Rheumatics heart disease using stratified Cox Proportional Hazard model with time-varying covariate effect

Nurhasniza Idham Abu Hasan, Nor Azura Md Ghani, Norazan Mohamed Ramli, Khairul Asri Mohd Ghani, Khairul Izan Mohd Ghani @ Mamat

Research output: Contribution to journalArticle

Abstract

The most common way of analyzing prognostic factors in clinical data is by using Cox Proportional Hazard (PH) model. It is a well-recognized statistical technique for exploring the relationship between the survival of patient and several explanatory variables. The proportionality of the hazards is a critical assumption in the PH analysis and implies that the influence of covariates effect remains similar over time. However, this assumption is often violated. Therefore, different model should be used to deal with non-proportionality of hazards assumption. The aim of this study was to compare the Cox PH Model with non-PH Model as well as to identify the risk of death among Rheumatic Heart Disease (RHD) patients. In this study, we used secondary data in which a retrospective cohort study of 721 RHD patients that were obtained from University Kebangsaan Malaysia (UKM) and National Heart Institute (Institut Jantung Negara, IJN), Malaysia. Both Stratified Cox PH with and without non-PH covariate interaction were performed as non-PH Models. By using Akaike's Information Criterion (AIC) and Deviance, the efficiency of the model performance were compared and then, the most suitable model was determined. Based on these values, the stratified Cox PH Model with no-interaction is the best model for RHD dataset. Five statistically significant prognostic factors that contribute to the risk of death among RHD patients were identified, namely those who diagnosed as emergency status (Opstatus), performed with mitral valve repair alone have Hypertension (HPT), redo operation (Opepisode) and had longer Coronary Pulmonary Bypass (CPB). The non-PH Models fit better than Cox PH Model with respect to the lowest for both deviance and AIC values. This stratified Cox PH Model serves as an alternative approach that can cope with the non-PH situation.

Original languageEnglish
Pages (from-to)9205-9209
Number of pages5
JournalJournal of Engineering and Applied Sciences
Volume12
Issue numberSpecialissue11
DOIs
Publication statusPublished - 1 Jan 2017

Fingerprint

Hazards
Repair

Keywords

  • Cox propohonal hazard model
  • Non-proportional hazard
  • Rheumatic heart disease
  • Situation
  • Stratified Cox propohonal hazard model
  • Time-dependent covariate effect

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Rheumatics heart disease using stratified Cox Proportional Hazard model with time-varying covariate effect. / Hasan, Nurhasniza Idham Abu; Ghani, Nor Azura Md; Ramli, Norazan Mohamed; Ghani, Khairul Asri Mohd; Mohd Ghani @ Mamat, Khairul Izan.

In: Journal of Engineering and Applied Sciences, Vol. 12, No. Specialissue11, 01.01.2017, p. 9205-9209.

Research output: Contribution to journalArticle

Hasan, Nurhasniza Idham Abu ; Ghani, Nor Azura Md ; Ramli, Norazan Mohamed ; Ghani, Khairul Asri Mohd ; Mohd Ghani @ Mamat, Khairul Izan. / Rheumatics heart disease using stratified Cox Proportional Hazard model with time-varying covariate effect. In: Journal of Engineering and Applied Sciences. 2017 ; Vol. 12, No. Specialissue11. pp. 9205-9209.
@article{15b3ffb13326461cbf60987504049798,
title = "Rheumatics heart disease using stratified Cox Proportional Hazard model with time-varying covariate effect",
abstract = "The most common way of analyzing prognostic factors in clinical data is by using Cox Proportional Hazard (PH) model. It is a well-recognized statistical technique for exploring the relationship between the survival of patient and several explanatory variables. The proportionality of the hazards is a critical assumption in the PH analysis and implies that the influence of covariates effect remains similar over time. However, this assumption is often violated. Therefore, different model should be used to deal with non-proportionality of hazards assumption. The aim of this study was to compare the Cox PH Model with non-PH Model as well as to identify the risk of death among Rheumatic Heart Disease (RHD) patients. In this study, we used secondary data in which a retrospective cohort study of 721 RHD patients that were obtained from University Kebangsaan Malaysia (UKM) and National Heart Institute (Institut Jantung Negara, IJN), Malaysia. Both Stratified Cox PH with and without non-PH covariate interaction were performed as non-PH Models. By using Akaike's Information Criterion (AIC) and Deviance, the efficiency of the model performance were compared and then, the most suitable model was determined. Based on these values, the stratified Cox PH Model with no-interaction is the best model for RHD dataset. Five statistically significant prognostic factors that contribute to the risk of death among RHD patients were identified, namely those who diagnosed as emergency status (Opstatus), performed with mitral valve repair alone have Hypertension (HPT), redo operation (Opepisode) and had longer Coronary Pulmonary Bypass (CPB). The non-PH Models fit better than Cox PH Model with respect to the lowest for both deviance and AIC values. This stratified Cox PH Model serves as an alternative approach that can cope with the non-PH situation.",
keywords = "Cox propohonal hazard model, Non-proportional hazard, Rheumatic heart disease, Situation, Stratified Cox propohonal hazard model, Time-dependent covariate effect",
author = "Hasan, {Nurhasniza Idham Abu} and Ghani, {Nor Azura Md} and Ramli, {Norazan Mohamed} and Ghani, {Khairul Asri Mohd} and {Mohd Ghani @ Mamat}, {Khairul Izan}",
year = "2017",
month = "1",
day = "1",
doi = "10.3923/jeasci.2017.9205.9209",
language = "English",
volume = "12",
pages = "9205--9209",
journal = "Journal of Engineering and Applied Sciences",
issn = "1816-949X",
publisher = "Medwell Journals",
number = "Specialissue11",

}

TY - JOUR

T1 - Rheumatics heart disease using stratified Cox Proportional Hazard model with time-varying covariate effect

AU - Hasan, Nurhasniza Idham Abu

AU - Ghani, Nor Azura Md

AU - Ramli, Norazan Mohamed

AU - Ghani, Khairul Asri Mohd

AU - Mohd Ghani @ Mamat, Khairul Izan

PY - 2017/1/1

Y1 - 2017/1/1

N2 - The most common way of analyzing prognostic factors in clinical data is by using Cox Proportional Hazard (PH) model. It is a well-recognized statistical technique for exploring the relationship between the survival of patient and several explanatory variables. The proportionality of the hazards is a critical assumption in the PH analysis and implies that the influence of covariates effect remains similar over time. However, this assumption is often violated. Therefore, different model should be used to deal with non-proportionality of hazards assumption. The aim of this study was to compare the Cox PH Model with non-PH Model as well as to identify the risk of death among Rheumatic Heart Disease (RHD) patients. In this study, we used secondary data in which a retrospective cohort study of 721 RHD patients that were obtained from University Kebangsaan Malaysia (UKM) and National Heart Institute (Institut Jantung Negara, IJN), Malaysia. Both Stratified Cox PH with and without non-PH covariate interaction were performed as non-PH Models. By using Akaike's Information Criterion (AIC) and Deviance, the efficiency of the model performance were compared and then, the most suitable model was determined. Based on these values, the stratified Cox PH Model with no-interaction is the best model for RHD dataset. Five statistically significant prognostic factors that contribute to the risk of death among RHD patients were identified, namely those who diagnosed as emergency status (Opstatus), performed with mitral valve repair alone have Hypertension (HPT), redo operation (Opepisode) and had longer Coronary Pulmonary Bypass (CPB). The non-PH Models fit better than Cox PH Model with respect to the lowest for both deviance and AIC values. This stratified Cox PH Model serves as an alternative approach that can cope with the non-PH situation.

AB - The most common way of analyzing prognostic factors in clinical data is by using Cox Proportional Hazard (PH) model. It is a well-recognized statistical technique for exploring the relationship between the survival of patient and several explanatory variables. The proportionality of the hazards is a critical assumption in the PH analysis and implies that the influence of covariates effect remains similar over time. However, this assumption is often violated. Therefore, different model should be used to deal with non-proportionality of hazards assumption. The aim of this study was to compare the Cox PH Model with non-PH Model as well as to identify the risk of death among Rheumatic Heart Disease (RHD) patients. In this study, we used secondary data in which a retrospective cohort study of 721 RHD patients that were obtained from University Kebangsaan Malaysia (UKM) and National Heart Institute (Institut Jantung Negara, IJN), Malaysia. Both Stratified Cox PH with and without non-PH covariate interaction were performed as non-PH Models. By using Akaike's Information Criterion (AIC) and Deviance, the efficiency of the model performance were compared and then, the most suitable model was determined. Based on these values, the stratified Cox PH Model with no-interaction is the best model for RHD dataset. Five statistically significant prognostic factors that contribute to the risk of death among RHD patients were identified, namely those who diagnosed as emergency status (Opstatus), performed with mitral valve repair alone have Hypertension (HPT), redo operation (Opepisode) and had longer Coronary Pulmonary Bypass (CPB). The non-PH Models fit better than Cox PH Model with respect to the lowest for both deviance and AIC values. This stratified Cox PH Model serves as an alternative approach that can cope with the non-PH situation.

KW - Cox propohonal hazard model

KW - Non-proportional hazard

KW - Rheumatic heart disease

KW - Situation

KW - Stratified Cox propohonal hazard model

KW - Time-dependent covariate effect

UR - http://www.scopus.com/inward/record.url?scp=85047479266&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85047479266&partnerID=8YFLogxK

U2 - 10.3923/jeasci.2017.9205.9209

DO - 10.3923/jeasci.2017.9205.9209

M3 - Article

VL - 12

SP - 9205

EP - 9209

JO - Journal of Engineering and Applied Sciences

JF - Journal of Engineering and Applied Sciences

SN - 1816-949X

IS - Specialissue11

ER -