Date of Award

1-1-2018

Embargo Period

1-1-2020

Document Type

Dissertation

Degree Name

Doctor of Health Administration

College

College of Graduate Studies

First Advisor

Kit Simpson

Second Advisor

Annie Simpson

Third Advisor

Steven Yakubov

Fourth Advisor

James S Zoller

Abstract

Background: A promising safe and effective therapy for stroke prevention in patients with non-valvular atrial fibrillation (NVAF) is left atrial appendage closure (LAAC). However, its cost-effectiveness compared to oral anticoagulation with warfarin, antiplatelet therapy (APT), or novel oral anticoagulants (NOACs) remains a topic of limited research in the US. One recently published network meta-analysis assesses the safety and efficacy of LAAC compared with these other strategies for stroke prevention in patients with NVAF and provides foundational data for analyzing the effectiveness of the interventions. However, this meta-analysis does not provide information on which interventions are most cost-effective, given the costs and practice patterns in the US. Objective: The objective of this study is to examine whether the costs used in a published Markov model are reflective of actual total Medicare payments for LAAC with the Watchman device and potential acute events in patients with NVAF. Methods: The published Markov model by Reddy et al., 2015, was chosen to examine whether the costs used to reach their conclusion are reflective of actual estimated Medicare payments. The cost and clinical inputs, which define the events, are obtained from the Markov model. The comparison cost payment estimates are derived from the 2015 Truven MarketScan database. We extracted hospital admissions by diagnostic related group (DRG) from the 2015 Truven Health MarketScan database. Each DRG event occurring to patients with NVAF was retained, and all associated Medicare (Part B) payments were also extracted for each event. Using the estimated mean costs for all the DRGs used in the published model we compared model cost weights to actual DRG payments by event. This prevents us from performing statistical tests on any observed costs differences. However, since a 10% cost difference is usually tested in a model sensitivity analysis, we reported all cost differences higher than 10% as substantial cost factors, which must be considered in any cost-effectiveness discussion. We present the original model cost weights and the new estimates in Table form. Results: The cost inputs in the published Markov model reviewed are unreasonably low. The mean payments by DRG for the LAAC procedure were 331% higher for Medicare patients > 65 years old and 391% higher for privately insured patients < 65 years old than the cost weights used in the Reddy et al., (2015) Markov model. The mean costs for acute ischemic stoke (the event which LAAC attempts to prevent) for patients with a comorbid code of atrial fibrillation were between 228–248% higher for Medicare patients and 195- 290% higher for privately insured patients than the cost weights used in the model. Conclusions: The DRG cost inputs in the published Markov model for LAAC and potential acute events were consistently much lower than the actual total Medicare payments for the same DRG. Our results suggest the need for a critical look at cost weights used as inputs for economic analyses assessing the cost-effectiveness of stroke prevention interventions, because the costing approaches in two current peer-reviewed modeling studies may be inadequate to serve as models for future studies. Further studies are needed to substantiate LAAC’s long-term cost effectiveness with head-to-head comparisons to NOACs and LAAC.

Rights

All rights reserved. Copyright is held by the author.

Download

Share

COinS