Volume 8, No. 4 • Winter 1993

In My Opinion: A Debate – Should Epidurals be Used in Patients Who Get Systemic Anticoagulation?

John P. Williams, M.D.; Sture G. Blomberg, M.D., Ph.D; John Tinker, M.D.

Should Epidurals be Used in Patients Who Get Systemic Anticoagulation?

‘Yes: Regional Anesthesia and Anticoagulants are OK’

The use of any technique in anesthesiology is accompanied by some quantifiable risk to the three P’s: public, patient and practitioner. In order to rationally assess the relevance of the type of risk imposed by a technique, one must first decide which of these groups is at risk, what specific risk is imposed (i.e., what is the incidence of the selected outcome), what possible benefit(s) is/are derived and finally what risks/benefits are posed by alternative techniques. Unfortunately, time and space constrain this debate only to considerations of practicality.

From all these perspectives, the greatest risk is neuraxial hematoma formation. However, the issue is not whether administration of neuraxial anesthetics in the presence of anticoagulation poses significant risk. More properly stated it is: does the combination (of neuraxial anesthetics and anticoagulation) increase the risk of neuraxial hematoma above that of either alone and if so is that increase additive, geometric or exponential?

Several studies have documented the safety of neuraxial anesthetics: Dripps’ (approximately 10,000 neuraxial anesthetics), a similar Chinese study (10,000 neuraxial anesthetics), Lund’s review of 150,000 neuraxial anesthetics, Kane in his landmark study of the incidence of neurological injury following spinal and epidural anesthesia noting a 1:45,783 incidence of paraplegia directly attributable to the anesthetic. Presumably, a few of these anesthetics were administered in the presence of anticoagulation (either pharmacological or pathological); if so, one would expect to find a noticeable increase in neurological injury if the additional burden of anticoagulation exponentially or geometrically increased the risk of neuraxial hematoma. Thus if an increased risk exists, it is only additive.

More direct evidence is available from the work of Rao and El Etr who noted no increase in the risk of neurological impairment when administering epidural anesthesia prior to system heparinization for vascular surgery (approximately 4,000 neuraxial anesthetics). A corroborative report by Odoom and Sih supports Ns finding (approximately 1,000 neuraxial anesthetics). Although Yeager did not intend to examine the role of neuraxial anesthetics with regard to the incidence of neuraxial hematoma, he did not detect an increase.

More recently, Blomberg (1) and Liem (2) examined the role of neuraxial anesthetics in ischemic heart disease: the former in the setting of unstable angina, the latter for coronary artery surgery. Neither investigator noted an increase in the incidence of neurological impairment. Dr. Blomberg’s epidurals were inserted after discontinuing heparin for four to six hours. Dr. Liem’s patients underwent instrumentation on the evening prior to surgery. Both of these represent traditionally ‘high-risk’ patient groups for neuraxial instrumentation; yet, neither investigator noted any evidence of neuraxial hematoma formation.

The benefits of neuraxial regional anesthesia during coronary artery bypass surgery include: more thorough rewarming following cardiopulmonary bypass, more rapid extubation, shorter stays in ICU, superior analgesia, decreased incidence of postoperative ischemia, improved patient communicative skills, and shorter awakening times.

The benefits of thoracic epidural anesthesia in unstable angina include: decreased utilization of intravenous heparin and nitroglycerin, increased exercise tolerance on treadmill tests, decrease in the number and duration of anginal episodes, an increase in the global ejection fraction, and an increase in the resting diameter of stenosed coronary arteries. Perhaps most impressively, these epidurals were inserted in patients who were refractory to conventional medical therapy (intravenous nitroglycerin and heparin). While this was not a prospective randomized study, it is tempting to speculate that some of these patients did not proceed to infarction principally because of this regional sympathetic blockade.

The benefits of neuraxial anesthetics as noted by Yeager’ and his group were: shorter ICU stays (although after log-transformation of the original data the difference was not significant), decreased morbidity from several causes (CHF, pulmonary, sepsis), shorter hospital stays (similar comment as regards ICU stays), and decreased hospital charges. Indeed, the Yeager group terminated their trial early because of the dramatic difference in morbidity between the general and regional anesthetic groups.

What benefits does general anesthesia provide? In head to head comparisons with regional neuraxial anesthesia techniques none. Is the risk of neuraxial hematoma formation absent with general anesthesia? No. A recent paper in the Japanese literature reports the occurrence of an epidural hematoma following cardiopulmonary bypass in the absence of any manipulation or instrumentation of the neuraxis (i.e., a spontaneous epidural hematoma).

Where does this leave us? Clearly the public has much to gain from the use of neuraxial techniques. Improved pain relief, the rapid ability to communicate with practitioners and family members, and a decrease in the incidence and severity of myocardial ischemic and other co-morbid events. In return, they (and we) assume a risk of neuraxial hematoma. Is this risk made greater by the addition of neuraxial instrumentation? We do not know this as yet. Only large-scale prospective comparisons or retrospective meta-analyses will give us that answer.

It is interesting to look back on the history of endotracheal intubation for comparison. The incidence of unrecognized esophageal intubation was and continues to be a risk to the use of endotracheal anesthesia (currently approximated at 1:100,000). However, this low incidence is at least partly a result of years and years of dedicated research into less expensive (read easily available), more reliable methods of detecting this complication; not by recognizing that endotracheal intubation was a problem and ending the discussion. As an old adage states, “Never were my eyes so open as when I first began to see.’ Let’s open our eyes and search for new methods of detection; not shut them and complain about the light!

Drs. Williams and Blomberg are on the faculty of the Department of Anesthesiology, UCLA School of Medicine, Los Angeles, CA.


1. Blomberg S, Emanuelsson H, Kvist H, Lamm C, Ponten J, Waagstein F, Ricksten S-E. Thoracic epidural anesthesia and central hemodynamics in patients with unstable angina pectoris. Anesth Analg, 69:558-562,1989.

2. Liem TH, Booij LH, Hasenbros MA, Gielen MJ. Coronary artery bypass grafting using two different anesthetic techniques: Parts 1-3j. Cardiothorac Vasc Anesth 6:148-167,1992.

3. Yeager MP, Glass DD, Neff RK, Brinck-Johansen T. Epidural anesthesia and analgesia in high-risk surgical patients. Anesthesiology 66:729-736,1987.

‘No: A Case Against Thoracic Epidurals in Systemically Anticoagulated Patients’

by John Tinker, M.D.

The case I wish to make against this practice involves risk versus benefit. I am sure that my worthy opponents in this ‘debate’ will carefully delineate all the real and/or potential benefits to be achieved by this practice, e.g., better analgesia, which is above all more humane but which may also have the not-so-side-benefits of better pulmonary toilet, shorter intensive care and then hospital stay, etc. The problem here is not with the benefits, which I’m sure in good hands are substantial, but with the risk.

When we think of risk benefit ratios, we tend to think of risk in terms of incidence, rather than rnagnitude, of complications, because incidence is much easier to quantify and contemplate. Actually, for most uncommon or rare complications, it is the magnitude about which we probably should be concerned. A thoracic hematoma leading to an infection and/or paraplegia is possible with this form of therapeutic intervention. It is not likely. It is not even remotely likely. It is probably not even quantifiably likely, at least not reliably so. Lawyers have argued that such rare complications are so unlikely that, if they occur, there must have been some sort of negligence (a questionable premise, but one which has carried the day in several cases).

Although I think the above lawyers’ argument is specious, there probably is validity in the contention that if a devastating complication is very rare in a general population, and if the treatment associated (rarely) with this complication is a complex treatment requiring superior mechanical skills (and this is the case for thoracic epidurals), then that rare and/or devastating complication might be expected to be more likely if the skill level of the practitioner was not of the highest degree or, perhaps, a reasonably skilled practitioner was having an “off day.’ Before you make fun of my “off day,’ think of any professional athlete of your choice. The greatest baseball players have slumps at the plate. Sometimes a golden glove outfielder will make two or three sophomoric blunders in the same game. How many times have you had two or three failed attempts at arterial cannulation in a row? I do believe there is validity to my contention that if a therapeutic intervention requires first class mechanical skills, then complications arising therefrom can be reasonably expected to occur with greater frequency if the practitioner either has less experience, or a lower mechanical skill level, or an ‘off day.’

Devastating Complications

Whether the above is valid or not, the fact that these devastating complications can occur is crucial to my argument that the magnitude as well as the incidence of complications must affect the ‘acceptability’ of any resultant risk benefit ratio. For this particular procedure (and all others), there is a real risk benefit ratio. That ratio is almost never known and may change over time for a given procedure, when done by a given practitioner, or because of many factors, but there is such a ratio, always present at any particular time.

With any rare and devastating complication, therefore, the practitioner will have trouble with a question something Re ‘how many of these rare but devastating complications am I willing to tolerate in order to gain real but not necessarily life-saving benefit in most of my patients so treated?’

Some will opine that I am posing no particularly difficult ethical question, since we make such decisions all the time. Deciding to cross the street or get on an airline entails risk of getting killed by a bus, or a crash, yet we often take these risks for decidedly ‘elective’ reasons. The difference, I would point out, is that we are exposing ourselves to these well-known risks. Even if we kid ourselves that the incidence is lower than it really is, we are, as lay people, able easily to understand the magnitude. Deciding to perform a thoracic epidural for pain relief, doing it on a stranger who is unlikely to understand the magnitude of a hematoma or infection or even paralysis (do you really even use the dreaded word, in your ‘informed’ consent? I’ll bet you usually don’t!), is to decide for someone else to take the risk an entirely different proposition.

Let me take a different example. If performance of deliberate hypotension during scoliosis repair operations in young patients saves blood (and that is controversial, but it probably does), and if there are rare complications from said hypotension, e.g., unilateral blindness, paraplegia, etc.), then how many paraplegic young patients are you willing to accept (per thousand or per million) in order to save two or so units of blood per patient? In this particular example, 15 years ago the ‘saving’ of two units of blood might well have been considered far less important than similar savings of blood would be today, based on the AID’s virus, far greater expense, non A-non B hepatitis, etc. Today, it might be easier to justify performance of deliberate hypotension for scoliosis repair than it was 15 years ago.

What about thoracic epidurals for postoperative analgesia in patients who have been Oven anticoagulants in the course of their surgery, albeit after the placement of the epidural catheter? How many thoracic epidural hematomas per thousand patients so treated are you willing to accept? How many epidural space infections? How many paraplegias? Because the risk of these complications is so small, it is unlikely to ever be possible to actually calculate a risk benefit ratio. Such a risk benefit ratio might be meaningless anyway if the complication is dependent to a considerable extent upon individual practitioner skill level.

Perhaps a better test is to ask yourself what you would like to hear from your anesthesiologist about this procedure? Would you want your anesthesiologist to tell you that there is a rare but non zero likelihood of paraplegia, or a year in pain, bed rest with an epidural space infection? Do you tell that to your patients prior to performance of this procedure? Perhaps this is the right ethical ‘test.’ If you do tell your patients something like the above in all candor and frankness, and if you then can state with equal candor that despite this very remote possibility, you strongly believe that the pain relief afforded by the procedure is worth this small risk of this devastating complication, then it seems to me that you have properly come to grips with the ethics of this problem. If, on the other hand, you are telling your patients that the risk is ‘small’ and you are deemphasizing the magnitude of the complication, however rare, then you should rethink not only your process of informed consent but your internal rationale for performance of the procedure in the first place. Many anesthesiologists say something glib at this critical juncture in the informed consent process, like “the risk is smaller than getting struck by lightning.’ I have heard this nonsense stated frequently. The reason is because the denominator is not the same! In the risk of being struck by lightning, the denominator is the U.S. population, compared to a denominator which must include only the (U.S.) population of patients having thoracic epidurals for surgery in which anticoagulants were used!

Pain control can be achieved other ways. I remain unconvinced that the epidural route of administration of narcotics has conclusively been shown to be more effective than opioids administered intravenously in a ‘patient controlled” fashion, if proper dosages are given. Because I believe this, I have difficulty telling patients that the “benefit’ of this is so great that the risk of a devastating complication can be justified.

There is always in medicine a tendency toward anecdotism, as opposed to scientific comparisons. Many colleagues of mine have read the various studies comparing, for example, epidural analgesia with intravenous PCA, and have stated that they intend to continue the epidural route because, despite the studies, they ‘know’ their method is better. They say things like ‘all you have to do is go out and see the patients.’

Thomas Preston, in his 1977 book Cardiac Surgery A Critical Review (Raven Press, NY), stated: ‘The history of medical therapeutics shows that most medical treatments do not stand the test of time, especially those that are controversial within their own times.” This particular therapy, when used with anticoagulated patients, will Rely, I predict, fit Preston’s wise characterization, especially if payors (reading the studies, not listening to the anecdotes) won’t pay for it.

Dr. Tinker is Professor and Chairman, Department Anesthesiology, University of Iowa