[Clinical research XX. From clinical judgment to multiple logistic regression model].


The complexity of the causality phenomenon in clinical practice implies that the result of a maneuver is not solely caused by the maneuver, but by the interaction among the maneuver and other baseline factors or variables occurring during the maneuver. This requires methodological designs that allow the evaluation of these variables. When the outcome is a binary variable, we use the multiple logistic regression model (MLRM). This multivariate model is useful when we want to predict or explain, adjusting due to the effect of several risk factors, the effect of a maneuver or exposition over the outcome. In order to perform an MLRM, the outcome or dependent variable must be a binary variable and both categories must mutually exclude each other (i.e. live/death, healthy/ill); on the other hand, independent variables or risk factors may be either qualitative or quantitative. The effect measure obtained from this model is the odds ratio (OR) with 95 % confidence intervals (CI), from which we can estimate the proportion of the outcome's variability explained through the risk factors. For these reasons, the MLRM is used in clinical research, since one of the main objectives in clinical practice comprises the ability to predict or explain an event where different risk or prognostic factors are taken into account.

Cite this paper

@article{BereaBaltierra2014ClinicalRX, title={[Clinical research XX. From clinical judgment to multiple logistic regression model].}, author={Ricardo Berea-Baltierra and Rodolfo Rivas-Ruiz and Marcela P{\'e}rez-Rodr{\'i}guez and Lino Palacios-Cruz and Jorge Moreno and Juan Osvaldo Talavera}, journal={Revista médica del Instituto Mexicano del Seguro Social}, year={2014}, volume={52 2}, pages={192-7} }