Claims-Based Analysis of Cost-Effectiveness of Cyclers vs Switchers in Patients With Less-Than-Adequate Tumor Necrosis Factor Inhibitors

Jeffrey Curtis, MD, a rheumatologist from Birmingham, Alabama, discussed the ways in which patients with rheumatoid arthritis (RA) who demonstrate a less-than-adequate response to a tumor necrosis factor inhibitor (TNFi) can be treated with another disease-modifying antirheumatic drug (DMARD), either by changing to another TNFi (“cyclers”) or by changing to a new mechanism of action (MoA) (new MoA “switchers”). Acknowledging that clinicians have many choices in the management of RA, and citing growing analytical evidence on which option might be best physically, the study by Curtis et al focused on 1 recommendation that evaluated what steps to take once a patient fails the TNFi therapy.

The study covered patients from 2010 to 2014, with 8517 patients meeting inclusion criteria. Although he cited several data sources, he relied mainly on Truven Health Analytics Market Scan, with more than 40 million people studied. Inclusion criteria consisted of patients with RA having manifested moderate or high disease activity for at least 3 months. Patients excluded from the study included those with another TNFi therapy for any other disease, as well as those with a history of cancer.

During the study, patients also must have used a TNFi for at least 12 months and then have been part of the study for a subsequent 12 months after starting a new therapy. Among additional requirements, patients could not stop taking the drugs and also could not change to a new MoA DMARD. If the patient was on steroids, he or she could not increase that dosage. If researchers saw any of these issues or other issues, the patient would have been classified as having failed the therapy.

Costs (targeting DMARD/biological treatments) were standardized to December 2015 dollars and total RA-related healthcare costs. The study’s main outcome variable was the cost per “effectively treated” patient in the first year after TNFi cycling or new MoA switching, with treatment effectiveness being defined according to an identified algorithm. Curtis et al examined the average 12-month postindex cost per patient divided by the percent of patients categorized as effectively treated. For example, if main costs per treatment cohort were $50,000, but only 50% of patients were effectively treated, then the cost per effectively treated patient was $100,000.

Patient characteristics included the fact that MoA switchers were slightly older and more likely to have used corticoids, plus they cost more from their health plan.

Concerning treatment effectiveness, 23.3% of cyclers met all the criteria compared with 26% of new MoA switchers. Speaking to why the former group had a higher failure percentage, Curtis et al found that some patients moving to the next therapy just gave up. They also found small but significant differences in other drugs.

Most RA costs are predicated on the cost of the drug treatment. Over the course of the year, the overall cost per effectively treated patient—$165,200 (cyclers) and $126,991 (switchers)—was lower for the switcher cohort. What accounts for the difference, he said, is the adherence issue and not switching to a new MoA.

In summing up the results, Curtis noted that a higher proportion of new MoA switchers were effectively treated, and had significantly lower costs per effectively treatment patient than TNFi cyclers. He did point out, however, that the direction of bias is always against a new drug, so if differences favor the new drug, they are more likely to be real.

Nevertheless, there are limitations to keep in mind, including that this was strictly an observational study, wherein causality could not be claimed. He added that true costs and true differences may be of greater magnitude. Additionally, they did not include clinical outcomes in using the claims-based treatment effectiveness algorithm, meaning the researchers did not know, for example, why a patient stopped using the drug or switched. That information was not captured in the data sourced.

Dr Curtis also acknowledged that the perspective in this analysis is somewhat different from previous studies, in that it examined the cost of medications over time, relative to whether patients were effectively treated.