Based on the historically high rates of fatal PE, orthopedic surgeons specializing in TJR have always feared postoperative PE. Moreover, as PE is a thrombotic phenomenon, there has been a general belief that potent anticoagulation should be used routinely for prophylaxis. This practice is deeply rooted in the orthopaedic community. In the United States, warfarin has been the preferred form of chemoprophylaxis until the introduction of low-molecular-weight heparins in the 1990s and modern factor Xa inhibitors in the last decade. The routine use of potent anticoagulation for prophylaxis has been reinforced since the 1990s by guidelines conceived by a panel of experts and endorsed by the American College of Chest Physicians (Geerts et al. 2008). Guidelines were subsequently adopted and enforced by regulatory bodies like Centers for Medicare and Medicaid Services in the United States and the National Institute for Health and Care Excellence in the United Kingdom.

In addition, as PE has historically been regarded as a “preventable” complication, orthopedic surgeons have been compelled to routinely prescribe anticoagulants following surgery due to the need to comply with the previously mentioned guidelines, fear of litigation, and decreased reimbursement.

With the increased use of potent anticoagulation, the orthopedic community witnessed a raise in the number of postoperative bleeding complications. Local bleeding has resulted in excessive wound drainage, hematoma formation and infection leading to reoperation, and neurologic complications, among others. Systemic complications have resulted in substantial morbidity and rarely in fatal bleeding and heparin-induced thrombocytopenia (HIT). Leaders in the field of joint replacement surgery voiced their concern about the increase in bleeding complications associated with the routine use of anticoagulation for prophylaxis (Callaghan et al. 2005). The concerns of the orthopedic community and awareness of the commercial bias of the majority of authors of guidelines (Johanson 2011) have resulted in the recent reformulation of the ACCP guidelines (Falck-Ytter et al. 2012) and the development of guidelines for thromboprophylaxis by the American Academy of Orthopedic Surgeons (Jacobs et al. 2012) that advocate preoperative risk stratification of the thromboembolic risk (Beksaç et al. 2006) and the use of effective anticoagulation in patients with a higher VTE risk.

In the last four decades, efforts have been made to understand the pathogenesis of thromboembolic disease and perfect its prophylaxis. In our institution, a large body or clinical and applied research has been built since the 1970s (Salvati et al. 2007). The basic and applied investigations hereby discussed provided the basis for our “multimodal approach for thromboprophylaxis.” The multimodal approach emphasizes nonpharmacological gestures to address the pillars of the Virchow’s triad (Virchow 1856) and favors the use of safe, inexpensive aspirin for postoperative chemoprophylaxis in the majority of patients (approximately 80 %).

In the early 1970s, we observed that patients undergoing a THR had a more pronounced drop in the blood levels of antithrombin III than in patients undergoing general surgery. This study suggested that an intense activation of the clotting cascade occurred during THR. However, at that time, we were unable to determine the exact time of maximal thrombogenic activity (Gitel et al. 1979).

In the early 1990s, and with the advent of faster markers of thrombosis (Dahl 1997; Sharrock et al. 1995b), we demonstrated that thrombosis is strongly activated when the femoral canal is invaded and progressively increased with rasping and cementation, peaking at the time of femoral stem insertion (Sharrock et al. 1995b). The activation of the clotting cascade is triggered by intramedullary procoagulants that are forced from the femoral canal into the venous circulation during the femoral work. The effect is more pronounced with cemented fixation than with cementless fixation as a result of greater extravasation of intramedullary contents to the venous circulation caused by cement pressurization (Sharrock et al. 1995b).

The thrombogenic stimulus at the time of the femoral work is further compounded by venous stasis: The flow in the femoral vein is obstructed at the same time when the extreme limb position needed for these steps (flexion, abduction, and internal rotation in the posterolateral approach) (Binns and Pho 1990; Plánes et al. 1990; Stamatakis et al. 1977) kinks the femoral vein.

In order to obliterate thrombogenesis during the femoral work, we studied and implemented the administration of a bolus of intravenous unfractionated heparin (10–15 U/kg) immediately before femoral canal preparation. Our own investigations in the 1990s demonstrated that heparin effectively suppresses intraoperative fibrin formation (Sharrock et al. 1995b, 1999).

Based on the results of the investigations performed in our and other institutions, the senior authors conceived a series of measures that are implemented by internists, anesthesiologists, surgeons, nurses, and physical therapists. Such measures are implemented before, during, and after surgery and are independent of the type for postoperative chemoprophylaxis. The measures aim at combating the pillars of Virchow’s triad and when implemented together are known as “multimodal thromboprophylaxis.” The measures include preoperative VTE risk stratification and discontinuation of procoagulant medications, the use of regional anesthesia (a bolus of intraoperative heparin is used before the femoral work in THR patients), expeditious surgery, postoperative use of pneumatic compression devices and rapid mobilization. When all these mentioned measures are observed, the use of aggressive postoperative anticoagulation becomes unnecessary. In fact, aspirin is our preferred form of chemoprophylaxis in the majority of patients with a low risk of VTE.

The most recognized clinical risk factors for VTE after TJA, in order of importance, include the diagnosis of hip fracture, malignancy, particularly if associated with chemotherapy, antiphospholipid syndrome, immobility or reduced mobility, history of VTE, tamoxifen and raloxifene therapy, American Society of Anesthesiologists (ASA) Physical Status Classification greater than 3, oral contraceptives, estrogen replacement therapy, stroke, atherosclerosis, advanced age, and obesity. The following risk factors are controversial: diabetes mellitus, certain cardiovascular conditions (congestive heart disease and atrial fibrillation), varicose veins, and smoking (Beksaç et al. 2006).

However, we observed that 50 % of patients who develop VTE after THR have no clinical predisposing factors. In a matched, controlled study, we defined that such patients with a low clinical risk of VTE may carry genetic predispositions that increase the risk of VTE. We identified the deficiency of antithrombin III (<75 %) and protein C (<70 %) and prothrombin gene (G20210A) mutation to be strong predisposing factors for postoperative VTE (Salvati et al. 2005). Preoperative genetic screening in conjunction with the identification of the recognized clinical risk factors can help determine the individual risk of postoperative VTE, and postoperative anticoagulation can be selected accordingly (Salvati et al. 2007).

One of the most important gestures that can be done prior to surgery is the discontinuation of procoagulant drugs according to their known ½ life. In an effort to reduce postoperative bleeding complications, antiplatelet medication (clopidogrel, rivaroxaban, warfarin, etc.) should be stopped unless otherwise indicated by the patient’s primary care physician or cardiologist.

Spinal anesthesia reduces blood loss and the prevalence of VTE by 40–50 % in comparison to general anesthesia (Salvati et al. 2007; Davis et al. 1989; Modig 1989; Modig et al. 1983; Prins and Hirsh 1990; Sculco and Ranawat 1975; Sharrock et al. 1993a, 1997; Sharrock and Salvati 1996; Wille-Jorgensen et al. 1989). Epidural anesthesia increases lower extremity blood flow, minimizing both venous stasis and thrombosis especially if low-dose epinephrine is used (Sharrock et al. 1993c; Sharrock and Salvati 1996; Sharrock et al. 1997; Bading et al. 1994; Mineo and Sharrock 1993). We have previously shown that a difference of 10 mmHg (50 versus 60 mmHg) in the mean arterial pressure, achieved through a high lumbar blockade (L1–2 or higher), has a measurable effect on intraoperative blood loss (Sharrock et al. 1993b). In addition, hypotensive anesthesia has been shown in prospective studies to be safe in the elderly patient population with comorbidities (Williams-Russo et al. 1999).

Our current protocol includes an epidural injection in the L1–L2 disk space of 2 % lidocaine with epinephrine (or 0.75 % bupivacaine) to ensure blockade to approximately L4. This is followed by intravenous infusion of epinephrine adjusted to maintain a mean arterial pressure between 50 and 60 mmHg to ensure hemodynamic stability. This approach allows the surgical team to proceed expeditiously on a bloodless field with excellent observation of the anatomic structures and without the delay produced by persistent bleeding. In a previously discussed retrospective study performed at our hospital, we observed that the 30-day mortality from a suspected PE after 15,559 TJRs declined from 0.12 % with the use of general anesthesia to 0.02 % with epidural anesthesia (Sharrock et al. 1995a).

During the early 1970s, we noticed patients undergoing THR were more susceptible to VTE than those who had other surgical procedures. Therefore, we began administering intraoperative unfractionated heparin intravenously and conducted three clinical trials. The heparin was administered at different times and in different doses. Venographic postoperative deep vein thrombosis (DVT) was reduced in all three studies in the heparin group in comparison to the control group (Sharrock et al. 1990; Huo et al. 1992a, b).

In the early 1990s, and with the advent of faster markers of thrombosis, we demonstrated that thrombogenesis is strongly activated as soon as the femoral canal is invaded, increasing progressively with rasping, cementation, and insertion of the femoral stem. Accordingly, we tested the efficacy of a low dose (1,000 U) of unfractionated heparin administered intravenously a few minutes before femoral preparation and found it suppressed fibrin formation during femoral preparation and insertion of the femoral component (Sharrock et al. 1995b). In a subsequent dose-response study, 10 U/kg of unfractionated heparin inhibited fibrin formation, whereas 20 U/kg completely suppressed fibrin formation (Sharrock et al. 1999).

The virtue of this single, small intravenous dose of unfractionated heparin, given a few minutes before femoral work when the thrombogenesis is maximally activated, is its short half-life (±30 min) so the risk of bleeding is minimal. With epidural hypotensive anesthesia, no additional intraoperative or postoperative bleeding is evident (Salvati et al. 2007).

Patients with regional anesthesia do not recover active motion of the lower extremities for a few hours after surgery, which increases venous stasis. Thus, we apply intermittent pneumatic compression (IPC) early in the recovery room. We have shown that the application of IPC immediately after the operation increases the velocity and volume of venous flow, preventing or minimizing the formation and propagation of clots (Westrich et al. 2000). We evaluated the hemodynamic effects of several commercially available IPC devices in a crossover study involving ten patients who underwent TKR (Westrich et al. 1998). Newer pulsatile calf and pulsatile calf-foot IPC with a rapid inflation time produced the greatest increase in peak venous velocity, whereas sequential compression of the calf and thigh showed the greatest increase in venous volume. In a randomized, prospective study, we compared the rate of asymptomatic DVT in patients who were treated with or without IPC after joint replacement surgery. The rate of occlusive thrombi detected using MRV was 2 % (1 of 50) in the IPC group and 10 % (5 of 50) in the control group (p = 0.04) (Ryan et al. 2002).

Active ankle dorsi-plantar flexion increases femoral venous flow by 50 % compared with baseline resting values (Markel et al. 1997). Thus, we strongly encourage such exercise throughout the entire recovery period. The postoperative pain relief provided by the patient-controlled analgesia facilitates early mobilization, which starts with the physical therapist on the day of surgery or in the morning of the first postoperative day.

Aspirin is administered (325 mg twice daily) to patients with no predisposing factors for VTE and who mobilize promptly to provide mild suppression of thrombogenesis in the postoperative period. Aspirin also reduces the risk of heterotopic ossification (Bek et al. 2009).

Patients who are intolerant to aspirin or have a predisposition for VTE are given warfarin, which is initiated on the night of the surgical day aiming at obtaining an INR of 2 for 4–6 weeks.

Patients with an unusually high risk of VTE can be treated with low-molecular-weight heparin (LMWH) or an oral factor Xa inhibitor (Brown and Huo 2013). When the risk of VTE is high, we start both coumadin and LMWH postoperatively. If the epidural catheter is maintained for pain control, we delay the LMWH for 6 hours after its removal. The LMWH is discontinued after 3–4 days, once the patient’s INR is >1.8. If the administration of LMWH is prolonged for 5 days or more, a platelet count is indicated in view of the risk of heparin-induced thrombocytopenia (HIT) (Warkentin and Greinacher 2004). The 2012 ACCP Guidelines recommend against the use of vena cava filters in patients who are not candidates for pharmacologic or mechanical prophylaxis (Falck-Ytter et al. 2012).

In order to determine if multimodal thromboprophylaxis is as efficacious as other regimes that rely only on the routine administration of an anticoagulant, we performed a systematic review of peer-reviewed studies published between 1998 and 2007 that included a 6-week or 3-month incidence of all-cause mortality and symptomatic PE. In an aggregate of over 25,000 patients, we compared 3 different thromboprophylaxis protocols (Sharrock et al. 2008): Group A (15,839 patients) received potent, fast-acting anticoagulants such as LMWH, ximelagatran, fondaparinux, and rivaroxaban; group B (7,193 patients) received multimodal thromboprophylaxis (as outlined above); and group C (5,006 patients) received a slow-acting anticoagulant (warfarin). All-cause mortality was significantly lower in the multimodal group (0.19 %) when compared to the potent anticoagulation group (0.41 %) and the warfarin group (0.40 %). Similarly, the symptomatic PE rate was lower in the multimodal group (0.35 %) when compared to the potent anticoagulation group (0.59 %) and the warfarin group (0.52 %) (Fig. 5.2). We concluded that multimodal thromboprophylaxis is associated with the lowest all-cause mortality and PE rates after joint replacement in the lower extremity and that PE occurs despite the use of potent anticoagulation.