Prediction or explanation: a clinician's guide to logistic regression

Logistic regression is a statistical method used for exploratory and predictive purposes in clinical research. While it is widely used, logistic regression results reported across studies have been inconsistent, particularly in terms of model estimation, validation, and interpretation, even though most studies have clarified their analytical purposes. Reports frequently do not adequately describe how the logistic regression models were evaluated in a purpose-specific manner, particularly regarding underlying assumptions, diagnostic procedures, and performance assessments. Therefore, the interpretability, reproducibility, and clinical applicability of the results presented are frequently limited. We reviewed papers published in the Korean Journal of Anesthesiology (KJA) between 2021 and 2024 that used logistic regression and examined reporting practices, focusing on whether the model evaluation was consistent with the stated analytical purpose. Based on our findings, we then outline the core assumptions required for the appropriate use of logistic regression and introduce key diagnostic tools with practical guidance on their application according to analytical purposes. We also describe statistically robust approaches for reporting the results of logistic regression analyses used for exploratory and predictive purposes, as well as study designs that intentionally incorporate both these aims. With this article, we hope to help researchers establish clear analytical objectives and to adopt systematic reporting practices when using logistic regression, thereby strengthening the basis for research findings to serve as scientifically credible evidence for clinical decision-making.