Συγκριτική μελέτη επιβίωσης ασθενών με καρδιαγγειακή νόσο με τον γενικό πληθυσμό, με τη χρήση παραμετρικών μοντέλων επιβίωσης για την περιοχή της Κέρκυρας
A comparative study of survival between patients with cardiovascular disease and the general population, from the area of Corfu Island, with the use of parametric survival models

View/ Open
Keywords
Συγκριτική μελέτη ; Επιβίωση ασθενών ; Μοντέλα επιβίωσης ; Κέρκυρα ; Καρδιαγγειακή νόσοςAbstract
Although Cox's proportional hazards (PH) model is widely used in research, proportionality assumption and goodness of fit are not evaluated in most studies. This study aims to compare different parametric survival models in the general population. All data are derived from a subsample of the Seven Countries Study. This study examined the effect of various factors on the 40-year mortality (from all causes and from cardiovascular disease (CVD) separately) in 529 farmers (age 49.8±5.7 years) from Corfu Island. Multivariate survival analysis models (PH Cox and accelerated failure time (AFT) with Exponential, Weibull, Loglogistic, Lognormal distributions) were used. PH Cox models did not fit well based on Cox-Snell residual graphs and were found to be inferior to AFT models. Among all AFT models, Weibull distribution exhibited the most favorable properties, based on Cox-Snell graphs & AIC criterion in both overall mortality, and CVD mortality. Based on the AFT Weibull multivariate model, age decreased the relative probability of total survival, as did the smoking, while physical activity increased it. Age also decreased the CVD survival, as did systolic blood pressure. AFT model is a good alternative when the conditions of the PH Cox model are not met, with the Weibull distribution probably being the best choice.