This is the English translation (or version) of the German guidelines published in "Anaesthesiologie und Intensivmedizin 2007 Oct (48) 124-129.

introduction

Since the publication of the first and second versions of the recommendations on conducting neuraxial regional anaesthesia and thromboembolism prophylaxis/antithrombotic medication by the German Society for Anaesthesiology and Intensive Care [Deutsche Gesellschaft für Anästhesiologie und Intensivmedizin, DGAI] in 1997 and 2003 [1], new anticoagulant agents have been introduced and further findings have been obtained regarding the risk of neuraxial regional anaesthesia with simultaneous thromboembolism prophylaxis. At the same time, more frequent use of coronary stents is increasing the perioperative cardiac risk for patients when treatment with platelet aggregation inhibitors is interrupted perioperatively.

It should be emphasized that due to the rarity of spinal epidural haematomas, recommendations regarding neuraxial regional anaesthetic procedures with simultaneous antithrombotic medication, and medical thromboembolism prophylaxis in particular, are not based on scientific evidence, but rather on case reports of haematomas and on expert opinion. Accordingly, the recommendations are not based on prospective randomized studies, but are mainly based on pharmacokinetics and the recommended dosage intervals for the individual agents concerned.

These national recommendations may, however, have contributed to the avoidance of a similar frequency of complications as that observed in the USA and Sweden.

Risk of spinal epidural haematomas

Spinal epidural haematomas often occur spontaneously – i.e., without a temporal connection with neuraxial regional anaesthesia. The absolute risk of spinal bleeding during simultaneous thromboembolism prophylaxis is not known. Following central nervous blockades, they represent such a rare – although severe – event that their incidence cannot be investigated using prospective randomized studies. Case series published in recent years show, however, that the risk of spinal epidural haemorrhage is possibly much higher than was previously thought.

After a higher dose of low-molecular-weight heparins (enoxaparin 2 x 30 mg/day) was introduced for thromboembolism prophylaxis in the USA, there were increasing reports of spinal epidural haematomas, which led almost without exception to permanent paraplegia and led to a warning issued by the U.S. Food and Drug Administration (FDA). Following this accumulation of cases, the risk of spinal epidural haematomas during simultaneous administration of low-molecular-weight heparins in the USA was calculated as one in 40,800 for spinal anaesthesia, one in 6600 for single-shot epidural anaesthesia, and one in 3100 for catheter epidural anaesthesia [2]. The differences in the incidence of bleeding complications were mainly attributed to different dosages of enoxaparin (in Europe, 1 x 40 mg/day; in the USA, 2 x 30 mg/day) and the lack of recommendations at that time regarding the observance of time intervals between neuraxial puncture/catheter removal and thromboembolism prophylaxis. The recommendations that have been introduced in the USA in the meantime, which are stricter than those in place in Europe, propose that thromboembolism prophylaxis at the dosages of low-molecular-weight heparins customary in the USA should be avoided during the entire period in which epidural catheters are in place, although the increased risk of thrombosis is not adequately taken into account [3]. A retrospective analysis in Sweden came to similar conclusions, despite low dosages of enoxaparin (1 x 40 mg). In the latter study, the risk was found to be one in 18,000 after epidural anaesthesia and one in 156,000 after spinal anaesthesia, with bleeding complications occurring much more rarely after epidural anaesthesia in obstetrics (one in 200,000) than in female orthopaedic patients (one in 3600) [4]. Risk factors for spinal haematoma after neuraxial regional anaesthesia were identified as a lack of guidelines, administration of antithrombotic agents, female sex, and difficult puncture conditions.
However, an increased risk of haemorrhage during simultaneous administration of antithrombotic agents is not a new finding. Vandermeulen et al. reported as early as 1994 [5] that 68% of patients with spinal epidural haematomas had received anticoagulants. Thrombocytopenia or drug-induced platelet function disturbances were demonstrated in 20% of the patients. In 1996, Wulf reported a case analysis with similar results [6]. In relation to its frequency of use, aspirin was rarely mentioned as a risk factor, but fibrinolytic agents and Bechterew disease were frequent. In a more recent analysis of 99 cases, coagulation disturbances were found in 78% of patients [7]. Additional risk factors include bloody or traumatic punctures and anatomic changes (e.g., spina bifida, Bechterew’s disease). Due to the smaller needle diameter, the risk of haemorrhage is lowest in spinal anaesthesia and highest in catheter epidural anaesthesia. Nearly half of all cases of bleeding occur during the removal of an epidural catheter [5], so that this phase must be regarded as critical as catheter placement.

These case reports show that spinal epidural bleeding complications are not restricted to recent medication with low-molecular-weight heparins, but can occur generally with any agent that interferes with haemostasis. Traditional coagulation parameters are not helpful for assessing the bleeding risk during thromboembolism prophylaxis, as these parameters usually remain unchanged. Observing the recommended dosages and time intervals helps in carrying out the puncture as near as possible to trough levels, in general the risk of haemorrhage is lowest the greater the time interval between the puncture and the administration of antithrombotic agents (or the presence of effective drug levels) and the lower the anticoagulant dosage. Accordingly, the recommended time intervals are based on the pharmacokinetics and pharmacodynamics of the individual agents concerned. In addition, the risk of haemorrhage becomes more difficult to assess the larger the number of different antithrombotic agents that are being administered (e.g., low-molecular-weight heparin plus acetylsalicylic acid), so that particular caution is required when there is a combination of several agents or if there are additional risk factors.

Even when time intervals were observed there have been several case reports of spinal epidural haematomas in older patients in recent years, with renal insufficiency being an important risk factor [8,9]. All of the agents used for thromboembolism prophylaxis, with the exception of argatroban, are eliminated renally and accumulate in patients with renal impairment, which often remains unidentified in everyday practice. The recommended time intervals therefore only apply to patients with normal renal function; in those with reduced renal function, either dose adjustment of the antithrombotic agents or longer time intervals are required in order to reduce the risk. Calculation of creatinine clearance may be helpful for risk assessment [10].

Unfractionated heparins

Low-dose thromboembolism prophylaxis does not lead to an increased risk of bleeding in regional anaesthesia procedures, provided that a certain time interval is observed between heparin administration and puncture. For example, during low-dose administration of unfractionated heparins, an interval of 4 hours should be observed between heparin administration (usually 5000 IU s.c.) and epidural puncture or catheter removal, in order to avoid bleeding complications. Any repeat administration of low-dose heparin should then follow at the earliest after 1 hour. With the exception of a platelet count to exclude heparin-induced thrombocytopenia (treatment > 5 days), coagulation analyses are not required during thrombosis prophylaxis with unfractionated heparins.

In contrast to heparinization at prophylactic dosages, an increased risk of bleeding does occur at therapeutic dosages, so that puncture or removal of a catheter should not be carried out during treatment. If, after careful consideration, a regional anaesthesia procedure or the removal of an epidural or spinal catheter has to be carried out, the intravenous heparin administration should be interrupted at least 4 hours earlier and the coagulation parameters – activated partial thromboplastin time (aPTT) and/or activated clotting time (ACT) – must have returned to normal.

• Regional anaesthesia procedures and intraoperative heparinization

  A planned intraoperative heparinization does not necessarily represent a contraindication to a regional anaesthetic procedure. Rao and El-Etr [11] reported in 1981 that in patients undergoing vascular surgery, the risk of haemorrhage after epidural anaesthesia and subsequent heparinization is not increased if the heparinization is carried out at the earliest 1 hour after spinal/epidural puncture and is closely monitored. By contrast, in the context of diagnostic lumbar punctures, a very high rate of paraplegia of 2% occurred when patients received intravenous heparinization less than 1 hour after the puncture with simultaneous administration of acetylsalicylic acid [12].
  
  Removal of the epidural catheter should be carried out at the earliest 4 hours after the end of heparin administration and after normalization of coagulation.
  
  If a bloody puncture occurs in patients in whom intraoperative heparinization is planned, it can be recommended that low-dose anticoagulation (e.g., 5000 IU heparin) should be avoided for 1–2 hours and that intraoperative full heparinization should be avoided for 6–12 hours, with the operation being delayed for at least 12 hours if necessary. Alternatively, to avoid delays, epidural catheter placement can be carried out on the evening before. This is particularly recommended in cardiac surgery using extracorporeal circulation [13,14].

• Low-molecular-weight heparins

  Low-molecular-weight heparins are used both for perioperative thromboembolism prophylaxis and in the treatment of thromboembolism. Coagulation parameters such as ACT and aPTT remain largely unaffected, they are thus not useful for monitoring the anticoagulant effect. However, if required, the anticoagulant effect of low-molecular-weight heparins is easily determined by measuring plasma anti-factor Xa activity using functional tests (anti-Xa activity, Hep test). The advantages of low-molecular-weight heparins lie in their high level of bioavailability (ca. 100%) after subcutaneous administration and their long half-life of 4–7 hours, so that they only need to be administered once a day. For thromboembolism prophylaxis, they are superior to unfractionated heparins in patients who are at high risk (e.g., with hip prostheses or trauma), without leading to increased bleeding complications [15]. They are regarded as the gold standard for thromboembolism prophylaxis in high-risk patients. Although the individual low-molecular-weight heparins differ in their molecular weight and pharmacokinetics and have been approved for different ranges of indications, no difference in clinical efficacy has yet been demonstrated between the individual preparations. Heparin-induced thrombocytopenia is caused much less often by low-molecular-weight heparins than by unfractionated heparins [16]. Despite this, they should not be administered to patients with heparin-induced thrombocytopenia, as there is a high level of cross-reactivity of 90%.
  
  Following subcutaneous administration of low-molecular-weight heparins, maximum efficacy levels are reached after approximately 4 hours, and the terminal elimination half-life in patients with healthy kidneys is 6 hours [17]. In patients with severe renal insufficiency, there is both a higher maximum anti-Xa activity level and also an increase in the elimination half-life to up to 16 h in steady state [17]. In contrast to unfractionated heparins, low-molecular-weight heparins show a higher promotion of fibrinolytic activity and less platelet interaction [18]. This is reflected in a greater thrombus regression in the treatment of deep venous thromboses [19]. To avoid bleeding complications, there should be a time interval of no less than 12 hours in patients with healthy kidneys between subcutaneous administration of low-molecular-weight heparins at prophylactic dosages and the placement or removal of an epidural catheter [5,20]. This means that low-molecular-weight heparins can be administered for thrombosis prophylaxis on the evening before the operation, without interfering with planned regional anaesthesia.
  
  By contrast, if low-molecular-weight heparins are being administered at a therapeutic dosage once or twice daily, one should wait for at least 24 hours after the last administration. Whether a 24-hour interval is acceptable in relation to the thromboembolism risk needs to be considered in the individual case, and if there is a high risk of thromboembolism – one should refrain from neuraxial blockade and continue the administration of LMWH.
  
  Following spinal/epidural puncture, or after removal of a spinal/epidural catheter, a repeat administration of low-molecular-weight heparins should be carried out at the earliest after 2–4 hours.

• Thromboembolism prophylaxis in non-elective procedures

  Patients undergoing trauma surgery have an increased risk of thromboembolism, so that prophylaxis is usually started preoperatively, immediately after admission. On the other hand, these patients also have an increased risk of aspiration, making locoregional anaesthesia preferrable. In accordance with the German guidelines for thromboembolism prophylaxis in surgical patients, these patients should therefore receive 5000 IU unfractionated heparin as early as possible after in-patient admission, in order to obtain as long as possible a time interval between heparin administration and neuraxial puncture [21]. Low-molecular-weight heparins are only administered postoperatively, at the earliest 7–9 hours after the administration of unfractionated heparins. In this special situation, observation of the 4-hour interval required in elective procedures between the administration of unfractionated heparins and neuraxial puncture can be disregarded after careful individual risk–benefit analysis, requiring especially close neurological monitoring. In these cases, spinal anaesthesia should be preferred over epidural anaesthesia due to the diminished associated risk of spinal epidural haematoma.

Fondaparinux (Arixtra®)

Fondaparinux is a synthetic pentasaccharide. It leads to selective inhibition of factor Xa, mediated by antithrombin, without inhibiting thrombin (factor IIa). It can be monitored using anti-factor Xa tests. Platelet aggregation is not affected. There is no cross-reactivity with heparin-induced antibodies. Individual studies and a meta-analysis have shown that fondaparinux is superior to low-molecular-weight heparins in thromboembolism prophylaxis in patients with knee prostheses and hip prostheses or hip fractures [22]. Fondaparinux is started postoperatively, in contrast to the frequent practice of beginning thromboembolism prophylaxis preoperatively in Europe. To avoid surgical bleeding complications, a time interval of more than 6 hours after the end of the operation is recommended. The recommended dosage is 2.5 mg; higher dosages increase the rate of bleeding complications without reducing the thromboembolism rate and are only approved for therapeutic anticoagulation. Due to its long half-life of 18 hours, fondaparinux only needs to be administered once a day. Even in patients with healthy kidneys, plasma levels only reach a stable plateau 2–3 days after the start of fondaparinux administration [23]. Although the elimination is mainly renal, a dose adjustment above a creatinine clearance of 30–50 mL/h is not recommended by the manufacturer; below a creatinine clearance of 30 mL/h substantial accumulation occurs, with a doubling of the plasma levels, which are still within the therapeutic range even after 24 hours [24]. Fondaparinux is therefore contraindicated in patients with severe renal insufficiency.

Due to the postoperative approach, this agent has advantages in neuraxial regional anaesthesia, since coagulation is not affected at the time of puncture. The rising plasma levels during the initial days of treatment and the accumulation in patients with renal insufficiency need to be taken into account during removal of an epidural catheter in particular. Experience from the EXPERT study in patients undergoing continuous regional anaesthetic techniques following orthopaedic procedures has now been published [25]. This study, with a total of 5387 patients, included 1428 patients undergoing regional anaesthesia procedures in whom fondaparinux administration was omitted once on the evening before the planned catheter removal. This provided a time interval of 36–42 hours before catheter removal, which did not lead an increased rate of thromboses. No cases of spinal epidural haematoma occurred.

In therapeutic anticoagulation with fondaparinux using 5–10 mg subcutaneously, neuraxial regional anaesthesia should not be performed due to the substantial potential for accumulation.

Danaparoid (Orgaran®)

Danaparoid is a glycosaminoglycan mixture consisting of 84% heparin sulphate, 12% dermatan sulphate and 4% chondroitin sulphate. Its effect occurs mainly through antithrombin III–mediated inhibition of factor Xa. Danaparoid is used for thrombosis prophylaxis and treatment in heparin-induced thrombocytopenia, although cross-reactivity with heparin-induced antibodies can occur in 10–20% of the patients. The efficacy of the agent in perioperative thromboembolism prophylaxis and the bleeding risk are comparable with those of low-molecular-weight heparins [26]. As the terminal half-life of danaparoid may be markedly prolonged in patients with renal insufficiency, dose adjustments are necessary. Cases of severe bleeding have been observed with danaparoid, there is no antidote, and the substance cannot be haemofiltered, although it can be removed using plasmapheresis [27]. Coagulation monitoring is only possible using anti-Xa activity. Despite its very long half-life, the agent is administered twice daily for thromboembolism prophylaxis, so that genuine trough levels are probably not achieved. Recommendations for thrombosis prophylaxis state that danaparoid should first be administered 2 hours preoperatively. Although neuraxial regional anaesthesias have been carried out in a very small number of patients 1 hour after danaparoid administration, this approach cannot be recommended, since at this time point a high level of anti-Xa activity has already been reached [28]. Instead, preoperative danaparoid administration should be avoided when neuraxial regional anaesthesia procedures are being planned. Due to its very long half-life and accumulation in patients with renal insufficiency, it is preferable to carry out single-shot regional anaesthesias and avoid the use of catheters.

Direct thrombin inhibitors

Other agents for perioperative thromboembolism prophylaxis and for therapeutic anticoagulation include thrombin inhibitors, which in contrast to heparins inactivate fibrin that has already bound to thrombin, thus inhibiting further thrombus growth. Binding to thrombin results in inhibition of proteolytic properties, without antithrombin or other cofactors being necessary. Thrombin inhibitors influence all functional haemostasis tests based on fibrin formation to various extents, and particularly aPTT, which is usually used for laboratory controls. The ecarin clotting time (ECT) is more specific and should be used at therapeutic dosages of thrombin inhibitors. The most important side effect of direct thrombin inhibitors at higher dosages – particularly in combination with other antithrombotic agents or platelet aggregation inhibitors – is an increased bleeding tendency [29]. There is no specific antidote; hirudins and argatroban can be eliminated by dialysis.

• Desirudin (Revasc®), lepirudin (Refludan®)

  The recombinant hirudins desirudin (Revasc®) and lepirudin (Refludan®) are direct thrombin inhibitors that are administered parenterally. They are indicated for thrombosis prophylaxis (desirudin) and treatment (lepirudin) in patients with type II heparin-induced thrombocytopenia (HIT II).
  
  Following a single intravenous or subcutaneous administration, aPTT is prolonged even at low dosages. In contrast to heparins, a prolonged aPTT appears to be required for effective thromboembolism prophylaxis. Following a subcutaneous injection of desirudin, there is a fast rise in the aPTT, which is measurable within 30 minutes and reaches a maximum after 2 hours [30]. A prolonged aPTT is still measurable 8 hours after subcutaneous administration of 15 mg hirudin (prophylactic dosage 2 × 15 mg/day) [31]. The elimination half-life is 2–3 hours and is markedly prolonged in patients with reduced renal function, due to the mainly renal elimination.
  In general, it appears to be advisable to observe as long an interval as possible (e.g., at least 8–10 hours) between the administration of these agents and neuraxial puncture, in accordance with the dosing interval (hirudin 12-hourly, dosage 2 × 15 mg), and also to avoid combinations with other antithrombotic agents.
  
  Hirudins accumulate in patients with renal insufficiency. For desirudin, monitoring of prophylaxis using aPTT is recommended in patients with creatinine clearance levels between 30 and 90 mL/min; in those with creatinine clearance levels below 30 mL/min, desirudin is contraindicated. The approved dosage of lepirudin needs to be corrected downwards; bolus administration is now avoided and the initial treatment in patients with healthy kidneys is started at 0.1 mg/kg/h. For lepirudin, a dose reduction of up to 85% is recommended in patients with severe renal insufficiency. Treatment should be monitored using aPTT or ECT. Antibody formation may develop in approximately 40% of patients when lepirudin is administered for several days; this delays the elimination of lepirudin and leads to substantial and incalculable prolongation of its activity [32]. Bleeding complications are therefore often observed during lepirudin treatment.
  
  Although desirudin (Revasc®) was administered in a small number of patients immediately after neuraxial puncture without spinal epidural haematomas occurring, this approach is not advisable, due to the agent’s pharmacokinetics. A delay of at least 2–4 hours after carrying out neuraxial regional anaesthesia should be observed. Time intervals of 8–10 h before puncture only apply in patients with normal renal function and no antibody formation, and it is therefore recommended that aPTT should be checked before puncture. A case of spontaneous epidural haematoma during lepirudin treatment has been reported [33].

• Argatroban (Argatra®)

  Argatroban is a direct thrombin inhibitor, which has been approved for the treatment of HIT II. Argatroban is exclusively eliminated through the liver, with no renal elimination, so that it is preferable in patients with renal insufficiency. The recommended dosage is 0.5–2.0 µg/kg/min i.v., and treatment can be monitored using aPTT, with the aim to achieve a 1.5–3-fold prolongation of aPTT. Dose reduction is required in patients with reduced hepatic function. In patients without reduced hepatic function, normalization of the aPTT is achieved 2–4 h after the end of argatroban infusion, due to the agent’s short half-life of 35–45 min [34].
  
  When deciding whether to carry out neuraxial regional anaesthesia, a distinction needs to be made between patients with a history of HIT, who only require thromboembolism prophylaxis – usually low-dose danaparoid, hirudin, or fondaparinux (off-label) subcutaneously – and patients with acute HIT II, in whom therapeutic anticoagulation treatment is required. In the latter patients, there is a high risk of thromboembolism if anticoagulation is interrupted in the peri-interventional period. On the one hand, these patients frequently suffer from multiple organ failure  including coagulation disturbances, may be ventilator dependent, making neuraxial blockade not advisable.

Vitamin K antagonists (phenprocoumon, warfarin)

Therapeutic anticoagulation with vitamin K antagonists represents an absolute contraindication to regional anaesthesia procedures. Since the normalization of coagulation parameters can take several days even after these agents have been withdrawn, normalization has to be awaited and coagulation has to be checked using laboratory tests. Faster normalization can be achieved by administering vitamin K, fresh plasma, or coagulation factors (PPSB). However, this should only be carried out with an appropriate indication and following an individual risk–benefit analysis. Normalization of coagulation using coagulation factors purely in order to carry out neuraxial regional anaesthesia is not indicated.

In contrast to most European countries, postoperative use of vitamin K antagonists for thrombosis prophylaxis is widespread in the USA. In comparison with unfractionated or low-molecular-weight heparins, thrombosis prophylaxis with vitamin K antagonists is less effective. Horlocker et al. [35] reported on 188 patients who received low-dose thrombosis prophylaxis with warfarin postoperatively after epidural analgesia. The epidural catheters were removed during continuing warfarin therapy, without an epidural haemotoma occurring. This is probably explained mainly by the small number of patients included and by the very early removal of the epidural catheter, before the onset of effective anticoagulation (INR < 1.4). It is inadvisable to withdraw an epidural catheter when vitamin K antagonists have already taken full effect, and accordingly vitamin K antagonists should on principle only be administered after the catheter has been removed.

Another analysis of 950 patients undergoing epidural analgesia in whom vitamin K antagonist administration was started preoperatively also did not observe any bleeding complications. However, no detailed information was provided regarding the patients’ coagulation status [36]. Apart from these two studies, several case reports on spinal epidural haematomas following neuraxial regional anaesthesia with simultaneous intake of vitamin K antagonists have been published [7]. In view of the high rate of bleeding complications in patients receiving therapeutic dosages of vitamin K antagonists, this is not surprising. As in Europe, therapeutic oral anticoagulation treatment is also regarded in the USA as being a contraindication to neuraxial regional anaesthesia; in contrast to Europe, however, perioperative thrombosis prophylaxis with warfarin is still regarded as possible there, despite the problems associated with catheter removal [3].

Acetylsalicylic acid

• Acetylsalicylic acid and the risk of bleeding

  Even after a single administration, acetylsalicylic acid dose-independently leads to irreversible inhibition of platelet function as a result of cyclooxygenase inhibition. Due to the lack of a nucleus, platelets are not able to newly synthesize cyclooxygenase, so that the effect persists even after the withdrawal of acetylsalicylic acid for the lifetime of the platelets – usually 7–10 days. With healthy bone marrow, however, more than 30% of the irreversibly inhibited platelets are replaced within 3 days. With a normal platelet count, this is usually sufficient to largely normalize haemostasis. Analyses of medical patients have shown that the risk of spontaneous bleeding complications in those taking acetylsalicylic acid is doubled, although the risk is generally very low. A total of 800 patients would thus need to be treated annually for one additional bleeding event to be observed [37]. A slight increase in bleeding complications is also observed in surgical patients receiving acetylsalicylic acid, but in most cases transfusions are not required. In cardiac surgery, the risk of relevant bleeding is increased 1.4-fold, and thus less relevant than the duration of extracorporeal circulation or the presence of renal insufficiency [38]. In a meta-analysis of surgical procedures other than those involving cardiac surgery, 1.5-fold higher bleeding rates were reported in patients receiving acetylsalicylic acid, although with the exception of intracranial procedures, prostate resections, and tonsillectomies, no severe cases of bleeding were observed [39].
  
  The safety of neuraxial regional anaesthesia in patients receiving acetylsalicylic acid is mainly based on three studies [40–42]. Although the Collaborative Low-dose Aspirin Study in Pregnancy (CLASP) included a total of 9364 pregnant women, only 2783 of the women underwent epidural analgesia, 1422 of whom had taken acetylsalicylic acid during their pregnancy, only half of whom had continued it up to delivery, so that approximately 700 women were left for safety assessment. The much lower risk of suffering an epidural haematoma among pregnant women, in comparison with other groups of patients, needs to be taken into account [4]; the reduced risk is most likely due to activation of the coagulation system during pregnancy and the lack of additional antithrombotic medication. The study by Horlocker et al. [41] included a total of 924 orthopaedic patients, 193 of whom had taken acetylsalicylic acid preoperatively. Intake of acetylsalicylic acid was defined as intake within the previous week, although the extent of platelet aggregation inhibition declines measurably 3 days after the withdrawal of acetylsalicylic acid [43]. The number of neuraxial punctures in patients receiving acetylsalicylic acid was thus correspondingly small; only 22 of 1000 patients simultaneously received thromboembolism prophylaxis with heparins, and no epidural haematomas were observed. In another study including 1035 patients with no accompanying thromboembolism prophylaxis, again no epidural haematomas were observed in 158 patients who had taken acetylsalicylic acid within the previous week [42].
  
  Although the administration of acetylsalicylic acid alone does not appear to lead to increases in bleeding complications, a higher rate of complications has been observed in both surgical and medical patients when heparins were administered simultaneously. A Canadian research group calculated that with simultaneous administration of heparins for thrombosis prophylaxis, the risk of spinal epidural haematomas in those taking acetylsalicylic acid was one in 8,500 after epidural anaesthesia and one in 12,000 after spinal anaesthesia [44].
  
  Non-steroidal anti-inflammatory drugs also inhibit cyclooxygenase and thus platelet aggregation. The effect is reversible and continues in accordance with the half-life of the agent used. With the exception of tenoxicam and piroxicam, platelet function normalizes within 12–24 hours after withdrawal [45]. If an interaction between non-steroidal anti-inflammatory drugs and platelet function during puncture is to be safely excluded, interrupting the agents from the evening before the planned puncture or catheter removal is sufficient. For selective COX-2 inhibitors, there is no evidence of any relevant effects on platelet aggregation capacity or an increased bleeding tendency [46]. The non-opioid analgetic agents Paracetamol (acetaminophen) and metamizole (dipyrone) have not been linked to spinal epidural haematomas to date.
  In general, on the basis of the available data, it can be assumed that non-steroidal anti-inflammatory drugs – including acetylsalicylic acid – without accompanying thromboembolism prophylaxis using anticoagulants does not lead to an increased risk of spinal epidural haematomas and thus does not represent a contraindication.

• Aspirin and the risk of cardiovascular events

  Acetylsalicylic acid reduces the risk of repeated cardiovascular events in patients with a previous acute coronary syndrome, stroke, or peripheral arterial occlusive disease. The mortality is reduced by acetylsalicylic acid by approximately 15%, and the rate of repeated cardiovascular events is reduced by around 30% [47]. Case series in recent years suggest that the morbidity and mortality – particularly in patients with recently implanted coronary stents or unstable coronary syndromes – is already markedly increased if acetylsalicylic acid is briefly interrupted before a surgical procedure [39,48,49], since cardiovascular events can already occur within a few days after the withdrawal. This may be explained by increased coagulation activity in patients with cardiovascular disease, additionally, a rebound phenomenon has been discussed [50,51]. The risk of late stent thromboses is increased in patients with drug-eluting stents in particular [52]. In summary, the preoperative withdrawal of acetylsalicylic acid is unnecessary in many operations and is associated with a high risk of acute thrombosis. It is now recommended that patients with an acute coronary syndrome or stent implantation should continue to take acetylsalicylic acid on a lifelong basis [53]. Particularly in patients with drug-eluting stents, acetylsalicylic acid should only be withdrawn before a planned operation in exceptional cases – i.e., when there is a life-threatening bleeding risk [54]. It should be noted that the simultaneous intake of ibuprofen can lead to a reduction in the efficacy of acetylsalicylic acid [46]. It is not at present known whether this also applies to other NSAIDs.

The decision for or against regional anaesthesia in patients taking acetylsalicylic acid requires a careful risk–benefit analysis, noting any history of bleeding, and with a physical examination (e.g., for signs of petechiae or haematomas), and/or laboratory controls (platelets, possibly a PFA-100 test), and it should be taken on an individual basis. Since the combination with thromboembolism prophylaxis can increase the risk of bleeding, it is recommended when a puncture is being planned in patients taking acetylsalicylic acid that perioperative thromboembolism prophylaxis with longer-acting anticoagulants (e.g., low-molecular-weight heparins) should be omitted on the day before puncture or only started postoperatively; the same applies to the time of catheter removal in order to avoid the combination of the two groups of antithrombotic agents. The resulting time window between the administration of low-molecular-weight heparins or fondaparinux and neuraxial puncture/catheter removal would amount to 36–42 hours (Fig. 1).

Thienopyridines

Ticlopidine (Ticlid®) and clopidogrel (Iscover®, Plavix®) are platelet aggregation inhibitors belonging to the group of thienopyridines. They act by antagonizing adenosine diphosphate (ADP) at the platelets’ purine receptors. ADP-induced aggregation is non-competitively and irreversibly inhibited, while the arachidonic acid metabolism is not affected. The gradual, dose-dependent full onset of effect taking place over a period of days is explained by the fact that the two agents have to be converted into active metabolites in vivo. Metabolization into the active agents takes place in the liver, and the active metabolites are eliminated renally [55]. Finally, a reduction in ADP-dependent activation of the glycoprotein IIb/IIIa receptor takes place, causing a reduction in fibrinogen formation and platelet cross-linking.

Ticlopidine. After oral administration of 500 mg ticlopidine per day, a maximum aggregation-inhibiting effect is exerted after 8–11 days. Due to the irreversible platelet function inhibition, with an elimination half-life of 24–32 hours, an effect on the platelets is still evident even more than 72 hours after the withdrawal of ticlopidine [56]. With long-term administration, the elimination half-life increases to over 90 hours, and normalization can only be expected 10–14 days after withdrawal of the agent [57]. In contrast to clopidogrel, ticlopidine can lead to neutropenia in up to 1% of patients, which limits the widespread use of the agent and makes regular blood count checks necessary in the initial weeks.

Neuraxial regional anaesthesia should only be carried out in patients who have received ticlopidine if a time window of 10 days can be observed between the last intake of ticlopidine and the puncture.

Clopidogrel. Following oral administration of 75 mg clopidogrel, the maximum platelet function–inhibiting effect is observed after 3–7 days, or after approximately 12–24 hours following initial bolus administration of 300–600 mg. Recovery of platelet function only occurs 6–7 days after the end of clopidogrel administration [58], so that neuraxial regional anaesthesia should only be carried out if a time interval of 7 days between the last intake of clopidogrel and the puncture is possible.

• Thienopyridines and the risk of bleeding

  Cardiac surgery conducted in patient receiving clopidogrel have shown that severe surgical bleeding can occur, with an increased need for transfusions and a 5–10-fold increase in the risk of repeat surgery, as well as a prolonged course of intensive-care treatment [59–61]. Severe perioperative bleeding complications thus occur much more frequently with clopidogrel than with acetylsalicylic acid. The incidence of severe bleeding is increased by simultaneous treatment with vitamin K antagonists, dextrans, or heparins [55]. The extent of bleeding complications in other surgical procedures without intraoperative full heparinization has not been adequately investigated. Immediate improvement of haemostasis is only possible by administering platelet concentrates.
  
  As there have already been case reports of spinal epidural haematomas in neuraxial regional anaesthetic procedures carried out in patients receiving clopidogrel [7], it is extremely inadvisable to carry out central nervous blocks with simultaneous intake of thienopyridines.

• Thienopyridines and the risk of cardiovascular events

  In comparison with acetylsalicylic acid, clopidogrel and ticlopidine appear to be superior in the prevention of ischaemic cerebral infarction, myocardial infarction, and deaths related to vascular events [62]. Patients with unstable coronary syndromes, previous percutaneous coronary interventions, and stent implantations benefit from long-term dual platelet aggregation inhibition treatment with acetylsalicylic acid and clopidogrel [63]. If dual platelet aggregation inhibition after coronary interventions is prematurely withdrawn, the risk of acute stent thromboses and myocardial infarctions is substantially increased, with a high mortality rate [64]. This also appears to be the case even when perioperative bridging is carried out using heparins and the platelet aggregation inhibitors are only withdrawn very briefly [65]. Patients with a drug-eluting stent are at risk for a particularly long period, due to late and incomplete endotheliazation [52], so that the American Heart Association currently recommends only using drug-eluting stents if no surgery is planned within the following 12 months and the patients show a high degree of compliance [54]. A cardiologist should be consulted before any interruption of platelet aggregation inhibition, elective operations should be postponed for at least 12 months, and clopidogrel plus acetylsalicylic acid should be administered during that period. With metal stents, dual platelet aggregation inhibition should be administered for at least 4 weeks, and with both types of stent the administration of acetylsalicylic acid should be continued on a lifelong basis and should not be interrupted perioperatively whenever possible [54].

Glycoprotein IIb/IIIa inhibitors

Inhibition of the common final pathway of platelet aggregation by inhibiting the glycoprotein IIb/IIIa receptor represents the maximum form of platelet aggregation inhibition. The aggregation inhibition is reversible. After intravenous administration, abciximab (ReoPro® at the standard dosage) leads to over 80% inhibition of ADP-induced platelet aggregation and to a reduction in thrombin generation. Abciximab also has additional antithrombotic properties that distinguish it from other agents in this group. It binds to platelets very quickly and can be demonstrated there even 2 weeks after the last administration; no significant renal elimination takes place. If CABG surgery has to be carried out after a coronary procedure, the administration of abciximab should be stopped as early as possible and the administration of low-molecular weight glycoprotein IIb/IIIa antagonists such as eptifibatide (Integrilin®) and tirofiban (Aggrastat®) should be stopped at least 4 hours preoperatively [66]. If severe bleeding occurs, transfusion of platelet concentrates is required, although redistribution of abciximab among the freshly infused platelets can also partly inhibit these. With eptifibatide and tirofiban, it can be assumed that there is 50% or 80% platelet aggregation capacity 4 hours after the end of intravenous administration [67].

The most frequent side effects are bleeding and thrombocytopenia [68]. The incidence of thrombocytopenia with abciximab is 0.3–1.0% [67]; it is most likely to be caused immunologically and occurs within the first 24 hours for a maximum of 1 hour, so that the platelets needs to be checked at this time [66]. In acute coronary interventions, glycoprotein IIb/IIIa inhibitors reduce the rate of myocardial infarction and also the mortality rate. Abciximab is more effective than tirofiban or eptifibatide [69]. In contrast to the acute effects of intravenous glycoprotein IIb/IIIa inhibitors, long-term administration of oral preparations does not appear to reduce cardiovascular complications; instead, an increased tendency to bleed and an increased mortality rate were observed [67].

Since glycoprotein IIb/IIIa inhibitors are used in patients with acute coronary syndrome in combination with anticoagulants and acetylsalicylic acid, and as cardiac surgery procedures are usually conducted as emergencies with continuing anticoagulation, central nervous blockade is contraindicated. If a catheter has to be introduced or removed after the administration of glycoprotein IIb/IIIa antagonists, most guidelines recommend waiting at least 48 hours after abciximab, or 8–10 hours after tirofiban or eptifibatide [70].

Neuraxial regional anaesthesia and thrombolysis

When thrombolysis is required, spinal epidural haematomas are occasionally observed even without previous neuraxial puncture. To prevent bleeding complications, neuraxial regional anaesthesia within the previous few weeks and neurosurgical procedures are therefore regarded as contraindications for thrombolysis [71].

As thrombolysis usually represents an emergency indication that cannot be postponed – e.g., in severe pulmonary embolism or myocardial infraction – time intervals cannot be observed when the epidural catheter is already in place. Taking into account the risk of spinal epidural haematomas, which are more frequent at the time of placement or removal of a catheter, it appears to be safer to leave the catheter in situ even during thrombolysis. In three case reports of spinal epidural haematomas after thrombolytic therapy, the administration of urokinase was already started intraoperatively shortly after neuraxial puncture [72–74]. Catheter removal should only be carried out after the thrombolytic effect has definitely worn off and coagulation has normalized. It must be borne in mind that the effect on coagulation may persist for much longer than the half-life of the individual thrombolytic agents; additional analyses such as assessment of the fibrinogen level or thromboelastography may be helpful.

Alternative medicine

Up to 50% of patients take alternative medicines preoperatively, as most patients do not voluntarily indicate this, it often remains undiscovered [75]. The main agents used include echinacea, Ginkgo biloba, garlic, ginseng, ephedra, aloe, and dwarf palm. Ginkgo, garlic and ginseng in particular have been linked to thrombocytopenia, platelet aggregation inhibition, and interactions with vitamin K antagonists and the development of spinal epidural haematomas. The assessment of alternative medicines is made more difficult by the fact that their manufacture is only regulated in Germany; by contrast, preparations manufactured in other countries often show the addition of other active substances, including non-steroidal anti-inflammatory drugs and acetylsalicylic acid, which may explain the coagulative effects observed in individual studies [76]. Case series on the side effects of alternative medicines report heavy metal intoxication, hepatic failure and allergic reactions as the most common side effects, whereas hardly any bleeding complications have been reported [77]. There is only one report of an epidural haematoma with simultaneous intake of garlic in the literature [78], despite the widespread use of these substances, making an increased risk extremely unlikely. An impairment of haemostasis  has been excluded for garlic and ginkgo [79,80]. Warnings against neuraxial puncture when alternative medicines are being taken and recommendations to withdraw these substances preoperatively are at present unjustified, particularly for preparations manufactured in Germany.

Avoidance of complications/patient monitoring

Neuraxial regional anaesthesia has to be carefully and appropriately explained to the patient, and requires an anaesthetist with experience in the technique. After performance of the block, the patient should be monitored at least until the effect of the regional anaesthesia is clearly declining – e.g., with a reduction in the extent of sensory block by two segments or a return of motor function. Particular attention should be given to persistent sensory or motor deficits, radicular back pain, pressure sensitivity in the puncture area and bladder dysfunction. When there is a clinical suspicion of neuraxial haematoma, appropriate diagnostic or treatment measures must be started immediately.

When continuous or patient-controlled analgesia techniques are used, regular visits by the acute pain service and a high degree of attentiveness on the part of all staff caring for the patient, as well as by the patient himself or herself, are required. Postoperative epidural analgesia should be carried out with a low-concentration local anaesthetic as a differential block with primary sensory analgesia, so that motor function can be used as a sign of spinal or surgical complications. It may be advantageous to establish thoracic epidural analgesia, due to the lack of motor blockade of the lower extremities.

The diagnostic method of choice when there is a clinical suspicion of spinal bleeding is magnetic resonance imaging, as this allows precise localization of the bleeding and its extent. However, inability to carry out magnetic resonance imaging must never be allowed to delay diagnosis and thus treatment. In this case, computed tomography should be carried out immediately as an alternative. This also helps to avoid unnecessary patient transport and the resulting time loss for the patient. The only effective treatment is a decompressive laminectomy as quickly as possible. Less than 6 to 12 hours should pass between the appearance of symptoms and surgical decompression [81]. Ideally, the appropriate algorithms should already have been agreed in advance with neurosurgical colleagues and the radiologist.

Summary

Conducting central nervous blocks in patients in whom anticoagulation treatment has been carried out or is planned continues to be a matter of controversy. The risk of spinal haematomas is extremely low, but it can have dramatic neurological consequences for patients, particularly in those with reduced coagulation.

Despite this, neuraxial blockade is safe if the cautionary measures described are observed, an atraumatic technique is used, and an individual risk–benefit analysis is carried out, even in patients who are to receive antithrombotic agents perioperatively. Acetylsalicylic acid probably only increases the risk of spinal epidural haematoma after neuraxial regional anaesthesia to a minor extent if the time intervals relative to the simultaneous administration of anticoagulants for thromboembolism prophylaxis are increased. Withdrawal of platelet function inhibitors such as acetylsalicylic acid or clopidogrel in patients after coronary artery interventions carried out less than 12 months previously increases the risk of cardiovascular events. This should be taken into account during the individual risk–benefit analysis.

Both unfractionated and low-molecular-weight heparins increase the risk of spinal epidural haematoma if these agents are administered at high dosages or with too short a time interval from the puncture or removal of an epidural catheter. In patients receiving therapeutic anticoagulation treatment with heparins or vitamin K antagonists, puncture and the removal of an epidural catheter continue to be contraindicated.

The lack of case reports on spinal epidural haematoma in neuraxial regional anaesthesia and more recent agents used for thromboembolism prophylaxis do not allow to conclude that these agents lead to less severe bleeding complications, as the numbers of patients studied are often too small (e.g., with hirudins). In everyday clinical practice, the time intervals tested in prospective and correspondingly carefully planned studies (e.g., with fondaparinux) must be observed.

Acknowledgements

The authors are grateful to all of their colleagues who, by taking part in discussions and providing their individual expertise during the development of these revised recommendations, have helped reflect the current state of knowledge as comprehensively as possible while constantly remaining in touch with current anaesthesiological practice. Particular thanks go to the members of the DGAI’s Scientific Working Group on Regional Anaesthesia.

ACT, activated clotting time; aPTT, activated partial thromboplastin time; ECT, ecarin clotting time; INR, international normalized ratio; IU, international unit; NSAIDs, non-steroidal anti-inflammatory drugs.
* All times given refer to patients with normal renal function.
† Maximum prophylactic dosages for low-molecular-weight heparins are listed in Table 2.
‡ Prolonged time interval in patients with hepatic insufficiency.
§ Interrupt low-molecular-weight heparins once; no low-molecular-weight heparins 36-42 h before planned catheter insertion or removal.