PREDICTING HEART FAILURE RECURRENCE AFTER AORTIC VALVE REPLACEMENT USING A COMPETING-RISKS MODEL
Author(s)
Chan V1, Anderson LH2, Martinson M3, Koullick M4, Ruel M5
1University of Ottawa Heart Institute, Ottawa, ON, 2Technomics Research, Ottawa, ON, Canada, 3Technomics Research, LLC, Ottawa, ON, Canada, 4St. Jude Medical, Ottawa, ON, Canada, 5University of Ottawa Heart Institute, Ottawa, ON, Canada
OBJECTIVES: Congestive heart failure (CHF) is a major health burden with an increasing prevalence and incidence. Risk factors associated with recurrent CHF following surgical AVR has been previously described in a multivariate model by Ruel et al. The objective of this study was to re-create the previously described model without relying on the initial patient-level data. METHODS: The semiparametric Cox proportional hazards models described by Ruel et al. was based upon 1563 patients who underwent AVR at the University of Ottawa Heart Institute between 1976 and 2001. Adjusted hazard ratios and mean covariate values were reported. Several distributions, including exponential, Weibull, and competing-risks, were tested to determine, which would better reproduce the recurrent CHF function. RESULTS: All candidate models successfully reproduced the published results by Ruel et al. One model was converted into a customizable excel model. Clinically relevant variables included in the re-created model included valve size, mean transprosthesis gradients, patient age, atrial fibrillation, preoperative NYHA class, body surface area, coronary artery disease, and smoking. After adjustment of patient and valve characteristics, the model predicted freedom from recurrent HF in the Ruel Cohort at 1, 5, 10, and 15 years as 98.4%, 90.8%, 73.2%, and 43.5% compared with published values of 98.6%, 88.6%, 73.9%, and 45.2%, respectively. CONCLUSIONS: The re-created model accurately predicted CHF recurrence. Statistical models validated using large patient cohorts can be useful in population management.
Conference/Value in Health Info
2014-05, ISPOR 2014, Palais des Congres de Montreal
Value in Health, Vol. 17, No. 3 (May 2014)
Code
PRM58
Topic
Methodological & Statistical Research
Topic Subcategory
Modeling and simulation
Disease
Cardiovascular Disorders