Discussion
Dr. Michaud says rheumatologists have had years to optimize DMARD therapies based on patients’ preferences and medical histories, comorbidities, risk factors and other considerations, meaning they’ve been able to increasingly refine their prescription strategies. “What that tells me is that when you look at observational data and you see that everybody has a similar risk, it is a similar risk based on real-world ways of prescribing,” he says.
The observational studies, Dr. Kim emphasizes, cannot prove causality, but instead attempted to control for multiple confounders to point out associations. She noted the high confidence intervals in Dr. Michaud’s study, while conceding that her own study’s estimates and conclusions were limited by the “quite small” number of patients initiating triple therapy, even within a very large population-
based insurance claims database.
More broadly, Dr. Michaud agrees the small percentage of RA patients taking triple therapy is complicating the difficult task of conducting properly controlled safety analyses. Confounding by indication can intrude as well: New patients at high risk of infection are less likely to be prescribed a TNFi, while patients averse to taking multiple pills will be less likely to have a triple therapy prescription. Early on, patients with worse disease activity often received TNFi because they weren’t responding well to existing therapies and needed something new; their higher disease activity, though, already put them at higher risk for serious infection.
“It’s incredibly important that we conduct prospective, longitudinal studies looking at the safety of these drugs,” Dr. Michaud says. Patient cohorts can also change fundamentally in their makeup. Twenty years ago, those taking csDMARDs would have been the norm, but such patients are now more of an exception, whether due to preference, comorbidities, age, financial pressure or other factors.
After 10-plus years of data collection and efforts to adjust for confounding factors, seeing “no obvious signals of concern” regarding the risk of the assessed biologics compared with conventional therapies should be reassuring, Dr. Michaud says.
“It doesn’t mean that all of these drugs are safe, but we’ve got a general sense that the overall risks are somewhat comparable,” he says. “We will continue to monitor it, and the doctors will continue to learn what’s better and what’s not for their patients. It is definitely learning as we go.”
Some past observational studies have suggested a higher risk of serious infections among patients taking TNFi’s. Dr. Kim, though, points out that the studies used different reference groups and that relative measures can depend on which patient groups are used for comparison and on how their baseline characteristics are accounted for.
“Getting appropriate controls is by far the most difficult aspect of safety registries,” Dr. Michaud adds. For each patient, the ideal comparison for the therapy of choice might be the first runner-up. But, based on the myriad variables that go into choosing the best options and thanks to the ever-greater variety of options to choose from, the top two choices in any group of 10 patients may be vastly different.
“That’s what makes it so challenging,” Dr. Michaud says. He’s encouraged by the expansion of administrative datasets, but stresses their value also hinges on continued improvement in capturing granular data, such as comorbidities, disease activity and other factors that help drive prescription choices and patient outcomes.
Dr. Kim says her paper is only one part of a wider effort to conduct multiple, real-world, comparative safety studies of various bDMARDs and targeted synthetic DMARDs.
