient care.
Understanding these measures also connects to clinical communication. When counseling a patient about a genetic risk factor identified through GWAS or family studies, you need to translate statistical measures like odds ratios into language patients can understand. Knowing what an odds ratio of 2.5 actually means, and what it does not mean, is essential for accurate risk communication.
Key Concepts
- Odds ratio (OR): the ratio of the odds of exposure in cases vs. controls; used in case-control studies
- Relative risk (RR): the ratio of the risk of disease in exposed vs. unexposed groups; used in cohort studies
- 2x2 table calculation: setting up and calculating OR and RR from study data
- When to use each measure: OR for case-control studies, RR for cohort studies; OR approximates RR when the disease is rare
- Interpreting values: OR or RR of 1.0 means no association; greater than 1.0 indicates increased risk; less than 1.0 indicates decreased risk (protective)
- Confidence intervals: if the 95% CI includes 1.0, the association is not statistically significant
Association Analyses
Odds Ratios and Relative Risk combines both measures in a single integrated leaf: the mathematical definitions, the study designs each belongs to (OR for case-control, RR for cohort), the construction of 2x2 tables, and the rare-disease assumption under which OR approximates RR. The interpretive layer is just as important as the math: what a given OR or RR value means for an individual patient, the role of confidence intervals in assessing statistical significance, and the limits set by confounding and the association-vs-causation distinction.
This chapter complements Clinical Test Performance. Sensitivity, specificity, and predictive values evaluate individual tests; odds ratios and relative risk evaluate the association between a genetic factor and a disease outcome at the population level.
GWAS and Polygenic Risk Scores
Genome-wide association studies (GWAS) scale single-variant association testing to the entire common-variant genome (~1M independent SNPs), with significance set at P < 5 × 10⁻⁸ after Bonferroni correction. Polygenic risk scores (PRS) aggregate the small effects of many GWAS hits into a single per-individual score for risk stratification. This leaf covers the standard GWAS study design (case-control + SNP array + imputation + logistic regression with PC adjustment), the critical distinction between tag SNPs and causal variants (with HLA-A3 in hemochromatosis as the textbook LD-tagging example), why the MHC dominates autoimmune Manhattan plots, the ancestry-portability barrier to PRS clinical translation, and the common-variant vs rare-variant disease architecture continuum.