Date of Award

Spring 5-2019

Degree Name

Doctor of Public Health (DrPH)



Second Advisor


Third Advisor



Rheumatoid arthritis (RA) is a chronic inflammatory disease of the joints affecting over 1.3 million Americans with annual societal costs estimated at $39.2 billion, rising faster than medical inflation. Therapy with tumor necrosis factor inhibitors (TNFi) has greatly improved the management of patients with rheumatoid RA; however, substantial numbers of patients do not experience an adequate response to these drugs, necessitating a change in treatment regimen. There are two basic approaches for TNFi failure: cycling (switching to a second TNFi) or swapping (to a drug with another mechanism of action) but the choice is controversial due to questions of comparative efficacy and pervasive resource constraints. The initial goal of this study was to follow the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement and systematically review the methodology of decision analyses aimed at determining the most cost-effective sequence of treatment for refractory RA in order to gauge best modeling practices and sources of disagreement in terms of techniques and parameters. The second goal was to analyze the Truven Health Analytics MarketScan® Databases to obtain real-world estimates for cost and drug survival parameters for all ten targeted drugs. Ultimately, the aim was to apply the lessons learned from the systematic review and the estimates calculated from claims data in order to develop an original decision analysis model that will assist physicians and patients in determining the most cost-effective course of care. Seven publication met the criteria for inclusion into the systematic review. They had a largely homogenous model structure and their efficacy estimates were from the same set of randomized clinical trials. Reporting quality was fair and the median ICE for the swapping strategy was $70,332/QALY. The claims analysis demonstrated that 63% of patients cycle to a second TNFi but those who swap to a non-TNFi drug are more likely to persist on treatment, even after controlling for covariates. There were no differences in time to discontinuation for subsequent lines of drugs. While non-TNFi drugs seem to be more effective, they are more costly. Adalimumab and abatacept are the most common second-line TNFi and non-TNFi respectively. Lastly, we built a Markov microsimulation model based on the Truven cohort and conclude that swapping to a non-TNFi is likely to be cost-effective at a $100,000/QALY threshold across a variety of scenarios. Probabilistic sensitivity analysis estimates that the basecase has an 80% probability of being cost-effective at $100,000/QALY. Our results calibrate well with those seen in the systematic review and have the advantage of being based on long-term follow up of a large real-world cohort.