Yang reports no relevant financial disclosures. Please see the study for all other authors’ relevant financial disclosures.
Prediction models for rheumatoid arthritis outcomes, including infection and stroke, demonstrated strong performance and may be viable as a study tool for clinical use, according to data published in Seminars in Arthritis and Rheumatism.
“Although the management and prognosis of RA has improved in recent decades, identification of RA patients at high risk of adverse health outcomes remains a major challenge,” Cynthia Yang, MD, of Erasmus University Medical Center, in Rotterdam, the Netherlands, and co-authors wrote. “The European Alliance of Associations for Rheumatology (EULAR) and the American College of Rheumatology (ACR) recommend initiating methotrexate (MTX) monotherapy (with glucocorticoids) as soon as possible after the diagnosis of RA, making this the most common treatment for RA worldwide.
“MTX treatment implies screening or monitoring efficacy and side-effects, as with most disease modifying antirheumatic drugs (DMARDs),” they added. “Using prediction models to evaluate patient-level risks in RA patients initiating first-line MTX monotherapy could allow clinicians to target those at high risk of adverse health outcomes for increased screening or monitoring throughout the course of treatment.”
To draft and validate prediction models for adverse health outcomes in patients with RA starting first-line methotrexate monotherapy, Yang and colleagues analyzed 15 large-scale claims and electronic health record databases across nine countries and four continents, using the Patient-Level Prediction Framework from the Observational Health Data Sciences and Informatics initiative. Researchers used data from one database for development, and data from the 14 others for external validation.
Patients with RA were included if they were 18 years or older and had at least 365 days of observation before the initiation of methotrexate. Additionally, patients were required to have been diagnosed with RA 5 years prior to starting methotrexate, as well as have no disease-modifying antirheumatic drug use before initialization, no other DMARD use on or within 7 days of initiation, no record of cancer before initiation and no record of other inflammatory arthritis.
The authors targeted outcomes for which patients with RA were at an increased risk, as well as outcomes that can be monitored throughout the course of treatment. The first instance of several events, including leukopenia, pancytopenia and infection, occurring up to 90 days after index were included. Additionally, myocardial infarction and stroke events through 2 years were included, as well as cancer through 5 years. Patients with instances of these complications within 90 days before beginning therapy were excluded.
In all, the model development used data from 21,547 patients with RA, with 131,928 patients contributing to validation. The models for serious infection, myocardial infarction and stroke demonstrated “good discrimination and adequate calibration.” The other outcome models showed “modest internal discrimination,” and were not validated externally.
“The models had particularly good external validation performance on the USA databases and appeared to perform well (with wider confidence intervals) on the international databases too,” Yang and colleagues wrote. “RA patients identified at high risk of serious infection, MI, or stroke could be targeted for increased screening or monitoring throughout the course of treatment, complementary to current screening or monitoring strategies. In this way, the models may enable clinicians to provide better personalized care to RA patients initiating first-line MTX monotherapy.”