Risks of postoperative infections following arthroplasty were similar among patients with rheumatoid arthritis (RA) treated with various different biologics, but were increased for patients using glucocorticoids even in modest doses, a large retrospective study found.
There were no significant differences in the rates of hospitalized infection within 30 days post-surgery or in prosthetic joint infection within 1 year for patients receiving tumor necrosis factor inhibitors or other types of biologics who had hip or knee replacements, according to Michael D. George, MD, of the University of Pennsylvania Perelman School of Medicine in Philadelphia, and colleagues.
In contrast, patients taking prednisone in dosages of 5-10 mg/day had a predicted increased risk of 30-day hospitalized infection of 8.76% (95% CI 7.16%-10.66%), while the predicted increased risk was 13.25% (95% CI 9.72%-17.81%) for more than 10 mg/day compared with a risk of 6.78% for non-use of prednisone, the researchers reported in Annals of Internal Medicine.
Biologics and Infection
Biologics have revolutionized the treatment of RA, and the goals of treatment today are remission and normal quality of life. However, these medications are associated with increased risks of infection, particularly after surgery, which can be problematic in RA patients who have high rates of knee and hip replacements.
But the available biologics differ from one another in target molecules and pathways as well as in mechanisms of action and dosing. Little is known about the comparative differences in infection risks with the individual agents, as well as the influence that combination therapy with methotrexate or prednisone might have.
Accordingly, George and colleagues reviewed the medical records of RA patients in Medicare claims and Truven MarketScan databases from 2006 to 2015 who were being treated with etanercept (Enbrel), infliximab (Remicade), adalimumab (Humira), abatacept (Orencia), tocilizumab (Actemra), or rituximab (Rituxan). Abatacept recipients were considered the reference group, and propensity scores were calculated for each treatment group.
The analysis included 7,138 patients in Medicare who underwent 7,929 arthroplasties (either primary or revision), and 2,773 patients in MarketScan who had 2,994 procedures.
Mean age was 65, and more than 80% were women. Almost 90% of the surgeries were primary.
During the 3 months before surgery, 45.6% of the patients were receiving methotrexate along with a biologic and 43% were on glucocorticoids, at a median dosage of 12.1 mg/day.
Hospitalized infections within 30 days occurred following 717 procedures (9%) in the Medicare cohort and in 140 (4.7%) of the MarketScan group. The most common types of infection were urinary tract infections, skin and soft tissue infections, and pneumonia.
In the Medicare group, the predicted risks for hospitalized infections ranged from 7.94% (95% CI 4.55-11.34) for tocilizumab to 10.13% (95% CI 6.29-13.98) for rituximab, while in MarketScan the predicted risks ranged from 3.11% (95% CI 1.85-4.36) for adalimumab to 5.22 (95% CI 3.86-6.57) for etanercept.
Prosthetic joint infections developed after 192 procedures in the Medicare group, for a 1-year cumulative incidence of 2%, and following 53 procedures in the MarketScan group, for a cumulative incidence of 2%. Non-urinary tract infections developed after 479 procedures (6%) in the Medicare cohort and after 89 (3%) in the MarketScan group. For 30-day readmissions, the rates were 6.2% and 2.5% in the Medicare and MarketScan groups, respectively.
The rates of all these outcomes were similar among the biologics, with rates of 30-day readmissions among the Medicare group ranging from 5.4% for infliximab to 7.1% with abatacept, and from 2.1% with infliximab to 7.6% for rituximab in the MarketScan cohort.
Rates of 1-year prosthetic joint infections also were similar for most of the biologics, but the number of patients receiving rituximab or tocilizumab having this outcome was small and precise estimates could not be made for these agents, the researchers said.
A Strong Signal
Concomitant methotrexate use was not associated with greater risks for any of the outcomes, but a dose-dependent increase was seen for glucocorticoid use. Compared with no prednisone use, the odds ratio for hospitalized infection was 1.32 (95% CI 1.06-0.64) for daily dosages of 5-10 mg and 2.10 (95% CI 1.48-2.98) for 10 mg or more. Rates of prosthetic joint infections by 1 year were numerically higher with a 5-10 mg daily dosage (HR 1.36, 95% CI 0.90-2.04) and significantly higher with doses higher than 10 mg (HR 1.86, 95% CI 1.02-3.37).
“The main message of the study is that glucocorticoids like prednisone seem to be associated with a lot of risk in people having surgery. There was a very strong signal that glucocorticoids were a major risk for a bad outcome,” George told MedPage Today.
With prednisone dosages above 10 mg/day, the predicted risk for non-urinary tract hospitalized infection was 9.41% compared with 4.34% for no prednisone use, and predicted risks for 30-day readmission were 6.79% versus 4.23%.
Recent guidelines from the American College of Rheumatology/American Association of Hip and Knee Surgeons recommend that biologics be withheld for one dosing interval before arthroplasty and that prednisone be avoided in dosages higher than 20 mg/day.
“The results of this study suggest the postoperative risk may be increased even with lower dosages of glucocorticoids (5 to 10 mg/day),” George and co-authors observed.
In an accompanying editorial, Bheeshma Ravi, MD, PhD, and Gillian Hawker, MD, both of the University of Toronto, stated that the study “provides compelling evidence … of the substantial risk for serious postoperative infections associated with glucocorticoid use.”
The study authors agreed, concluding: “Minimizing glucocorticoid exposure before surgery should be a primary focus of perioperative medication management.”
“Medicines like prednisone are among the most commonly used, but sometimes are forgotten when we are thinking about the newer medications like biologics, when glucocorticoids are probably more important,” George said in an interview.
A limitation of the study, the researchers said, was the possibility of residual confounding despite the use of propensity scores.
The study was funded by the Rheumatology Research Foundation, the National Institute of Arthritis and Musculoskeletal and Skin Diseases, the Patient-Centered Outcomes Research Institute, and Bristol-Myers Squibb.
The authors reported financial relationships with Bristol-Myers Squibb, AbbVie, the Veterans Affairs Office of Research and Development, the American Federation for Aging Research, Pfizer, UCB, Eli Lilly, Galapagos, GlaxoSmithKline, Amgen, Corrona, Janssen, Myriad, and Regeneron; several authors were employees of Bristol-Myers Squibb.
Editorialists Ravi and Hawker disclosed no conflicts of interest.