Reflexión acerca de la regresión logística y las decisiones clínicas

Introduction. The application of risk study through multiple logistic regression necessarily involves making such estimates with the necessary information to the clinician about all the limitations that such work has. Problem. It is possible that works that apply multiple logistic regression do not include the use of the tests of Goodness of Adjustment or the Coefficient of Determination or both. Development: there may be high values of relative risk or odds ratio together with unadjusted models or low values of the coefficient of determination or both at the same time. Therefore, if the clinician does not have this information, in relation to a given entity, it cannot know whether the estimates of elevated risk values are of clinical importance or not and, therefore, the absence of reports of these estimators could imply a therapeutic intervention whose results may not be as expected. Conclusion: Researchers applying the logistic regression model must necessarily report the results of the estimators that give consistency to said model.