Logistic regression analysis
OVERVIEW
Logistic regression analysis is used to predict a dependent binary outcome (yes/no, dead/alive) based on one or more predictor variables
- most commonly used to perform multivariate analysis (controlling for various factors)
- ICU mortality predictions are based on logistic regression analysis
VARIABLES
- Logistic regression can be binomial (yes/no) or multinomial (good/neutral/bad)
- the dependent variably is categorical (yes/no, dead/alive)
- the predictor variables are usually continuous (but can be categorical)
ESTIMATION AND GOODNESS OF FIT
- The regression coefficients are usually estimated using maximum likelihood estimation (there are other methods)
- Goodness of fit can be determined by multiple methods (e.g. Chi-square goodness of fit tests and deviance, Hosmer-Lemeshow tests, Classification table, ROC curves, Logistic regression R2 and Model validation via an outside data set or by splitting a data set)
REGRESSION COEFFICIENTS
- regression coefficients allow the contribution of different predictor variables to be analysed
- likelihood ratio test and the Wald statistic can be used
Critical Care
Compendium
Chris is an Intensivist and ECMO specialist at The Alfred ICU, where he is Deputy Director (Education). He is a Clinical Adjunct Associate Professor at Monash University, the Lead for the Clinician Educator Incubator programme, and a CICM First Part Examiner.
He is an internationally recognised Clinician Educator with a passion for helping clinicians learn and for improving the clinical performance of individuals and collectives. He was one of the founders of the FOAM movement (Free Open-Access Medical education) has been recognised for his contributions to education with awards from ANZICS, ANZAHPE, and ACEM.
His one great achievement is being the father of three amazing children.
On Bluesky, he is @precordialthump.bsky.social and on the site that Elon has screwed up, he is @precordialthump.
| INTENSIVE | RAGE | Resuscitology | SMACC