Navideh Nikmohammadi1, Neda Gilani1, S Morteza Shamshirgaran2, Parvin Sarbakhsh3. 1. Department of Statistics and Epidemiology, Faculty of Health, Tabriz University of Medical Sciences, 5165665931, Tabriz, Iran. 2. Epidemiology Healthy Ageing Research Center, Department of Statistics and Epidemiology, Faculty of Health, Neyshabur University of Medical Sciences, Neyshabur, Iran. 3. Department of Statistics and Epidemiology, Faculty of Health, Tabriz University of Medical Sciences, 5165665931, Tabriz, Iran. p.sarbakhsh@gmail.com.
Abstract
BACKGROUND: Stroke is a global public health challenge. Frailty models can detect and consider the effects of the unknown factors influencing survival along with other known factors. This study aims to evaluate health care providers' effect, along with the demographic and clinical factors, on the stroke patients' survival by using the shared frailty survival models. METHODS: In the 2-year follow-up, a total of 1036 patients with first-ever stroke were recruited from 2013 up to 2015 with census sampling method from two hospitals of Iran, as the health care providers. For model selection, we fitted parametric and semiparametric survival models with parametric shared frailty and used the goodness of fit criteria to compare the models. RESULT: The median follow-up was 730 days. The rate of mortality was 38% during the follow-up period. The Weibull model with gamma frailty had a better fit than the other survival models. The significant variables from the Weibull model were NIHSS score as the stroke severity (score < 5: reference category; scores 5-19: HR = 2.99, p value < 0.001; score ≥ 20: HR = 5.66, p value < 0.001) and age (HR = 1.03, p value < 0.001). Even with the incorporation of the demographic and clinical factors in the survival model, the effect of health care providers as the shared frailty effect was significant (p < 0.001). CONCLUSIONS: Despite considering the known demographic and clinical prognostic factors, health care providers' effect on the patients' survival after stroke was still significant. This may be due to the existing difference between two hospitals in facilities, management, coordination, and efficiency of treatment.
BACKGROUND: Stroke is a global public health challenge. Frailty models can detect and consider the effects of the unknown factors influencing survival along with other known factors. This study aims to evaluate health care providers' effect, along with the demographic and clinical factors, on the stroke patients' survival by using the shared frailty survival models. METHODS: In the 2-year follow-up, a total of 1036 patients with first-ever stroke were recruited from 2013 up to 2015 with census sampling method from two hospitals of Iran, as the health care providers. For model selection, we fitted parametric and semiparametric survival models with parametric shared frailty and used the goodness of fit criteria to compare the models. RESULT: The median follow-up was 730 days. The rate of mortality was 38% during the follow-up period. The Weibull model with gamma frailty had a better fit than the other survival models. The significant variables from the Weibull model were NIHSS score as the stroke severity (score < 5: reference category; scores 5-19: HR = 2.99, p value < 0.001; score ≥ 20: HR = 5.66, p value < 0.001) and age (HR = 1.03, p value < 0.001). Even with the incorporation of the demographic and clinical factors in the survival model, the effect of health care providers as the shared frailty effect was significant (p < 0.001). CONCLUSIONS: Despite considering the known demographic and clinical prognostic factors, health care providers' effect on the patients' survival after stroke was still significant. This may be due to the existing difference between two hospitals in facilities, management, coordination, and efficiency of treatment.