MODEL REGRESI COX STRATIFIED PADA DATA KETAHANAN

Abstract

Stratified Cox model on the events are not identical is a modification of the Cox Proportional Hazard models when there are individuals who experienced more than one incident. This study aims to form a stratified Cox regression models for repeated occurrences of data are not identical and their application to cases of hemorrhagic stroke disease recurrence and to determine the factors that affect the case. Parameter Estimation in Stratified Cox models using Partial Maximum Likelihood Estimation (MPLE). Stratified Cox model building procedure consists of six stages: (1) identification data, which specify the variables that will be used in the Cox models. (2) Estimated Cox Proportional Hazard model parameters. (3) The test parameters for each variable using the Wald test. (4) Testing Proportional Hazard assumptions. (5) stratification variables. (6) Interpretation Stratified Cox models. This study uses data of patients who experienced a hemorrhagic stroke unspecified with 7 independent variables such as age, sex, blood pressure, blood sugar, triglycerides, cholesterol and replications. Based on the testing parameters obtained three variables that influence such as age, cholesterol levels and repeat. Furthermore, in assuming Proportional Hazard showed that replicates variable Proportional Hazard did not meet the assumptions that need to be stratified. Unspecified hemorrhagic stroke patients aged over 50 years admitted to 3.230 times longer than the patients were under 50 years old. Unspecified hemorrhagic stroke patients with high cholesterol levels are treated 0.182 times faster than patients with normal cholesterol levels. Keywords: Stratified Cox, Cox Proportional Hazard, MPLE, Haemorrhagic Stroke, Recurrent Events