Model Cox Proportional Hazard Pada Kejadian Bersama (Ties) dengan Metode Breslow

(Studi Kasus: Pasien Rawat Inap Demam Berdarah Dengue (DBD) di Rumah Sakit Dirgahayu Samarinda Periode Juli 2016 s.d Juni 2017)

Abstract

Cox proportional hazard model at the tied incident is a modification from Cox model while there are two or more individual that experienced tied. Parameter estimation in procedure formation of cox generally uses maximum partial likelihood estimation (MPLE) that maximised the function of partial likelihood. In incident with ties case, the modification at partial likelihood is done with breslow approach. The data used in this research were Dengue Hemorrhagic Fever patients (DHF) who were hospitalized at Dirgahayu Hospital Samarinda from July 2016 until June 2017. There are 100 patients with 5 independent variables that are suspected to affect the healing of Dengue Hemorrhagic Fever (DHF) patients, namely sex, age, platelet count, hematocrit count, and duration of fever before hospitalized. From the calculation of data used R 2.11.1 software, Cox proportional hazard model using Breslow method, it is obtained that significant variable, are the count of platelets and hematocrit, and the duration of fever before hospitalized, patient who had normal platelet count had a chance of healing 2.359 times faster than patients had a thrombocytopenia (low platelet). For the amount of heamatocrit each patient who had a normal hematocrit had a chance of healing 2.364 times faster than patients who had under-normal hematocrit, and each one day addition to duration of patient's long history fever before hospitalized, decreased the chance of healing by 0.849 times.