Secondary Logo

Journal Logo

The Open Mind: The Open Mind

A Haunting Abstract

Durieux, Marcel E. MD, PhD

Author Information
doi: 10.1213/ANE.0000000000000523
  • Free

More than 15 years ago, I published an abstract in Anesthesia & Analgesia that has haunted me ever since.1 With coauthors Bernard Lo and Christian Hönemann, I reported (overly) preliminary results of a clinical trial in progress. The trial studied the effects of ketamine and magnesium on postoperative opioid consumption in patients undergoing abdominal hysterectomy. The design was straightforward: patients were randomized to receive placebo, ketamine (0.15 mg/kg), magnesium (30 mg/kg), or both, about 10 minutes before incision. All participants were blinded. Patients received 1 μg/kg fentanyl at induction, 0.15 mg/kg morphine during the procedure, and 30 mg ketorolac at emergence. In the postanesthesia care unit (PACU), morphine was used for pain control. We measured morphine requirements and assessed pain by visual analog scale (VAS) during the first 2 hours after surgery.

Our abstract reported results of the first 4 patients in each group (Fig. 1). The results were remarkable: the placebo group required approximately 0.16 mg/kg morphine in the PACU (10–15 mg for a typical patient). In patients who received either ketamine or magnesium, this requirement was reduced by about 50%; giving both drugs “virtually eliminated the need for narcotics in PACU.”1 There were no differences in VAS scores.a

Figure 1
Figure 1:
Effects of ketamine (0.15 mg/kg), magnesium (30 mg/kg), or both on early postoperative morphine requirements after abdominal hysterectomy.1

For reasons having nothing to do with ketamine or magnesium (in fact, I moved between institutions), the study was stopped for about 4 years. Meanwhile, I learned more about opiate tolerance and receptor signaling, and when I was in a position again to continue the trial, I decided not to do so, for reasons I will explain below. However, the abstract has continued to follow me ever since. On average, just about every 2 years, someone finds it in the literature and contacts me to ask whether the final study results have been published. Over the years, by having to formulate my answers to these questions repeatedly, I have come to believe that my reasons for not continuing this study may be important for other researchers investigating nontraditional analgesics such as magnesium and ketamine. Therefore, in this article, I will explain why I did not restart the trial, and why this matters.

First, let me assure you that the results presented in the abstract, although very preliminary, may very well be reproducible: the combination of magnesium and ketamine might indeed reduce the need for opioids in the PACU. The problem is that this may not matter.

Morphine, and most other opioids, are inexpensive drugs. We do not try to reduce opioid consumption to save money on analgesics; we do so because it should reduce opioid-induced side effects; in particular, reducing opiate consumption should reduce the risk of respiratory depression. Opioid-induced respiratory depression is a very troublesome but (luckily) relatively rare side effect. Its rarity makes it difficult to use as an end point in a clinical trial. If my study had been designed using the incidence of respiratory depression as the primary end point, it would have required thousands of patients. In fact, in that case, the haunting abstract would never have been written, because (unsurprisingly) none of the first 16 patients developed respiratory depression. Therefore, we studied instead the amount of opioids administered as a surrogate. Because the incidence of most opioid-induced side effects, including respiratory depression, is dose dependent, this seems to be a very logical thing to do; so logical, in fact, that reducing opioid requirements has almost become a goal in itself. But using surrogate end points can be dangerous.

Reductions in opioid requirements induced by many analgesics should indeed translate to a reduction in opioid side effects. For example, my study patients all received 30 mg ketorolac. What we did not know in 1998 but do know now2 is that ketorolac at this dose likely has no significant effect on postoperative opioid consumption. However, had we given 60 mg, we could have expected a reduction in the opioid requirement of a 1.64-mg morphine equivalent (ME).2 Although this is a remarkably small effect size, it should reduce the risk of opioid-induced side effects, and, in fact, it does: 60 mg ketorolac reduces postoperative nausea and vomiting (PONV) substantially (odds ratio, 0.5).2 Drugs that provide analgesia by a mechanism different from opiate signaling should reduce opioid requirements and opioid-induced side effects.

Unfortunately, this may not be true when talking about drugs such as ketamine and magnesium, and the reason is that they have an unconventional mechanism of action: they interact with opiate signaling. These drugs are antagonists at the N-methyl-D-aspartate (NMDA)–type glutamate receptor. NDMA receptor signaling plays a major role in the development of opiate tolerance,3 and we now know that such opiate tolerance can develop very quickly, even during the operative period. For example, in rodents, tolerance development to morphine infusion is almost complete after 6 hours.4 At least part of the ability of NMDA receptor antagonists to reduce opioid requirements is explained by their ability to prevent this opiate tolerance.5,6

Let me try to translate this to my trial. All patients received intraoperative fentanyl and morphine. In the placebo group, these drugs induced some degree of opiate tolerance, and that these patients required 0.16 mg/kg morphine in the recovery room may reflect that they were partly tolerant. The patients who received ketamine, magnesium, or both would not have developed as much tolerance. Hence, they should obtain pain relief from a substantially lower dose of opiate in the PACU, and that is exactly what we found.

But the crux of the matter is that tolerance also develops to side effects. Therefore, a patient in the placebo group who receives 10 mg morphine to control her pain in the PACU is actually somewhat protected against respiratory depression by the opioids she received intraoperatively. NMDA receptor antagonists would be expected to prevent development of tolerance to opiate side effects, including tolerance to respiratory depression. A patient who received ketamine intraoperatively only requires 5 mg morphine in the PACU (because she did not develop tolerance intraoperatively), but she is not protected against opioid-induced respiratory depression. Therefore, it is completely conceivable that the risk of respiratory depression would be similar in a placebo group patient who received 10 mg morphine and in a ketamine group patient who received 5 mg morphine in PACU. Opioid sparing no longer automatically means fewer side effects. The surrogate end point of opioid use has become meaningless. And that is why I did not restart the trial.

The take-home message here is that drugs that influence opiate signaling may disrupt the relationship between opioid dose and risk of opiate side effects. This is important to consider when evaluating studies of such drugs. What would we predict to find in such a trial?

  • Finding 1: A notable reduction in opiate requirements (here, we would get excited);
  • Finding 2: Because opioids generally get titrated to pain level, we would find similar pain scores despite decreased opioid use (we would still be somewhat excited);
  • Finding 3: But we would also expect a similar side effect incidence despite decreased opioid use (and here, we would get much less excited).

One single example that, I hope, will illustrate why I feel that discussing a 1998 abstract still has some value is the recent meta-analysis of perioperative magnesium published in Anesthesiology by de Oliveira et al.7 It was accompanied by an excellent editorial by Naidu and Flood.8

Magnesium reduced opioid requirements, with a weighted mean difference of about 10 mg ME (Finding 1). Naidu and Flood state: “Magnesium does something. Is that something useful? If patients were more comfortable after surgery, then the utility of magnesium would be clear.”8 Based on the discussion above, I think the situation is more complicated. We could easily get these patients more comfortable by giving them more opioid; we do not need magnesium for that. The main reason why one would use magnesium is because its opioid sparing might reduce opioid-induced side effects.

But before we look at side effects, what was the effect of magnesium on pain? Actually, almost nothing: <1 point on a 10-point VAS with rest and with movement. As we would have anticipated, patients were titrated to similar pain relief with a substantially reduced dose of opioid (Finding 2). So far, so good.

So what about side effects? Respiratory depression is too rare to study, but 6 studies looked at PONV. There was not even a suggestion of a beneficial effect of magnesium (odds ratio, 1.00), even though the sample size was sufficient to detect a 15% difference in PONV with a power of 87%. Thus, magnesium administration does not appear to affect opioid side effects (Finding 3). Note that with magnesium, a reduction in requirements of 10 mg ME did not change side effect incidence at all; whereas ketorolac, which reduced opiate requirements by <2 mg ME, cut the risk of PONV in half. Why? Different mechanisms of action …

In short, the meta-analysis of magnesium suggests that the drug behaves like a typical NMDA receptor blocker: by preventing tolerance to both opioid effects and opioid side effects, it reduces opioid requirements, and at those reduced opioid doses, we find similar pain scores and similar rates of side effects. Nausea rarely leads to long-term adverse outcomes, but these data are also compatible with the hypothesis that the incidence of respiratory depression would be unchanged after magnesium administration, despite a substantial reduction in opioid usage. However, this hypothesis remains to be tested.

I want to make clear that I am not arguing that NMDA blockers are without clinical benefit; I am just making the point that with these drugs, a reduction in opioid requirements does not necessarily imply improved outcomes. But true outcome improvements have been shown with these compounds in several studies. In opioid-tolerant patients undergoing spine surgery, intraoperative ketamine reduces not only opioid consumption (even 6 weeks after surgery) but also pain throughout this time frame.9 In patients undergoing outpatient segmental mastectomies, magnesium reduces opiate consumption and improves Quality of Recovery-40 scores 24 hours after surgery.10 Results of both of these studies are compatible with the hypothesis proposed above, but they show clinical benefits directly using outcomes other than a reduction in opiate requirement. In fact, based on these and other trials, I use ketamine routinely in patients undergoing complex spine surgery and use magnesium selectively. What about patients undergoing abdominal hysterectomy, the target group of my 1998 abstract? Grady et al.11 reported recently in Anesthesia & Analgesia no benefit of intraoperative ketamine (0.35 mg/kg, then 0.2 mg/kg/h for the first 2 hours, and then 0.12 mg/kg/h for 24 hours) on functional outcome (6-minute walk distance). I am not aware of a study of functional outcome after magnesium administration in patients undergoing hysterectomy, but Ko et al.12 reported in Anesthesiology no benefit of magnesium on postoperative pain. Based on those results, I do not use the drugs in that population.

This is the story I have to explain every couple of years. It does not leave me very satisfied. Because of both simple scientific curiosity and a desire to provide the best care to patients, I would like to have a more complete answer. Can we address these issues in a definitive way and determine whether ketamine and/or magnesium have benefits or not? It is not feasible to construct a trial using frank respiratory arrest as an end point. But we do now have technology (such as continuous percutaneous CO2 measurement)13 that can assess even modest degrees of respiratory depression on a continuous basis. This would allow a trial with respiratory depression as an end point. The hypothesis of such a study would be that perioperative administration of ketamine and/or magnesium, although resulting in significant decreases in opioid requirements, does not reduce the incidence of postoperative opioid-induced respiratory depression. I am considering doing it.


Dr. Marcel Durieux is the section editor for Anesthetic Preclinical Pharmacology for Anesthesia & Analgesia. This manuscript was handled by Dr. Steven L. Shafer, editor-in-chief, and Dr. Durieux was not involved in any way with the editorial process or decision.


Name: Marcel E. Durieux, MD, PhD.

Contribution: This author is responsible for the design and content of this paper.

Attestation: Marcel Durieux approved the final manuscript.


a I am not particularly proud of this abstract. We should not have looked at the results until all data collection was complete, and we should not have published n = 4 groups without statistical analysis. But there it was, in print.
Cited Here


1. Lo B, Hönemann W, Durieux ME. Preemptive analgesia: ketamine and magnesium reduce postoperative morphine requirements after abdominal hysterectomy. Anesth Analg. 1998;89:A1163
2. De Oliveira GS Jr, Agarwal D, Benzon HT. Perioperative single dose ketorolac to prevent postoperative pain: a meta-analysis of randomized trials. Anesth Analg. 2012;114:424–33
3. Zeng J, Thomson LM, Aicher SA, Terman GW. Primary afferent NMDA receptors increase dorsal horn excitation and mediate opiate tolerance in neonatal rats. J Neurosci. 2006;26:12033–42
4. Ling GS, Paul D, Simantov R, Pasternak GW. Differential development of acute tolerance to analgesia, respiratory depression, gastrointestinal transit and hormone release in a morphine infusion model. Life Sci. 1989;45:1627–36
5. Trujillo KA, Akil H. Inhibition of opiate tolerance by non-competitive N-methyl-D-aspartate receptor antagonists. Brain Res. 1994;633:178–88
6. Mendez IA, Trujillo KA. NMDA receptor antagonists inhibit opiate antinociceptive tolerance and locomotor sensitization in rats. Psychopharmacology (Berl). 2008;196:497–509
7. De Oliveira GS Jr, Castro-Alves LJ, Khan JH, McCarthy RJ. Perioperative systemic magnesium to minimize postoperative pain: a meta-analysis of randomized controlled trials. Anesthesiology. 2013;119:178–90
8. Naidu R, Flood P. Magnesium: is there a signal in the noise? Anesthesiology. 2013;119:13–5
9. Loftus RW, Yeager MP, Clark JA, Brown JR, Abdu WA, Sengupta DK, Beach ML. Intraoperative ketamine reduces perioperative opiate consumption in opiate-dependent patients with chronic back pain undergoing back surgery. Anesthesiology. 2010;113:639–46
10. De Oliveira GS, Bialek J, Fitzgerald P, Kim JY, McCarthy RJ. Systemic magnesium to improve quality of post-surgical recovery in outpatient segmental mastectomy: a randomized, double-blind, placebo-controlled trial. Magnes Res. 2013;26:156–64
11. Grady MV, Mascha E, Sessler DI, Kurz A. The effect of perioperative intravenous lidocaine and ketamine on recovery after abdominal hysterectomy. Anesth Analg. 2012;115:1078–84
12. Ko SH, Lim HR, Kim DC, Han YJ, Choe H, Song HS. Magnesium sulfate does not reduce postoperative analgesic requirements. Anesthesiology. 2001;95:640–6
13. Soto RG, Davis M, Faulkner MJ. A comparison of the incidence of hypercapnea in non-obese and morbidly obese peri-operative patients using the SenTec transcutaneous pCO(2) monitor. J Clin Monit Comput. 2014;28:293–8
© 2015 International Anesthesia Research Society