Journal Logo

Commentary and Perspective

Absence of Evidence May Not Mean Evidence of Absence

Commentary on an article by Bernard H. van Duren, BEng, MBChB, DPhil, et al.: “The Effect of Perioperative Biologic Disease-Modifying Anti-Rheumatic Drugs on the Risk of Postoperative Complications. Surgical Site Infection, Delayed Wound Healing, and Disease Flares Following Orthopaedic Surgical Procedures”

Bass, Anne R. MD1,a

Author Information
The Journal of Bone and Joint Surgery: June 15, 2022 - Volume 104 - Issue 12 - p 1129
doi: 10.2106/JBJS.21.01296


Early treatment and the introduction of biologic disease-modifying anti-rheumatic drugs (bDMARDs) have dramatically improved outcomes for patients with rheumatoid arthritis (RA) and other forms of inflammatory arthritis. Nevertheless, patients with inflammatory arthritis comprise 2.3% of those undergoing knee arthroplasty, and these patients are at significantly higher risk for periprosthetic joint infection than patients with osteoarthritis1. Perioperative discontinuation of bDMARDs may be associated with arthritis flares and the need for corticosteroid treatment, whereas continuation of bDMARDs may increase the risk of infection. The 2017 American College of Rheumatology/American Association of Hip and Knee Surgeons guideline for perioperative medication management in patients undergoing arthroplasty recommended discontinuing bDMARDs for 1 dosing interval prior to surgery, but noted that the level of evidence supporting the recommendation was low2.

The systematic literature review and meta-analysis by van Duren et al. updates the literature on this important subject. The study included data from 11 retrospective cohort studies with a total of 7,344 patients, although data regarding delayed wound healing was only available for 2,150 patients and data regarding disease flares, for 75 patients. They showed a significantly lower likelihood of disease flare among patients who continued versus discontinued their bDMARDs (7.32% versus 25.71%, respectively; odds ratio [OR], 0.22; 95% confidence interval [CI], 0.05 to 0.95; p = 0.04), whereas the risk of surgical site infections did not differ significantly (3.06% versus 2.80%; OR, 1.11; 95% CI, 0.82 to 1.49; p = 0.49). These findings contrast with those of prior studies, which showed a higher risk of infection among patients who continued their bDMARDs perioperatively3. Although there were higher rates of delayed wound healing among patients who continued versus discontinued their bDMARDS, the difference was not significant (2.28% versus 0.99%; OR, 2.16; 95% CI, 0.48 to 9.85; p = 0.32).

The results of this analysis differ from previous studies largely because of the inclusion of 2 recently published studies, which showed no difference in the rate of surgical site infections among patients continuing their bDMARDs4,5. Together, these 2 studies contributed >80% of the patients to the meta-analysis by van Duren et al. The authors also limited their inclusion criteria to patients who either continued or discontinued their preoperative use of bDMARDs, which in some cases meant excluding subsets of patients from previously published articles. Some prior reviews have compared the rates of postoperative infection in RA populations either on or not on bDMARDs—an approach that introduces considerable bias, which was avoided by van Duren et al. Typically, patients with RA who are not on bDMARDs have either mild disease or comorbidities that prevent them from using bDMARDs. Often, these patients are instead on low-dose corticosteroids. Because these factors can also impact the risk of postoperative infection, wound healing, and disease flare, they bias the assessment of medication-attributable risk.

So can we conclude from the present study that infection risk is not increased among patients who continue bDMARDs perioperatively? I think not. Although the literature in this area has grown in recent years, it remains limited in scope, and the strength of the evidence is low because it is based exclusively on retrospective cohorts. The current meta-analysis included only 2,385 patients who continued their bDMARDs. Although this is a seemingly large denominator, complications such as surgical site infection and delayed wound healing occur in only a small percentage of patients, necessitating larger numbers to provide reliable estimates of risk. As the authors themselves acknowledge, the underlying literature also lacked important information regarding concomitant use of conventional DMARDs, and the duration of bDMARD discontinuation varied between studies. Methodologically, van Duren et al. utilized a fixed rather than random effects model for analyses in which study heterogeneity was low (I2 < 25%, as it was for both the surgical site infection and disease flare analyses), but given the heterogeneity that is inherent when patients are drawn from multiple retrospective cohorts, the use of fixed effects models may have artificially their narrowed the reported confidence intervals.

This study highlights the difficulty of measuring risk factors for rare events such as hospitalization for surgical site infection in patients with inflammatory arthritis on bDMARDs. Although a prospective trial might, by necessity, be even more limited in size, it would enable inclusion of some surgical site complications that are managed in the outpatient setting and thus missed in administrative data sets. Certainly, the current meta-analysis suggests that there is the clinical equipoise required to justify such a study.


1. Lenguerrand E, Whitehouse MR, Beswick AD, Kunutsor SK, Foguet P, Porter M, Blom AW; National Joint Registry for England, Wales, Northern Ireland and the Isle of Man. Risk factors associated with revision for prosthetic joint infection following knee replacement: an observational cohort study from England and Wales. Lancet Infect Dis. 2019 Jun;19(6):589-600.
2. Goodman SM, Springer B, Guyatt G, Abdel MP, Dasa V, George M, Gewurz-Singer O, Giles JT, Johnson B, Lee S, Mandl LA, Mont MA, Sculco P, Sporer S, Stryker L, Turgunbaev M, Brause B, Chen AF, Gililland J, Goodman M, Hurley-Rosenblatt A, Kirou K, Losina E, MacKenzie R, Michaud K, Mikuls T, Russell L, Sah A, Miller AS, Singh JA, Yates A. 2017 American College of Rheumatology/American Association of Hip and Knee Surgeons Guideline for the Perioperative Management of Antirheumatic Medication in Patients With Rheumatic Diseases Undergoing Elective Total Hip or Total Knee Arthroplasty. J Arthroplasty. 2017 Sep;32(9):2628-38.
3. Ito H, Kojima M, Nishida K, Matsushita I, Kojima T, Nakayama T, Endo H, Hirata S, Kaneko Y, Kawahito Y, Kishimoto M, Seto Y, Kamatani N, Tsutani K, Igarashi A, Hasegawa M, Miyasaka N, Yamanaka H. Postoperative complications in patients with rheumatoid arthritis using a biological agent - A systematic review and meta-analysis. Mod Rheumatol. 2015 Sep;25(5):672-8.
4. George MD, Baker JF, Hsu JY, Wu Q, Xie F, Chen L, Yun H, Curtis JR. Perioperative Timing of Infliximab and the Risk of Serious Infection After Elective Hip and Knee Arthroplasty. Arthritis Care Res (Hoboken). 2017 Dec;69(12):1845-54.
5. George MD, Baker JF, Winthrop K, Alemao E, Chen L, Connolly S, Hsu JY, Simon TA, Wu Q, Xie F, Yang S, Curtis JR. Timing of Abatacept Before Elective Arthroplasty and Risk of Postoperative Outcomes. Arthritis Care Res (Hoboken). 2019 Sep;71(9):1224-33.

Supplemental Digital Content

Copyright © 2022 by The Journal of Bone and Joint Surgery, Incorporated