In patients who use VKAs and have elective orthopedic surgery, the risk of perioperative anticoagulation management noncompliance was higher compared to patients who use a DOAC and have elective orthopedic surgery. This is in line with our expectations because perioperative anticoagulation management in DOAC users contains fewer steps than perioperative anticoagulation management in VKA users. Nevertheless, the noncompliance remains high in DOAC users, with 55% of patients having noncompliance with at least one step of perioperative anticoagulation management. Moreover, 28% of patients received conflicting information from various healthcare providers on the timing of interruption.

In VKA users, noncompliance was most prevalent in the preoperative steps of perioperative anticoagulation management. In 68% of these patients, the interruption interval was longer than recommended by the guidelines, and in most patients this was advised by the physician. The longer interruption interval was prompted by the experience that the advised interruption interval for VKAs was too short and resulted in too many cancellations of surgeries. Therefore, an interruption interval of 5 days for acenocoumarol and 7 days for phenprocoumon was used. This interruption interval may increase the risk of thromboembolic events in these patients. However, the risk of these events in patients with atrial fibrillation, which is the most prevalent indication of anticoagulants in our population, is low. In the BRIDGE trial, the risk of thromboembolism was 0.4% in patients with atrial fibrillation who were not bridged with an LMWH, and this risk was not significantly different compared to patients with atrial fibrillation who were bridged with an LMWH [1]. In the BRIDGE trial, patients were using warfarin, and warfarin therapy was stopped 5 days before surgery; the average CHADS2 score in this trial was 2.3. Although not the subject of this trial, it is to be expected that the thromboembolic risk is not substantially increased if VKA therapy is stopped several days earlier. This is especially the case in high-risk patients because anticoagulation therapy is bridged with an LMWH, reducing the thromboembolic risk. In 3% of patients, the interruption interval was shorter than recommended by the guidelines with the potential of a too-high INR before surgery, in most cases because of noncompliance by the patient. Because INR is measured preoperatively, patients with too-high INR levels before surgery are intercepted and treated with vitamin K as needed. Bridging was performed incorrectly in 13% of VKA users, mostly because of incorrect doses prescribed.

In DOAC users, the percentage of patients with a timing of interruption not according to the guidelines was 40% (91 of 229 patients) compared to 71% in VKA users. We noticed that superseded versions of guidelines on the perioperative management of DOACs were still used in the study period, such as the 2012 guidelines ‘guided introduction on new oral anticoagulants’ [25]. In DOAC users, renal function was incorporated in the judgment regardless of whether the interruption interval was in line with the guidelines as presented in Supplementary Table 1. The percentage of patients with an interruption interval shorter than recommended in the guidelines was 7% in DOAC users compared to 3% in VKA users. Because anticoagulation levels are not measured before surgery in DOAC users, unlike VKA users in whom INR is measured, too short of an interruption interval in DOAC users exposes them to a potential increased bleeding risk during surgery. In most cases, the patient executed the interruption interval as instructed by the physician, and the noncompliance was because of the physician. Moreover, we found that in 5% of patients, DOAC therapy was bridged with LMWH therapy—although not recommended in the guidelines—and in 13% of patients, LMWH was used postoperatively together with DOAC therapy erroneously. However, the overall noncompliance is lower in DOAC users than in VKA users; we found that DOAC users are exposed to a higher bleeding risk because of a too-short interruption interval and erroneous use of DOAC and LMWH combined. Contrary to this, we see in the secondary end points that perioperative blood loss during surgery and the administration of blood transfusions were lower in patients using DOACs, suggesting that DOAC use might be safer than VKA use. However, because the numbers were low and differences small, no hard conclusions should be drawn. Implementation of a simplified standardized perioperative management approach, as suggested in the PAUSE study, may result in higher guidelines compliance by the physician [5]. This approach has been implemented in the CHEST 2022 guidelines on perioperative management of antithrombotic therapy [11].

Previous studies on noncompliance with perioperative anticoagulation management guidelines have shown poor compliance, similar to the results found in our study [19,20,21,22]. Moesker et al. described guidelines compliance of 40–81%, depending on the step of the perioperative anticoagulation management [21, 22]. Contrary to our study, they found that the incidence of guidelines noncompliance was highest in the postoperative steps when VKA therapy was reinitiated. An explanation for this could be that we assessed whether the operator made a conscious decision concerning the day of reinitiating VKA therapy after surgery. In the study by Moesker et al., guidelines compliance in 48 DOAC users was analyzed, and the interruption interval was more frequently too long than too short, as compared to the guidelines [20]. For 3 of 34 patients (9%) in whom the interruption interval was analyzable, the interruption interval was too short. This percentage is comparable to the 7% found in our study.

Our study has several shortcomings and limitations. First, we retrospectively analyzed patient dossiers and depended on what was reported in them. The moment of stopping anticoagulation therapy before surgery was not reported for all patients, and assumptions had to be made. If a decision deviated from the guidelines, in most cases no reason was mentioned—and we do not know whether the noncompliance was intentional or erroneous. Moreover, we do not know whether patients complied with the instructions by their physician or reported their use correctly. Second, there were differences in baseline characteristics between the VKA users and DOAC users, which might have influenced the results. Because the percentage of DOAC users increased over time, a change in perioperative anticoagulation management compliance over time might have influenced the difference in guidelines compliance between VKA and DOAC users. However, we have not seen any trend in compliance over time. Third, in the patients using DOACs, more patients preoperatively received tranexaminic acid. During the study period, the guidelines concerning the administration of tranexaminic acid were changed, and the administration was not recommended in the new guidelines. Nevertheless, preoperative administration of tranexaminic acid was higher in patients using DOACs. It is possible that patients using VKAs had more comorbidities, such as valvular heart disease, and therefore more frequently had a contra-indication for the administration of tranexaminic acid. This might have influenced the risk of bleeding complications. Fourth, we analyzed peri- and postoperative bleeding complications as a secondary outcome. However, complications depend on many factors that were not part of the scope of this study, and we therefore cannot draw conclusions on this.

Our study shows that, although guidelines compliance is higher in the group of patients using DOACs, compliance with these guidelines is still limited. Moreover, we identified that one in four patients was given conflicting information about the timing of interruption. Decision support given by the hospital information system could be used to optimize perioperative anticoagulation management guidelines compliance. During the study period, several changes were made to the hospital information system, including mandatory questions about the anticoagulants used and the advised interruption period in the process of ordering a surgery. The orthopedic surgeon must complete these questions and therefore must be aware of whether a patient uses anticoagulants and make an explicit decision on the interruption interval. Because the guidelines recommendations are based on drug use, renal function, and the bleeding risk of the surgery, an algorithm could be developed that gives advice on the interruption interval. In addition, good coordination among healthcare providers is important because many providers are involved in perioperative anticoagulation management.