The case for invasive placebo – is the devil in the detail?

Abstract

In this issue of Anaesthesia we publish a paper by Kumar et al. evaluating the effect of pre-operative stellate ganglion block on postoperative tramadol consumption following surgery to fixate upper limb fracture 1. This research study was prompted by the 2011 case series from McDonnell et al. that showed a marked benefit of stellate ganglion block performed for similar indications with respect to postoperative pain scores and analgesic requirements 2. Two editorials accompanied McDonnell et al.'s paper; one discussed the potential for modulation of acute somatic pain by the autonomic nervous system 3, and the other called for robust substantiation of the findings before such an approach was incorporated into routine clinical practice 4. Kumar et al. have now performed a randomised, double-blind, placebo-controlled study to address the latter point. Subjects received a 3-ml stellate ganglion injection of either lidocaine 2% or saline before general anaesthesia and surgery. The authors report a statistically significant and clinically relevant reduction in tramadol consumption, administered via patient-controlled analgesia, over the first 24 h postoperatively. There are two conventional ways of assessing the benefit of an analgesic intervention: measuring pain scores and/or recording analgesic usage (preferably with a patient-controlled analgesia system rather than administered if and when requested). Whilst pain scores are commonly used as a research tool, a statistically significant difference in pain scores may translate poorly to an actual clinical benefit for the patient. Furthermore, pain scores are also hampered by intra- and inter-individual variation in scoring, making a change in median pain scores across a group difficult to interpret. In contrast, use of analgesics can be used as a surrogate marker for pain, if one assumes that the patient is ‘titrating’ the analgesic dose to provide an acceptable level of analgesia. Whilst this is not a direct evaluation of subjective pain, it reflects the patient experience, as the reduced use of analgesic medication can be thought to represent a reduction in the pain experienced. In addition, it addresses an important non-specific aspect of managing pain, also inherent in the concept of multi-modal analgesia: that of a reduction in analgesic doses in order to reduce side-effects. The aim of the optimal analgesic package is to get the best analgesia for the least ‘cost’, if we consider cost to include not just the financial tariff but also complexity and time, side-effects and risks. Therefore Kumar et al.'s study aims to address two questions. The first is the usefulness of stellate ganglion block before upper limb surgery in reducing postoperative pain as reflected by a reduction in self-administered analgesic requirements. The second, and possibly more important question, is its exploration of the relationship between the autonomic nervous system, nociception and acute pain. If asked, most clinicians would support the statement that a link exists between the autonomic nervous system and pain. Anaesthetic textbooks and review articles alike repeatedly reinforce this view by asserting that pain stimulates a sympathetic stress response. However, there is a lack of clinical evidence to substantiate this widely held belief 5. More importantly, there is even less evidence to support the concept that this interaction between nociceptive transmission and the autonomic nervous system works in a bidirectional fashion; that is, nociception stimulates the autonomic nervous system and also conversely the autonomic nervous system modulates nociceptive processing. In this regard, Kumar et al.'s work 1 is essential in exploring an area of pain and mechanistic nociceptive theory that is complex, incompletely defined and inadequately explored. Interestingly, this paper has re-ignited the emotive issue of the appropriateness of including an invasive placebo as part of a research study assessing a therapeutic intervention. The main arguments against the use of invasive placebos in clinical trials have been outlined in 2011 in this journal 6, 7 and are revisited by Cyna & Tan in their accompanying editorial published in this issue of Anaesthesia 8. These articles focus on the scientific and ethical implications of performing a placebo nerve block that is likely to yield no benefit to the patient, yet is accompanied by all the potential risks and complications. Although Cyna & Tan make a strong argument against the invasive placebo block, perhaps the issue is not quite as clear-cut as one might think at first. The potential reasons for performing a placebo block may be addressed by considering two questions: firstly, what is the purpose of the invasive placebo in this research context? And secondly, the more general question: are invasive placebo interventions ever appropriate in medical research? Although this argument could be accused of focussing on semantics, it is important to appreciate the true definition of a placebo in order to evaluate the appropriateness of the invasive placebo intervention. Sites & Neal, in their editorial, use the definition of a placebo from the Merriam Webster online dictionary: “an inert or harmless substance or procedure” 7. Ignoring the potential for complications of the nerve block to cause harm, how does the inertness of a substance square with the well-known ‘placebo effect’? In reality, the ability for a placebo to be inert only needs to refer to the context of the particular effect expected. Therefore, taking a more comprehensive definition from the field of psychology (the discipline integral in exploring and formulating our understanding of the placebo effect), a placebo is “a substance or procedure that has no inherent power to produce an effect that is sought or expected” 9, and the placebo effect is “a genuine psychological or physiological effect attributable to a substance or undergoing a procedure, but not due to the inherent powers of that substance or procedure”. Taking these two broader statements into account allows us to appreciate that an invasive procedure with a non-active component can meet the criterion of being a placebo. The placebo effect is recognised as being inconsistent and not associated with a singular outcome. Various mechanisms make up this psychobiological phenomenon, and the different potential effects are dependent upon the disease context, the system and the therapeutic intervention being studied 10. For the purposes of the study by Kumar et al. 1, the effect being evaluated is that of placebo analgesia, the administration of a placebo that leads to pain relief and/or reduced analgesic use. Much of the research exploring the placebo effect has been conducted in the field of pain and analgesia. It is thought that the phenomenon of placebo analgesia is comprised of psychological mechanisms including expectation, conditioning and social learning, along with neurobiological mechanisms involving specific neurotransmitters such as endogenous opioids and dopamine, specific brain and spinal cord regions, and finally the molecular substrate effects of the placebo analgesic itself 11. As we have mentioned, Kumar et al. aimed not only to explore the analgesic effects of the stellate ganglion block but also the interaction between the autonomic nervous system and the nociceptive system. The rationale behind this study came from the original case series by McDonnell et al. who specifically targeted the stellate ganglion as a means of isolating the contribution of the sympathetic system to acute nociceptive pain 2. When designing clinical trials to evaluate analgesic medications or procedures, investigators need to account for the uncertainty principle. This refers to the possibility that the drug or intervention is not directly modulating the pain pathway, but instead influencing an aspect of placebo analgesia, for example the expectation pathways 12. This may not be important in studies purely evaluating clinical outcomes such as self-reported pain intensity scores or analgesic doses, but is an important confounding factor if investigating the potential underlying mechanisms of autonomic modulation of nociception. In this regard, Kumar et al. were required to include some methodological safeguard to account for this confounder. Therefore, whilst Cyna & Tan have criticised this study design for including an invasive procedure “to promote the non-specific, non-biological (placebo) treatment effect” 8, there is sound scientific justification for its inclusion. Their suggestion that the study should also have included a non-treatment control group is, however, well founded; this would have allowed the evaluation of the effect of the needle insertion per se. The broader philosophical question as to whether invasive placebo interventions are ever justified in medical research requires more space than is available in this editorial. However, one should also remember the considerably greater intervention of sham surgery, which has a well-established place in surgical research. A well known example is a randomised study of internal mammary artery ligation for relief of angina pectoris. Some patients received the surgical incision but without the arterial ligation; they showed improvement in exercise tolerance and symptoms, and some also demonstrated improvement in ECG patterns 13. Sham surgery continues to be seen as necessary for the evaluation of benefit from surgical interventions; last year the New England Journal of Medicine published a multicentre, randomised, double-blind, sham-controlled trial in 146 patients 14. The authors argued that this experimental design increased the rigour of their trial, which questioned the usefulness of arthroscopic partial meniscectomy. If it is considered that Kumar et al. should not have used an invasive placebo, then alternative experimental pathways should be suggested. Cyna & Tan propose that the investigators could have blinded the patient, anaesthetist and assessor by performing a stellate ganglion block in the intervention group after induction of general anaesthesia, and subsequently placing identical dressings on all participants. This would address in part the ‘open-hidden’ paradigm often implemented whilst exploring the placebo effect. In this circumstance, the patient is unaware whether the treatment has occurred or not by its being administered in a ‘hidden’ manner; however, for true implementation of this rationale, a comparison would need to be made against an ‘open’ group who knew that they were receiving the stellate ganglion block. In addition, it could be argued that the active group would be exposed to greater risk by having the nerve block performed under general anaesthesia rather than while awake 15. Is this ethically preferable to the invasive placebo? The paper by Kumar et al. has some limitations. One could argue that their invasive placebo is not a placebo in the truest sense. They have used saline, a solution that is – as far as we know – as ‘inert’ as possible when injected into tissues. However, they could have injected this into an area near but not within the stellate ganglion, avoiding some of the criticisms outlined by Cyna & Tan 8. Furthermore, the authors draw broad conclusions from somewhat limited data. By confusing the term neuroinflammation (a process that is an integral part of inflammatory pain) with neuropathic pain (pain caused by a lesion or disease of the somatosensory nervous system) 16 and using them interchangeably, their explanation of how the autonomic sympathetic nervous system modulates neuropathic pain may be flawed; the assumption that rest pain reflects only neuropathic pain is also without foundation. Notwithstanding these problems, Kumar et al. have conducted a well-designed study to answer a very specific and scientifically important question. Cyna & Tan present their opinion on invasive placebos, which lies at one end of a spectrum 8, argued against by other authors whom they quote. Individual clinicians or research ethics committees might take up differing and even contradictory positions on these matters, reflecting in part the balance between respecting study participants’ autonomy (the right to decide for themselves if properly informed) on the one hand, and the desire to protect them from risk (albeit paternalistically) on the other 17. We are now left with Kumar et al's results to fit into our understanding of how the autonomic nervous system may influence acute nociceptive pain and how this could aid in optimising pain treatment. No external funding and no competing interests declared.