Predictors of Long-Term Survival in Patients with Idiopathic Pulmonary Fibrosis: Data from the IPF-PRO Registry.

PURPOSE: We used data from the IPF-PRO Registry of patients with idiopathic pulmonary fibrosis (IPF) to identify characteristics that predicted survival for a further > 5 years. METHODS: Participants had IPF that was diagnosed or confirmed at the enrolling center in the previous 6 months. Patients were followed prospectively. A Classification And Regression Tree (CART) was used to identify predictors of survival > 5 versus ≤ 5 years following enrollment. The following variables, assessed at enrollment, were considered: age; body mass index (BMI); former smoker; current smoker; time from first imaging evidence, symptoms, or diagnosis of IPF to enrollment; forced vital capacity (FVC) % predicted; diffusing capacity of the lungs for carbon monoxide (DLco) % predicted; antifibrotic drug use; supplemental oxygen use; history of cardiac disease; pulmonary hypertension; COPD/emphysema; and rural location. RESULTS: The analysis cohort comprised 819 patients, of whom 278 (33.9%) survived > 5 years. DLco % predicted, supplemental oxygen use and FVC % predicted were the most important variables for predicting survival > 5 versus ≤ 5 years after enrollment. The importance of these variables (scaled such that the most important had an importance of 100%) was 100%, 78.2% and 74.2%, respectively. The optimism-corrected area under the curve (AUC) of the CART was 0.72, with an accuracy of 0.72. CONCLUSION: Among patients enrolled in the IPF-PRO Registry, a decision tree that included DLco % predicted, oxygen use and FVC % predicted facilitated the prediction of survival > 5 years. Understanding predictors of longer-term survival may facilitate conversations with patients about their prognosis and treatment.

Duke Scholars

Author Jeremy Weber  

Author Laurie D. Snyder Medicine, Pulmonary, Allergy, and Critical Care Medicine

Author Megan Lee Neely Biostatistics & Bioinformatics, Division of Biostatistics