USES OF LOGISTIC REGRESSION TO PREDICT THE PROGNOSIS IN ACUTE MYOCARDIAL INFARCTION (AMI)
Vijaya.M.Sorganvi1,Rekha.Udagiri2 and Deepak.Kadeli3
|Received: 28 Feb-20132014 Accepted: 17 Mar-2013|
|Related article at Pubmed, Scholar Google|
Background: Logistic Regression is one of the important methods to perform the statistical model (s) in epidemiological & medical research. It allows the investigators to examine the relationship between binary dependent variable and a set of continuous & discrete independent variables. To understand utility & applicability of this method it was decided to study in hospital outcome (survival & death) of cases admitted due to AMI. Objectives: To develop the prognostic model & compare the result of bivariate and multivariate analysis. Study Design: Retrospective Record based study Study Period: 31ST March 2011 to 1st January 2010. Sample Size: 152 AMI Patients Statistical Analysis: Bivariate analysis, to assess the role of each variable on the outcome variable death or survival chi square test was used. Logistic regression analysis was perform to find the set of best prognostic variable. Odds ratio was obtain by both methods to compare the role of covariates in prognosis of the outcome variable. Result: Logistic Regression Analysis identified age, sex & resident have not shown association with prognosis, time gap in initiation of treatment, & length of hospital stay as most significant variables in the prediction of prognosis after AMI. Conclusion: Logistic regression has failed to fit the data& unable to explain the contribution of explainary variables to the expected level. Hence further investigation is needed before making any final comments on its application.