MODEL EXTENDED COX UNTUK MENGATASI NON-PROPORTIONAL HAZARD PADA DATA STUDI KANKER PARU-PARU

Abstract

Lung cancer is a major cause of cancer-related deaths in Indonesia, making it essential to identify factors influencing patient survival. This study aims to analyze lung cancer patient survival using the Extended Cox Model as an alternative when the Proportional Hazard (PH) assumption is not met. Secondary data from 137 lung cancer patients were analyzed using variables such as type of treatment, treatment history, cancer cell type, Karnofsky score, age, and time of diagnosis. The results showed that only the Karnofsky score was significant in Cox-PH, but the assumption test showed that the cell type and Karnofsky score variables violated PH. Therefore, the analysis was continued with the Extended Cox Model. The final results showed that cell type and Karnofsky score had a significant effect on survival. The Hazard Ratio showed that a certain cell type reduced the risk of death by 22.7%, and an increase of one unit in the Karnofsky score reduced the risk of death by 3.1%. Cancer cell type and Karnofsky score are important factors in the survival of lung cancer patients, and the Extended Cox model has been proven to provide more reliable estimates than Cox-PH when the PH assumption is not being followed.