Making sense of regression models in clinical research: a guide to interpreting beta coefficients and odds ratios

Regression models play a central role in clinical research by quantifying the relationship between outcomes and explanatory variables. Accurate interpretation of model outputs, such as beta coefficients in linear regression and odds ratios in logistic regression, is critical for drawing valid conclusions. This article focuses on the key principles for interpreting linear and logistic regression results in clinical research. A publicly available heart disease dataset was used to demonstrate this. Linear regression was applied to a continuous outcome (cholesterol), while logistic regression modelled a binary outcome (presence of heart disease). Model building involved purposeful variable selection to account for confounding. Model adequacy and multicollinearity were assessed using goodness-of-fit statistics and variance inflation factors, respectively. Interpretation of results focused on beta coefficients and odds ratios. Regression models are robust methods for analysing clinical data and identifying key predictors. The concepts presented in this manuscript provide a foundational framework for applying and understanding linear and logistic regression in clinical research.