The debate over the optimal strategy for venous thromboembolism (VTE) prevention after hip and knee arthroplasties has raged on for more than a decade; however, a dominant strategy has clearly emerged from clinical practice. Aspirin and mechanical prophylaxis, as opposed to anticoagulant prophylaxis, has become the leading strategy for 95 percent of practitioners in the postoperative setting of total hip and knee arthroplasties, according to a survey of hip and knee surgeons in 2021.1
The CRISTAL trial was a multicenter cluster-randomized, crossover, noninferiority trial performed at 31 centers across Australia with the goal of trying to definitively answer the question of anticoagulant versus antiplatelet therapy for prophylaxis after hip and knee arthroplasties. Centers were randomized to either aspirin (100 mg once daily) or enoxaparin (40 mg once daily) for 14 days after knee arthroplasty or for 35 days after hip arthroplasty. The primary outcome was symptomatic VTE within 90 days of surgery (no screening ultrasounds). To power the noninferiority study, the estimated rate of symptomatic VTE within 90 days was assumed to be 2 percent, and the noninferiority margin was set at 1 percent using statistical power of 90 percent and a one-sided α of 0.025. Secondary outcomes included joint-related readmission, joint-related reoperation, major bleeding, and mortality. The study was nested within the Australian Orthopaedic Association National Joint Replacement Registry and was funded by a Medical Research Futures Fund grant by the Australian federal government.
Although no interim analysis was initially planned, a serious adverse event prompted assembly of a data safety monitoring board (DSMB), which on initial review recommended continuation of the study. However, on a subsequent planned review, the DSMB applied the Haybittle-Peto stopping rule with a two-sided α of 0.001 and recommended stopping the trial. A total of 9,711 patients were included, and 9,203 were analyzed, of whom 5,416 received aspirin and 3,787 received enoxaparin. The median age of patients receiving aspirin and enoxaparin was 67 and 68 years, respectively; more than half of the patients were obese (body mass index > 30). Most procedures were unilateral total knee arthroplasties (52.4%), followed by unilateral total hip arthroplasties (36.7%), bilateral total knee arthroplasties (9.6%), and bilateral total hip arthroplasties (1.4%). Patients with a history of VTE were included in the study — 5.2 percent of patients in the aspirin group and 6.3 percent of patients in the enoxaparin group.
The primary outcome of symptomatic VTE occurred in 187 patients treated with aspirin (3.45%) and 69 patients treated with enoxaparin (1.82%), with an absolute difference of 1.97 percent (95% CI, 0.54%-3.41%). Aspirin did not meet the criterion for noninferiority and instead, enoxaparin demonstrated superiority (p=0.007). There were no significant differences in the secondary outcomes between aspirin or enoxaparin for major bleeding (0.31% vs. 0.40%), joint-related reoperation (2.1% vs. 1.9%), or joint-related readmission (2.4% vs. 2.3%). Audited adherence to the prophylactic regimen used was 85 percent for aspirin and 86 percent for enoxaparin. Although the occurrence of pulmonary emboli (PE; 1.1% of aspirin-treated vs. 0.6% of enoxaparin-treated patients) was numerically lower with enoxaparin, the outcomes were predominately driven by symptomatic below-knee deep vein thrombosis (DVT), with a difference of 1.49 percent favoring enoxaparin (p=0.004). The rate of proximal DVT was the same in each group (0.2%).
In Brief
The CRISTAL trial is a very well designed and conducted cluster randomized controlled trial across a broad spectrum of practices throughout Australia, which has one of the most sophisticated and respected total joint registries in the world. Much of the data that surgeons rely upon for VTE prevention are otherwise observational and retrospective. Two notable exceptions are EPCAT I and II, which both used hybrid anticoagulation/antiplatelet approaches where all patients started with anticoagulants for five to 10 days followed by either aspirin or continued anticoagulant use. The EPCAT study published in 2013 compared dalteparin to aspirin and concluded that aspirin was noninferior for the outcome of symptomatic proximal DVT or PE within 90 days.2 This study used an initial 10 days of dalteparin after unilateral hip arthroplasties in all patients, followed by randomization to 28 days of aspirin or dalteparin thereafter. A noninferiority margin of 2 percent was used, enrolling a total of 778 patients after being terminated early. The design of the initial lead-in period of dalteparin in the EPCAT study is similar to the design the EPCAT II study, which used rivaroxaban 10 mg for the first five days followed by either ongoing use of rivaroxaban or aspirin in hip and knee arthroplasties.3 The EPCAT II study indicated that there was no difference when the extended use of aspirin was compared to rivaroxaban for either VTE or major bleeding outcomes.
How do we interpret the results of the CRISTAL trial in the setting of the previous EPCAT studies? CRISTAL provides important information as it used a monotherapy approach, and demonstrates earlier-onset symptomatic VTE with aspirin regimens and more frequent symptomatic VTE overall. However, most of the difference was driven by distal DVT, which is less dangerous than PE or proximal DVT. The importance of symptomatic distal DVT clearly influences the interpretation and impact of this trial on clinical practice. Although not a part of the primary outcome of overall VTE in the EPCAT II, using an initial five days of rivaroxaban followed by aspirin was associated with a lower rate of distal DVT compared to the CRISTAL study in the aspirin-only group (0.55% vs. 2.4%; Table), indicating that the CRISTAL trial might have achieved similar outcomes if enoxaparin was used only for the first five to 10 days. The importance of distal DVT could be debated with increased vigor if there was a signal for potential harm with enoxaparin prophylaxis compared to aspirin; however, this was not the case, with every single secondary outcome including major bleeding and joint-related reoperations being the same. Unfortunately, important, and more frequent secondary outcomes such as minor bleeding and wound drainage were not reported. Importantly, there have not been mortality differences with different strategies. The overall low rates of PE and proximal DVT after hip and knee arthroplasties are a significant consideration and indicate that perhaps aspirin would be sufficient in a lower-risk population, whereas higher-risk patients would be more likely to benefit from prophylactic anticoagulation. In clinical practice, most patients with a history of VTE would likely get anticoagulant prophylaxis, and indeed in this study, the risk difference between enoxaparin and aspirin in those patients was much higher than in those without VTE history (risk difference, 6% vs. 1.7%; p value for interaction, 0.14). Perhaps we should not be treating all patients the same and maybe both strategies have a role; it seems that we are overdue for a risk adaptive approach to VTE prophylaxis in hip and knee arthroplasties.
Outcomes (symptomatic venous thromboembolism within 90 days) of the EPCAT I and II, and CRISTAL Studies
| . | EPCAT N=785) . | EPCAT II (N=1,904) . | CRISTAL (N=9,203) . | |||
|---|---|---|---|---|---|---|
| Ext Aspirin | Ext Dalteparin | Ext Aspirin | Ext Rivaroxaban | Aspirin | Enoxaparin | |
| PE | 0% | 0.8% | 0.22% | 0.22% | 1.1% | 0.6% |
| Proximal DVT | 0.3% | 0.5% | 0.11% | 0.11% | 0.2% | 0.2% |
| Distal DVT | 0% | 0.5% | 0.55% | 0.89% | 2.4% | 1.2% |
| Major bleeding | 0% | 0.3% | 0.33% | 0.33% | 0.31% | 0.40% |
| . | EPCAT N=785) . | EPCAT II (N=1,904) . | CRISTAL (N=9,203) . | |||
|---|---|---|---|---|---|---|
| Ext Aspirin | Ext Dalteparin | Ext Aspirin | Ext Rivaroxaban | Aspirin | Enoxaparin | |
| PE | 0% | 0.8% | 0.22% | 0.22% | 1.1% | 0.6% |
| Proximal DVT | 0.3% | 0.5% | 0.11% | 0.11% | 0.2% | 0.2% |
| Distal DVT | 0% | 0.5% | 0.55% | 0.89% | 2.4% | 1.2% |
| Major bleeding | 0% | 0.3% | 0.33% | 0.33% | 0.31% | 0.40% |
Abbreviations: DVT, deep vein thrombosis; Ext, extended-use; PE, pulmonary embolism.
Will This Change Practice (Surgeons Perspective)? In brief, the answer is likely to be “no” for most surgeons. There has been a dramatic shift to aspirin monotherapy for VTE prophylaxis in the United States after primary hip and knee arthroplasty and revision surgeries that do not involve mobilization restrictions. Aspirin is readily available, easy to use, inexpensive, does not require monitoring, and has secondary positive effects of arthrofibrosis and heterotopic ossification prevention. It also might be less likely to cause minor bleeding events that potentiate wound complications, infection, arthrofibrosis, and heterotopic ossification. Importantly, patients like using it. By contrast, patients frequently complain about or have difficulty with anticoagulants. They hate injections with low molecular-weight heparin. They hate blood draws and dose adjustments with warfarin. They hate it when direct oral anticoagulants are not covered by insurance. This is especially poignant with the shift to same-day discharge or short length of stay following hip and knee arthroplasties. In our practice, more than 50 percent of primary arthroplasty patients discharge the same day, and these numbers are higher in many private practices.
With this being said, surgeons should take the CRISTAL data seriously as they offer important lessons and considerations. Firstly, aspirin monotherapy seems to increase the risk for symptomatic VTE, and these events tend to occur earlier than in patients treated with anticoagulants. Patients often call in with reports of leg swelling after surgery, and it can be difficult to differentiate normal postoperative swelling from superimposed VTE. Surgeons continuing to use aspirin for VTE prophylaxis should have a lower threshold for evaluating patients showing signs of VTE postoperatively. Second, greater emphasis should be placed on more precise risk stratification. Aspirin is not optimal for every patient, and identifying those most likely to benefit from a hybrid approach or monotherapy with an anticoagulant should be a research and practice priority.
Competing Interests
Dr. Houghton and Dr. Wyles indicated no relevant conflicts of interest.
