Lung Cancer Risk and Genetic Factors

Regression Analysis

• Regression analysis is a statistical method used to evaluate risk factors, treatment effects, and outcomes in health research.
• It helps make predictions (e.g., risk of disease).
• It explores relationships between variables (e.g., smoking and lung cancer risk).
• It controls for variables that could bias results (confounding).
• It performs multivariable analysis; going beyond bivariate analysis.
• Examples include predicting patient outcomes and analyzing treatment effects.

Regression Modeling

• Regression models help make predictions (e.g., risk of disease).
• Explores relationships between variables (e.g., smoking/lung cancer risk).
• Controls for variables that could bias results (confounding).
• Regression models go beyond bivariable analysis, accommodating multivariable analysis.
• These models are commonly used to evaluate risk factors, treatment effects, and outcomes in healthcare research.

Variables in Regression Models

• Independent Variable (exposure/treatment): Predicts the value of the dependent variable; other terms include exposure, main predictor, or main risk factor.
• Dependent Variable (outcome/endpoint): Represents the outcome being studied, and is predicted by the independent variable.
• Covariates (confounders/third variables): Other terms include confounders or third variables. They are controlled for when assessing the effect of an independent variable on a dependent variable.

Types of Regression Models

• Linear Regression: Used for continuous outcomes (ratio or interval variables). Examples include predicting systolic blood pressure.
• Logistic Regression: Used for binary (binomial) outcomes. Examples include the probability of having a disease.
• Survival Analysis and Cox Proportional Hazard Models: Used for time-to-event outcomes. Examples include time until recovery or death.

Simple Linear Regression

• Equation: Y = β₀ + β₁X + ε
• Y: Outcome variable (e.g., weight)
• X: Predictor variable (e.g., exercise hours)
• β₀ (α): Intercept/constant (where line crosses Y-axis)
• β₁: Slope (change in Y per unit X)

Multiple Linear Regression

• More accurate models, adjusting for confounding variables (e.g., weight predicted by exercise and diet).
• Model Equation: Y = β₀ + β₁X₁ + β₂X₂ + ... + ε
• Each predictor has its own slope (β).

Simple Logistic Regression

• Used for outcomes like disease/no disease, with an S-shaped distribution.
• Commonly used in case-control studies.
• Odds Ratio is a measure of association.
• Model Equation: logit(p) or log(p/1-p) = β₀ + β₁X₁
• Where p is the probability of the outcome. Exponentiate of β₁ is the odds ratio.

Multiple Logistic Regression

• Model Equation logit(p) = β₀ + β₁X₁ + β₂X₂ +...
• Useful for controlling for multiple variables when exploring associations, for example, between smoking and lung cancer, controlling for exercise and family history.

Time-to-Event Outcomes

• Time-to-event outcomes measure the time until a specific event (e.g., death, disease progression, recovery). Events may or may not occur during the study period.
• Often analyzed using survival analysis methods (e.g., Kaplan-Meier, Cox proportional hazards model).
• Kaplan-Meier curves visualize survival differences between groups.

Cox Proportional Hazards Regression

• Analyzes time-to-event outcomes, such as overall survival or progression-free survival.
• Simple Cox PH model: Ln(h(t)/h₀(t)) = β₁X₁
• Multiple Cox PH model: Ln(h(t)/h₀(t)) = β₁X₁ + β₂X₂ + β₃X₃ +...
• Hazard Ratio (HR) refers to the risk of an event occurring at any time point, and it can be obtained from exponentiating the β coefficients.

Why Multiple Regression Models?

• Health outcomes are rarely influenced by a single factor.
• Multiple regression allows for adjusting confounding variables; isolating the effect of each predictor.
• For example, when studying the impact of exercise on heart health, variables like age, diet and smoking need to be included.

What is a Confounder?

• A confounding variable is associated with both the independent and dependent variables and potentially distorts the observed association between them.
• Example: Age is a confounder in a physical activity and heart disease study since older age is related to less activity and higher heart disease risk.
• Confounding bias occurs when the effect of the exposure on the outcome is mixed with the effect of the confounder, potentially overestimating or underestimating the association, or even showing a non-existent association.

Effect Modification

• Effect modification occurs when the strength or direction of the association between an exposure and an outcome changes depending on the level of a third variable.
• Unlike confounders, effect modifiers reveal how an exposure's effect varies across groups.
• Example: In a study on exercise and heart health, gender may be an effect modifier if exercise affects heart health differently in males and females.

Presentation of Results

• Presenting results from regression models often involves unadjusted and adjusted odds ratios (OR).
• Unadjusted ORs show the relationship between an exposure and an outcome without adjusting for other factors.
• Adjusted ORs show the relationship while controlling for other variables; this typically gives a clearer understanding of the association between the exposure and outcome.

3 Features of a Confounder

• A confounder must be associated with the independent variable and the dependent variable.

• It should not be an effect of the exposure itself.

• It should be a separate entity not part of the causal pathway.

This quiz explores the relationship between genetic susceptibility, smoking, and lung cancer risk. It covers concepts like odds ratios, statistical effect modification, and the significance of regression coefficients in the context of lung cancer studies. Test your understanding of these critical epidemiological factors.