Comparison of Surgical Site Infection Risk Between Warfarin, LMWH, and Aspirin for Venous Thromboprophylaxis in TKA or THA: A Systematic Review and Meta-Analysis

Infection following arthroplasty can have devastating effects for the patient and necessitate further surgery. Venous thromboembolism (VTE) prophylaxis is required to minimize the risk of deep venous thrombosis and pulmonary embolism. Anticoagulation has been demonstrated to interfere with wound-healing and increase the risk of infection. We hypothesized that different anticoagulation regimes will have differing effects on rates of periprosthetic joint infection. The aim of this study was to compare the surgical site infection risk between the use of warfarin, low-molecular-weight heparin (LMWH), and aspirin for VTE prophylaxis following total knee or hip arthroplasty. A systematic literature search was conducted in November 2018 using the PubMed, CINAHL, and Cochrane Central Register of Controlled Trials (CENTRAL) databases to identify studies that compared warfarin, LMWH, and/or aspirin with regard to surgical site infection rates following hip or knee arthroplasty. Meta-analyses were performed to compare the infection and VTE risks between groups. Nine articles involving 184,037 patients met the inclusion criteria. Meta-analysis showed that warfarin prophylaxis was associated with a higher risk of deep infection (or infection requiring reoperation) (odds ratio [OR] = 1.929, 95% confidence interval [CI] = 1.197 to 3.109, p = 0.007) and surgical site infection overall (OR = 1.610, 95% CI = 1.028 to 2.522, p = 0.038) compared with aspirin in primary total joint arthroplasty, with similar findings also seen when primary and revision procedures were combined. There was no significant difference in infection risk between warfarin and LMWH and between LMWH and aspirin. There was a nonsignificant trend for VTE risk to be higher with warfarin compared with aspirin therapy for primary procedures (OR = 1.600, 95% CI = 0.875 to 2.926, p = 0.127), and this was significant when both primary and revision cases were included (OR = 2.674, 95% CI = 1.143 to 6.255, p = 0.023). Thes