Current recommendations for palliative care in COPD patients are not perfect nor precise most because prediction of death is difficult We conducted a retrospective cohort study to develop a predictive index for 1-year mortality in COPD patients From Longitudinal Health Insurance Database 2005 we selected COPD cohort which included 21991 prevalent cases and 12452 incident cases All patients had complete-1 year follow up and the end point was mortality By applying logistic regression model we found risk factor for mortality were male gender old age lower income more hospitalization higher comorbidity score and lower numbers of outpatient department visits We developed prognostic index based on these variables and the c-statistics was 0 82 When applying the index in validation group the model has nearly the same discrimination power (c-statistics = 0 81) Best cut-off point based on Youden index was calculated to identify patients with high risk for death When comparing with actual mortality status out predictive index showed high negative predictive value (95%) but relatively low positive predictive value (25%)
Date of Award | 2021 |
---|
Original language | English |
---|
Supervisor | Chung-Yi Li (Supervisor) |
---|
Development of predicting indexes for 1-year mortality in chronic obstructive pulmonary disease patients
昇翰, 蔡. (Author). 2021
Student thesis: Doctoral Thesis