Cochrane Database Syst Rev. 2026 May 20. 5
CD015096
Ville Ponkilainen,
Teemu V Karjalainen,
Ilari Kuitunen,
Mikko Uimonen,
Renea V Johnston,
Antti Saarinen,
Samuel L Whittle,
Vanessa Glennon,
Liesl Grobler,
Diane Horrigan,
Sheila Cyril,
Rachelle Buchbinder.
RATIONALE: Disease-modifying anti-rheumatic drugs (DMARDs) are the cornerstone of pharmacologic treatment for inflammatory arthritis (rheumatoid arthritis, psoriatic arthritis and axial spondyloarthritis). DMARDs are immunomodulatory drugs which may increase the risk of infection, including surgical site infections; thus, some surgery guidelines recommend continuing some DMARDs and withholding others prior to surgery. On the other hand, discontinuation may result in worsening of the symptoms of the underlying inflammatory arthritis. Little is known about the optimal use of DMARDs during elective surgery.
OBJECTIVES: To assess the benefits and harms of perioperative interruption versus continuation of conventional synthetic (csDMARDs), biologic (bDMARDs) and targeted synthetic disease-modifying anti-rheumatic drugs (tsDMARDs) in people with inflammatory arthritis.
SEARCH METHODS: We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase and two trial registers up to 24 March 2025.
ELIGIBILITY CRITERIA: We included randomised controlled trials (RCTs) comparing temporary discontinuation or a dose reduction of DMARDs with continued medication perioperatively in adults with inflammatory arthritis (rheumatoid arthritis, psoriatic arthritis or axial spondyloarthritis) undergoing surgery (major orthopaedic surgery (including but not limited to joint arthroplasty and spinal fusion, or other)). When no RCTs were available for one or more types of DMARDs, we included observational studies, data from registries and insurance databases.
OUTCOMES: The critical outcomes were flare of the underlying inflammatory disease, postoperative infections, prosthetic joint infections, mean disease activity score, function, total adverse events and serious adverse events. The prespecified primary time point was up to and including four weeks. As no trial reported outcomes before six weeks, we extracted the shortest time point reported for each outcome.
RISK OF BIAS: We assessed risk of bias using the Cochrane risk of bias 1.0 tool.
SYNTHESIS METHODS: Two review authors independently used Cochrane methods for management of included studies. The main comparison was continuation of any DMARD versus discontinuation in the perioperative period. We synthesised effect estimates using a random-effects meta-analysis model.
INCLUDED STUDIES: Three trials (306 participants) met inclusion. All compared perioperative discontinuation with perioperative continuation of csDMARDs in adults with rheumatoid arthritis undergoing elective orthopaedic surgery. Participants were 56.2 years old on average, and 83% of participants were women. In two trials, the discontinuation group stopped csDMARDs (methotrexate) two weeks prior to surgery and continued two weeks after the surgery, and in one trial, the discontinuation group stopped csDMARDs (leflunomide) one week prior to surgery and continued one month after the surgery. All trials were at risk of selection, performance, detection and selective outcome reporting biases. We did not find any trials assessing the perioperative use of (any) DMARDs in adults with rheumatoid arthritis undergoing non-orthopaedic surgery or adults with psoriatic arthritis or axial spondyloarthritis undergoing any surgery. We also did not find any trials assessing the perioperative use of bDMARDs, or tsDMARDs in adults with rheumatoid arthritis undergoing orthopaedic surgery. We identified nine observational studies with b/tsDMARDs in orthopaedic and non-orthopaedic surgery; six included participants with rheumatoid arthritis and the other three included mixed inflammatory arthritis populations. Results from the observational studies are presented narratively.
SYNTHESIS OF RESULTS: As no trial reported outcomes at the designated primary time point of up to four weeks, we report the earliest time point for flare and postoperative infections (which varied across studies) and last follow-up for adverse events, serious adverse events and revision surgery. Low-certainty evidence indicates perioperative discontinuation of csDMARDs may increase the risk of flare, may have little or no effect on the number of people with postoperative infections, and may have little or no effect on the number of people reporting total or serious adverse events compared with perioperative continuation of csDMARDs. Only one trial measured prosthetic joint infections and reported no prosthetic joint infections in either group (very low-certainty evidence). None of the trials reported disease activity scores or function scores at follow-up. At eight months follow-up, 38/104 (36%) participants reported a flare in the discontinuation group compared to 0/120 (0%) in the continuation group (RR 32.99, 95% confidence interval (CI) 4.54 to 239.53, 2 studies, 224 participants, I2 = 0%). At 12 months follow-up, 5/145 (3%) participants in the discontinuation group had postoperative infections compared to 5/161 (3%) in the continuation group (RR 1.00, 95% CI 0.31 to 3.19, 3 studies, 306 participants). For total adverse events, 29/145 (20%) participants in the discontinuation group reported any adverse events compared to 17/161 (10%) in the continuation group (RR 1.89, 95% CI 0.20 to 18.00, 3 studies, 306 participants, I2 = 60%) after eight to 12 months follow-up. For serious adverse events, 9/113 (7%) participants reported events in the discontinuation group compared to 6/129 (5%) participants in the continuation group (RR 1.41, 95% CI 0.50 to 3.93, 2 studies, 242 participants, I2 = 4%) after 12 months follow-up. Evidence from observational cohort studies largely concurred with the findings from RCTs, indicating an increased risk of flare if DMARDs are discontinued, with no apparent reduction in the risk of postoperative infection. While we could not estimate the risk of prosthetic joint infections from trial data, observational data suggests that perioperative discontinuation of DMARDs does not reduce the likelihood of this outcome occurring.
AUTHORS' CONCLUSIONS: Perioperative discontinuation of DMARDs may increase the risk of flare, and may have little or no effect on the number of people with postoperative infections, the number reporting adverse events and serious adverse events. None of the studies reported on mean disease activity or function. The evidence is limited to rheumatoid arthritis and csDMARDs. Observational studies largely support the findings that there may be a risk of flare with discontinuation of csDMARDs in people with other inflammatory arthritis, but the risks associated with perioperative discontinuation of biologic or targeted synthetic DMARDs are less clear. Editorial note: This is a living systematic review. We search for new evidence approximately yearly and update the review when we identify relevant new evidence. Please refer to the Cochrane Database of Systematic Reviews for the current status of this review.
FUNDING: This Cochrane review had no dedicated funding.
REGISTRATION: Protocol (2022). DOI: 10.1002/14651858.CD015096.