1.6 Prevention of thrombosis in orthopaedic surgery

Despite a huge literature base, thromboprophylaxis in orthopaedics remains controversial. The scale of the problem is disputed and the cost–benefit, risk–benefit, and practicality of any particular protocol is uncertain.

Some surgeons question the very need for prophylaxis, citing very low death rates in series without routine prophylaxis (Box 1.6.1). With changes in orthopaedic practice (quicker surgery, regional anaesthesia, earlier mobilization) the problem of venous thromboembolism (VTE) appears to be receding. The rates of symptomatic VTE without prophylaxis are probably only in the region of 4% after joint replacement without prophylaxis and so there is much being spent, and a risk of bleeding complications being taken, for relatively small gain. No study has shown a longer-term risk of chronic venous insufficiency and pulmonary hypertension in orthopaedic patients who develop postoperative deep vein thrombosis (DVT) or pulmonary embolism (PE). The cost and potential risk of prophylaxis is thereby believed to be unjustified.

Even a low death rate, when multiplied by the very large number of procedures performed, represents a large number. Symptomatic DVT and PE are important outcomes; society will not tolerate avoidable risk and there is an estimated £68 million in litigation costs for VTE. On the balance of probability (the civil test for breach of duty), VTE can be reduced with prophylaxis. The cost of treating a symptomatic VTE is probably higher than routine thromboprophylaxis. A robust study to show, or exclude, chronic venous insufficiency after orthopaedic thrombosis has yet to be performed; until that study is available, the potential for this expensive and morbid complication should be avoided.

There are no studies which show death can be reduced with prophylaxis, or that one modality is any better than another. This is because death is so rare that an adequate sample size cannot be accrued.

There is some evidence that heparins reduce death in surgery. There is compelling evidence that thromboprophylaxis substantially reduces symptomatic VTE. That is a worthwhile effect in itself; the pyramid of VTE means that a proportionate reduction in death can be logically adduced. Most fatal PEs have no warning from a symptomatic DVT; there is very strong evidence that asymptomatic DVT (the usual endpoint in clinical trials) is reduced by suitable prophylaxis.

Unavoidable bleeding is unacceptable in a surgical patient—it carries a morbidity of its own and reflects on the surgeon’s skill. Any chemical carries an intrinsic risk of bleeding. If that risk is similar to the risk of symptomatic VTE that it aims to prevent, nothing has been gained yet a healthcare resource has been spent. Anecdotal experience and many studies do show a higher rate of bleeding with chemicals than placebo.

There is no evidence that postoperative bleeding compromises the longevity of an implant. If a patient develops VTE in the postoperative phase, the required therapeutic anticoagulation invokes an even higher risk of bleeding. Bleeding is correlated with the administration of the chemical too close to surgery; guidelines which delaying the chemical until the risk of bleeding has evaporated, covering the interim with a mechanical method, should allow safe prophylaxis.

It is cheap, easy to administer, and is used by a fair proportion of surgeons. There are studies to show it reduces the frequency of both radiological and symptomatic VTE. It is suitable for extended duration use because it is an oral agent. The American Academy of Orthopaedic Surgeons (AAOS) has recommended aspirin.

It is mainly an antiplatelet rather than antithrombotic agent; the risk reduction achieved is far less than that with low-molecular-weight heparin (LMWH), fondaparinux, or newer oral agents. It carries a risk of gastrointestinal complications and bleeding, particularly in the elderly. The Pulmonary Embolism Prevention (PEP) trial showed no benefit for symptomatic VTE in the subgroup of total hip replacement (THR) and total knee replacement (TKR) patients. The death rate was the same in the placebo and the aspirin group. The AAOS guidelines only addressed PE which revealed a dearth of studies to show any benefit, let alone superiority, for aspirin. The National Institute of Health and Clinical Excellence (NICE) and the American College of Chest Physicians (ACCP) specifically advise against its use.

These methods are inherently safe. Foot pumps are as effective as LMWH in the short-term for THR; intermittent pneumatic compression (IPC) appears effective in THR and TKR. Meta-analysis suggests that even simple TED (thromboembolic deterrent) stockings convey a significant risk reduction.

Compliance by patient and ward staff, especially as the patient becomes increasingly mobile, limits the duration of use to a period that may be shorter than the patient’s risk of thrombosis. VTE is not only promoted by venous stasis—activation of coagulation is an important aetiological factor which may persist beyond the period of venous stasis. There may be a protective effect with chemical prophylaxis from the cardiac mortality that accompanies joint replacement.

Major trauma and orthopaedic surgery predispose to VTE. Virchow’s triad of altered blood components, venous stasis, and endothelial damage are all represented. The soft tissue exposure, bone cutting, and reaming induce systemic hypercoagulability and fibrinolytic inhibition. Patients may be relatively immobile after surgery. During hip replacement, femoral vein blood flow is obstructed by the manoeuvres required to expose the femoral canal and acetabulum. This may not only damage endothelium, in the proximal femoral vein, but also cause venous distension and thus endothelial damage distally, particularly in valve pockets, but also allow the concentration of clotting factors. Anterior subluxation of the knee and the vibration may cause local endothelial damage during knee replacement.

Some orthopaedic procedures, such as minor upper limb surgery, probably carry no risk of thrombosis, whilst other procedures, such as complex lower limb trauma reconstruction or revision hip surgery, carry a particularly high risk.

Individual patients each have their own risk, determined by comorbidity (venous disease, obesity) and particularly by genetic predisposition. A previous PE or DVT is the strongest individual risk factor.

With modern surgical and anaesthetic techniques, but without prophylaxis, the death rate from PE after hip replacement or knee replacement is probably around 0.2%; perhaps slightly higher after hip fracture. With 1.2 million arthroplasties per year in Europe, that equates to 2400 deaths—a huge problem. The death rate after other orthopaedic procedures is unknown, but fatal PE is occasionally seen after lower limb trauma and pelvic trauma; there are case reports after ankle fracture, knee arthroscopy, and even elbow replacement.

Symptomatic event rates are summarized in Table 1.6.1.

Other factors can increase an individual’s personal risk (Table 1.6.2).

The frequency of chronic venous insufficiency, an important longer-term outcome, is unknown. It is likely to be rare after asymptomatic thrombosis (the majority of thrombosis after orthopaedic surgery) but common after symptomatic thrombosis. Chronic pulmonary hypertension is a potential sequelae for those who survive a symptomatic PE.

No study could be large enough to directly show a reduction in fatal PE because it is so rare. Reduction in fatal PE is, of course, not the only purpose in thromboprophylaxis. Symptomatic DVT and PE are a cause of considerable cost (both medical and medico-legal) and are likely to cause longer-term sequelae (chronic venous insufficiency and pulmonary hypertension). Furthermore, fatal PE usually has no warning signs in the leg, so that reduction in asymptomatic DVT is itself an important goal of prophylaxis.

Until the past few years, almost all prophylaxis studies relied on radiological surrogates assuming that a reduction in DVT demonstrable by imaging would correlate with a reduction in symptomatic endpoints. There is evidence to support this. A reduction in symptomatic DVT of 66% with heparin after orthopaedic surgery has been correlated with a similar reduction in fatal PE; a reduction in radiological DVT correlates with a reduction in symptomatic VTE when given for a prolonged time after hip replacement and hip fracture. This correlation is self-evident, given that thromboembolism represents a pyramid of the same disease (Figure 1.6.1).

Most studies refer to hip and knee arthroplasty patients; there are far fewer data on other orthopaedic procedures.

There is a good physiological premise, although only weak circumstantial evidence, to encourage early mobilization.

Anaesthetists are keen to use these techniques (spinal or epidural anaesthesia) as they reduce mortality and enhance perioperative analgesia. Furthermore, neuraxial anaesthesia also reduces the risk of VTE by about 50%, probably due to enhanced blood flow. Because of concerns that a spinal haematoma could develop with concomitant use of chemical prophylaxis and neuraxial anaesthesia, it is prudent to avoid giving neuraxial anaesthesia and chemical prophylaxis too close together. Local anaesthetic guidelines should be followed.

Rough surgical technique will potentiate thromboplastin release. Prolonged torsion of the dislocated hip whilst reaming during hip replacement, or aggressive dorsal retraction of the tibia during knee replacement, inhibit venous return and damage the endothelium.

There is no evidence that tourniquets promote thrombogenesis. The accumulation of clotting factors whilst the tourniquet is inflated is probably balanced by the fibrinolytic and valve-flushing hyperaemia on tourniquet deflation

This means that no prophylaxis is given around the time of surgery; an investigation (e.g. ultrasound or venogram) is performed some time after surgery and therapeutic doses of chemical are then given if the test is positive for a large clot; repeat screening for a small clot. Ultrasound is not too sensitive for small asymptomatic DVT and venography is too invasive. Thirty per cent of THR patients with a negative venogram 10 days after surgery will develop a DVT in the next 3 or 4 weeks. Major and even fatal VTE can occur within the first 10 days. Therapeutic doses of anticoagulant within the first 2 weeks of surgery invite a high complication rate. Prophylaxis is not so expensive whereas routine investigation and treatment of detected thromboses or later-emerging events is. This method of secondary prophylaxis is not recommended.

Mechanical methods are intuitively attractive to orthopaedic surgeons, who have to balance risk and benefit in the perioperative period, as they carry no bleeding risk. The most recent meta-analysis (through the UK NHS Health Technology Assessment process) reviewed 17 Graduate compression stocking (GCS) trials, 22 Intermittent pneumatic compression (IPC) trials, and three foot pump trials. Of these, 14 trials were in hip and knee surgery. The review concluded a 72% odds reduction for mechanical methods alone. All mechanical methods have the disadvantages of expense and compliance. Furthermore, they are not practical for, nor is there evidence for, extended duration prophylaxis.

These are commonly used. To work, they must be properly woven, well-fitted, and remain in place. The evidence is sparse for efficacy after orthopaedic surgery, but a meta-analysis of other surgical studies suggests a modest benefit. There is no clear benefit for above-knee rather than below-knee stockings.

These devices enhance venous flow and also have a fibrinolytic effect due to release of factors from the endothelium from compression. The peak venous flow varies with different devices: frequency of contraction; number of compartments; above- or below-knee design; inflation pressure. The ideal parameters have not been established; however, in general these devices are effective, with an overall risk reduction of about 26%.

These devices empty the venous plexus in the sole of the foot rhythmically, so flushing out the deep leg veins. The foot should be level or slightly dependent so that the plexus can fill (pre-load) prior to it being emptied by the impulse. Stockings probably do not enhance flow. The efficacy will also depend on factors such as the pressure and frequency and the impulses. The risk reduction is probably similar to LMWHs in hip arthroplasty but the evidence for knee arthroplasty is less secure.

These devices (‘umbrellas’) are inserted percutaneously through the femoral vein and are lodged in the inferior vena cava. They cannot prevent thrombosis in the leg, nor do they prevent embolism; they merely catch an embolus and prevents it from reaching the lungs. They are associated with a complication rate which includes death from proximal embolism and venous distension. Their role should be confined to the occasional case where anticoagulation is contraindicated yet the risk of embolism is high. The typical example would be a pelvic fracture transferred from one centre to another who has already developed a leg DVT yet needs a major surgical reconstruction.

Chemical methods such as LMWH, pentasaccharide, warfarin, direct thrombin inhibitors, and factor Xa inhibitors are effective in reducing the risk of VTE. They are generally easy to administer (tablet or injection) and can be used for an extended duration. Relative to the overall cost of surgery, they are fairly inexpensive.

These drugs all carry an inherent risk of bleeding—a risk which properly concerns orthopaedic surgeons.

Aspirin is familiar, cheap, available, and easy to use. However, it only has a weak antithrombotic effect (since it is an antiplatelet agent rather than an anticoagulant). The PEP trial (2000) showed a reduction of about a quarter for symptomatic VTE with aspirin compared with placebo, which is less than would be expected with LMWH (about two-thirds). It risks alternative complications (wound bleeding, transfusion, gastrointestinal bleeding). Although the AAOS recommend its use in THR and TKR, NICE and the two largest evidence-based consensus groups, ACCP and National Institute for Clinical Excellence (ICS), all specifically recommend against its use because if chosen as the sole prophylactic method, then patients are deprived of safer and more effective alternative mechanical and chemical methods. It is not even licensed for thromboprophylaxis in the United Kingdom (Table 1.6.3).

This class of drugs is readily bio-available and has a wide window of safety; therefore monitoring is not required. There are several different types but they are all broadly similar. The drugs are relatively cheap and easy to administer by injection once or twice daily, depending on half-life. LMWHs have been very widely studied and are at least as effective as warfarin, compression devices, and foot pumps. They are more effective than unfractionated heparin and far more effective than placebo. There is little evidence that if used safely they cause bleeding. However, particular care (dose amendment or alterative therapy) should be taken in those with reduced renal function.

Used carefully, it is effective in reducing venographic DVT; fatal PE is exceedingly rare as a reported outcome. It can be continued for as long as the patient is at risk for an extended duration. Meta-analysis shows that it is not as effective as LMWH. Although warfarin is used widely in North America, it is generally regarded as obsolete in Europe (Table 1.6.3).

These synthetic antithrombotic agents precisely inhibit factor Xa. Fondaparinux (Arixtra) is the first of this class to be widely studied and commercially available. It is excreted renally rather than metabolized by the liver. Because of its long half-life (15h) it can be administered by once-daily injection. The clinical trial programme shows that fondaparinux may be more effective than LMWH but may cause more bleeding. However, these differences may be least partly explained by differences in the proximity to surgery when the trial drugs were administered. The drug is not readily reversed and must be used carefully or avoided in those with poor renal function.

These drugs became available in 2008 and may transform thromboprophylaxis. They are given orally and have a broad therapeutic and safety window so that monitoring is not required. They offer a pragmatic solution to those who would otherwise need regular injections (LMWH, fondaparinux) or complex monitoring (warfarin). They are given after surgery and are continued for as long as the patient is at risk of VTE. The drugs are difficult to reverse. Presently, two are available: a. direct thrombin inhibitor (Dabigatran, Boehringer Ingelheim) and an anti-Xa inhibitor (Rivaroxaban, Bayer). Others are due to follow.

Thromboprophylaxis involves a balance of risk and benefit, like most medical interventions. If an effective dose of chemical is administered too close to surgery, bleeding will occur. If the drug is given before surgery, it provides an anticoagulant remedy to peroperative thrombogenic factors (tissue thromboplastins and venous stasis). If the drug is given too long before surgery, the serum levels will be too low for any prophylactic effect; if given too close to surgery then surgical bleeding can be expected. The alterative is to give the drug after surgery. If given too close after surgery, the soft tissues and bone will still be vulnerable to bleeding side effects; if given too late after surgery, then although the risk of surgical bleeding is diminished, the thrombogenic process will be well-established before the drug can act. The drug now has to function as a therapeutic rather than a prophylactic (Figure 1.6.2). The drug has to be given ‘just in time’ to be effective yet to avoid bleeding.

There is a trend for so called ‘stacked modalities’ in which a mechanical method is used to cover the perioperative phase, with a drug then started only when the surgeon and anaesthetist feel that in that individual patient the bleeding risk has decayed.

In the late 20th century, there tended to be a conflict between chemical methods and mechanical methods. This dichotomy was inappropriate, as each has advantages and disadvantages. Furthermore, the obsession with evidence-based medicine tended to allow recommendations only if supported by a particular randomized trial or meta-analysis. However, the randomized trial is only as valid as the unitary hypothesis studied and generalizability is hampered by the exclusion criteria applied to the studied sample. Therefore, clinicians should regard randomized trials and meta-analyses as showing the general direction rather than constraining sensible practice. A pragmatic view would support a combination of mechanical and chemical methods.

For these patients (e.g. with previous thrombosis, strong family history, or prolonged surgery), mechanical and chemical methods can both be used together for as long as possible.

For these patients, a mechanical method should be used until the bleeding risk is lessened; an effective chemical can be safely started and continued for as long as there is a risk of thrombosis.

For these patients (e.g. hip fracture and major trauma), the mechanical method can be started as close to the moment of trauma as possible—in the Emergency Department—and continued until such time after surgery as the clinician feels that a drug can be safely commenced.

In these patients, the cost and risk of prophylaxis may not be justified at all. If prophylaxis is deemed necessary then a mechanical or chemical method could be used, whichever is judged safer and most cost-effective.

If prophylaxis is used, it should be given for an appropriate duration of time. The risk of thrombosis depends on many factors, such as the particular procedure performed, the likelihood of postoperative immobility, and the individual’s inherited propensity to thrombosis. Thromboprophylaxis in the latter part of the 20th century was usually continued only whilst the patient remained in hospital. This was for pragmatic rather than scientific reasons. However, it is now apparent that the risk of thrombosis in hip or knee arthroplasty and hip fracture persists for much longer (Figure 1.6.3). Several sources show that half of symptomatic VTE after knee replacement and two-thirds after hip replacement and hip fracture occur beyond the second week. Randomized controlled trials clearly show that the risk of later symptomatic VTE can be reduced by about two-thirds if prophylaxis is continued for longer. The precise period depends on many factors, but current evidence supports 14 days for knee replacement and 4–5 weeks for hip replacement and hip fracture (Box 1.6.5). This is likely to be a cost-effective approach. Because patients are discharged from hospital sooner and sooner, then a system must be established to continue prophylaxis after discharge. The advent of the new oral agents which do not require monitoring will allow effective and practical extended duration prophylaxis.

There are now several authoritative guidelines available, based on a systematic literature review (Table 1.6.4). Guidelines should ensure that safe and effective prophylaxis is routinely given according to a protocol that has been accepted by the surgeons and anaesthetist, yet properly funded by the health system.

The guidelines have two common threads: to provide effective prophylaxis without causing harm and to use risk assessment. There are differences. In particular the AAOS guidelines only address PE and support the use of aspirin; the other guidelines address the spectrum of VTE and reject aspirin. The International Surgical Thrombosis Forum has suggested a model for the design and interpretation of guidelines in orthopaedics (Box 1.6.6).

In the author’s institution, guidelines have been implemented which are consistent with existing guidelines, to provide effective prophylaxis yet avoid potential bleeding by using mechanical prophylaxis first, then switching to chemical (Table 1.6.5). It is likely that the LMWH will be replaced with an oral anti-Xa or direct thrombin inhibitor.

Like so many other diseases in medicine, there is likely to be a genetic predisposition to both thrombosis and bleeding. Risk factor analysis should provide a clear understanding of those at greatest risk of thrombosis or bleeding; this would allow safe yet effective prophylaxis tailored to the individual’s critical thrombosis period.

The duration of risk for any procedure, let alone individual, is not known. Does the risk rebound when prophylaxis is stopped?

A combined approach seems sensible, but needs to be validated; the most effective, yet safe and affordable, protocol must be established.

A safe and effective oral agent will support simple protocols for extended prophylaxis in those with higher risk of VTE (e.g. day-case knee arthroscopy or plaster casts with additional risk factors).