Placebo responses and central neuropathic pain: from meta- to pooled analysis of clinical trial data

Defined as a reduction in pain without a biologically active therapeutic, the placebo response includes the placebo effect (i.e., psychobiological factors that cause a reduction in rating) in addition to other contributing factors, such as natural and expected fluctuations in pain [4]. To improve the design of future clinical trials, recent work has identified factors that predict placebo responses [3,5,8,10,11,13,15,17]. This has been done for several chronic pain conditions, and has uncovered a number of important relationships. For example, it is well established that people with higher pain ratings are also more likely to report greater reductions in pain [7,9].

The fundamental assumption of this area of research is that if placebo responses can be predicted, we can better design studies to maximize the detection of significant and clinically meaningful treatment effects. This is supported by the fact that “successful” clinical trials (i.e., those shown to significantly reduce pain) tend to have the smallest placebo responses [1].

In our recent study published in Pain, we focussed on understanding the placebo response among individuals with central neuropathic pain [6]. Central neuropathic pain arises due to damage in the spinal cord or brain (i.e., the central nervous system), and is common after stroke and spinal cord injury, and in patients with multiple sclerosis. Symptoms include continuous burning, as well as so-called “evoked pain” (e.g., feeling of pain to a stimulus that should otherwise not be considered painful). These symptoms are often very severe, negatively affect quality of life, and are very difficult to manage.

To address predictors of placebo responses in people with central neuropathic pain, we performed a meta-analysis. This type of analysis pools together the results from completed clinical trials, in order to produce an aggregate, overall estimate of an effect. To perform this analysis, we extracted pain ratings before and after treatment with placebo, as well as the number of male and females in the study, the average age of the group administered placebo, and characteristics of their pain. In total, 39 clinical trials involving individuals with neuropathic pain due to stroke, spinal cord injury, and multiple sclerosis were reviewed. Overall, we found significant reduction in pain due to placebo. Upon closer inspection, smaller placebo responses were seen for studies with a longer average duration of pain. In general, we think this observation fits with what is known about placebo responses: when individuals come to expect less benefit, placebo responses decrease [12,14]. In the case of central neuropathic pain, long durations may reflect, to some degree, that other medications have failed, and the individual now only expects a minimal benefit.

We also explored if damage in the spinal cord could interfere with placebo responses. We were interested in specifically addressing the role of the spinal cord because: 1) previous studies in healthy people have demonstrated an important role for the spinal cord in generating placebo, and 2) the research in our laboratory is focused on issues related to spinal cord injury ( Our hypothesis was that people with more severe damage in the spinal cord would have smaller placebo responses, owing to the disruption of central nervous tissue in the spinal cord. However, the severity of spinal cord injury was not an important factor: studies that included the most severe injuries demonstrated comparable placebo responses to studies incorporating less severe injuries. Therefore, based on the results of our meta-analysis, damage in the spinal cord does not change how a person responds to placebo treatment.

In theory, including only individuals with longstanding pain in clinical trials could serve as a strategy to yield lower placebo responses in randomized controlled trials. In turn, this could increase the likelihood of measuring a statistically significant treatment effect in the active treatment group. However, such a conclusion should be met with caution. First, a major concern is that patients with longstanding pain may also be the least likely to respond to an active treatment. Therefore, lowering placebo responses may inadvertently make it more difficult to assess treatment effects. Second, limiting inclusion into clinical trials to individuals with small placebo responses has dangerous implications in terms of the generalizability. The difficulty of generalizing trial outcomes increases as more specific and accurate criteria emerge to predict placebo responders. Eventually, treatments may only be tested and therefore proven effective in a very small cohort of patients.

Perhaps most importantly, there is the issue of the “ecological fallacy”. Meta-analyses operate on the assumption that group data reflect what is happening at the individual subject level. However, individual level relationships can be overlooked. This means, for example, that factors that did not predict placebo responses (e.g., injury severity), may actually be important to consider. The only way around the ecological fallacy is to consider raw clinical trial data. To date, surprisingly few analyses of raw placebo data from clinical trials has been performed [2,15,16]. The next step in our research will attempt to address the ecological fallacy by accessing completed clinical trials. We hope to include as many clinical trials involving individuals with central neuropathic pain as possible, both from industry sponsored and academic studies. A focus of this work will continue to investigate the potential role of the spinal cord in individuals with spinal cord damage, which we hope will yield new insights into the underlying physiology of placebo.

About the authors

John Kramer

John KramerDr. Kramer is an Assistant Professor in the School of Kinesiology at the University of British Columbia and Principal Investigator at ICORD (International Collaboration on Repair Discoveries).  Dr. Kramer has worked in the field of neuroscience, primarily in the area of spinal cord injury (SCI). The Kramer lab is chiefly focused on understanding mechanisms involved in central neuropathic pain after SCI, and the impact on neurological recovery. Dr. Kramer is a Michael Smith Foundation for Health Research and Rick Hansen Institute Scholar. Research in his lab is currently funded by support from Wings for Life and NSERC (National Science and Engineering Research Council of Canada).

Jacquelyn Cragg

cragg_jacquelynDr. Cragg is a Post-doctoral Fellow at the Harvard T.H. Chan School of Public Health and the University of British Columbia. Dr. Cragg is an experienced epidemiologist who specializes in neurological disorders such as spinal cord injury, Parkinson’s disease, ALS, and migraine. She is the ALS Canada Tim E Noël Postdoctoral Fellow, and also supported by the Michael Smith Foundation for Health Research.

Freda Warner

Freda WarnerMs. Warner is a PhD student in the School of Kinesiology at the University of British Columbia, and works in the lab of Dr. Kramer. After completing a Master’s degree in Public Health, she joined the Kramer lab to apply epidemiological techniques in the field of spinal cord injury, with an aim of understanding the effects of pain and pain management on neurological outcomes following spinal cord injury.


[1] Alexander MS, Anderson KD, Biering-Sorensen F, Blight AR, Brannon R, Bryce TN, Creasey G, Catz A, Curt A, Donovan W, Ditunno J, Ellaway P, Finnerup NB, Graves DE, Haynes BA, Heinemann AW, Jackson AB, Johnston MV, Kalpakjian CZ, Kleitman N, Krassioukov A, Krogh K, Lammertse D, Magasi S, Mulcahey MJ, Schurch B, Sherwood A, Steeves JD, Stiens S, Tulsky DS, van Hedel HJ, Whiteneck G. Outcome measures in spinal cord injury: recent assessments and recommendations for future directions. Spinal Cord 2009;47:582-591.

[2] Andreae MH, Carter GM, Shaparin N, Suslov K, Ellis RJ, Ware MA, Abrams DI, Prasad H, Wilsey B, Indyk D, Johnson M, Sacks HS. Inhaled Cannabis for Chronic Neuropathic Pain: A Meta-analysis of Individual Patient Data. J Pain 2015;16:1221-1232.

[3] Arakawa A, Kaneko M, Narukawa M. An investigation of factors contributing to higher levels of placebo response in clinical trials in neuropathic pain: a systematic review and meta-analysis. Clin Drug Investig 2015;35:67-81.

[4] Benedetti F. Mechanisms of placebo and placebo-related effects across diseases and treatments. Annu Rev Pharmacol Toxicol 2008;48:33-60.

[5] Cepeda MS, Berlin JA, Gao CY, Wiegand F, Wada DR. Placebo response changes depending on the neuropathic pain syndrome: results of a systematic review and meta-analysis. Pain Med 2012;13:575-595.

[6] Cragg JJ, Warner FM, Finnerup NB, Jensen MP, Mercier C, Richards JS, Wrigley P, Soler D, Kramer JL. Meta-analysis of placebo responses in central neuropathic pain: impact of subject, study, and pain characteristics. Pain 2016;157:530-540.

[7] Farrar JT, Troxel AB, Haynes K, Gilron I, Kerns RD, Katz NP, Rappaport BA, Rowbotham MC, Tierney AM, Turk DC, Dworkin RH. Effect of variability in the 7-day baseline pain diary on the assay sensitivity of neuropathic pain randomized clinical trials: an ACTTION study. Pain 2014;155:1622-1631.

[8] Hauser W, Bartram-Wunn E, Bartram C, Reinecke H, Tolle T. Systematic review: Placebo response in drug trials of fibromyalgia syndrome and painful peripheral diabetic neuropathy-magnitude and patient-related predictors. Pain 2011;152:1709-1717.

[9] Irizarry MC, Webb DJ, Ali Z, Chizh BA, Gold M, Kinrade FJ, Meisner PD, Blum D, Silver MT, Weil JG. Predictors of placebo response in pooled lamotrigine neuropathic pain clinical trials. Clin J Pain 2009;25:469-476.

[10] Irving G. The placebo response: relationship to outcomes in trials of postherpetic neuralgia. Clin Drug Investig 2010;30:739-748.

[11] Kamper SJ, Machado LA, Herbert RD, Maher CG, McAuley JH. Trial methodology and patient characteristics did not influence the size of placebo effects on pain. J Clin Epidemiol 2008;61:256-260.

[12] Lund K, Petersen GL, Erlandsen M, De Pascalis V, Vase L, Jensen TS, Finnerup NB. The magnitude of placebo analgesia effects depends on how they are conceptualized. J Psychosom Res 2015.

[13] Mbizvo GK, Nolan SJ, Nurmikko TJ, Goebel A. Placebo responses in long-standing complex regional pain syndrome: a systematic review and meta-analysis. J Pain 2015;16:99-115.

[14] Petersen GL, Finnerup NB, Grosen K, Pilegaard HK, Tracey I, Benedetti F, Price DD, Jensen TS, Vase L. Expectations and positive emotional feelings accompany reductions in ongoing and evoked neuropathic pain following placebo interventions. Pain 2014;155:2687-2698.

[15] Vase L, Vollert J, Finnerup NB, Miao X, Atkinson G, Marshall S, Nemeth R, Lange B, Liss C, Price DD, Maier C, Jensen TS, Segerdahl M. Predictors of the placebo analgesia response in randomized controlled trials of chronic pain: A meta-analysis of the individual data from nine industrially sponsored trials. Pain 2015.

[16] Vickers AJ, Cronin AM, Maschino AC, Lewith G, MacPherson H, Foster NE, Sherman KJ, Witt CM, Linde K, Acupuncture Trialists’ Collaboration. Acupuncture for chronic pain: individual patient data meta-analysis. Arch Intern Med 2012;172:1444-1453.

[17] Zhang W, Robertson J, Jones AC, Dieppe PA, Doherty M. The placebo effect and its determinants in osteoarthritis: meta-analysis of randomised controlled trials. Ann Rheum Dis 2008;67:1716-1723.

Commissioning Editors: Lorimer Moseley, Carolyn Berryman, Neil O’Connell 


  1. A quick reminder to folks who are commenting on BiM. As a courtesy to our authors and readers please keep your comments ON TOPIC and BRIEF.

  2. Unrelated bit of research here I thought was very good.

    I read this and I’m wondering what the future of acute injury management might look like: “Oh you hurt your back? Go home, have a few beers and forget about it! Drown your sorrows!”. A bit facetious, but really it’s not too much of a stretch since alcohol pumps your GABA pretty quickly and powerfully. I’d like to hear others’ comments on this paper and its implications.

    Graham Reply:

    The obvious problem when looking at GABA is that it is about 40% of synapses have GABA receptors .Groups of medications like benzodiazepines can have calming effects on anxiety but also having a lot of unwanted side effects where they act on other systems non selectively.
    We are at heart Complex Adaptive Systems that are bottom line trying to keep us alive and next attempting to utilize our energy as efficiently as possible.Pain is part of this process.
    I definitely feel that decatastrophizing “go home have a few beers and forget about it” can be a useful strategy for those with good levels of self efficacy but could be harmful to those with low levels and will probably see this as dismissive.
    Touch and education are to an extent working selectively on inhibiting GABA in the amygdala and hippocampus helping reframe your thoughts and allowing the environment to feel a safer place to explore.

    EG Reply:

    I agree, good quality touch and reassurance/confidence is preferred. But for those who can’t get to a therapist, maybe not a bad option.

  3. It’s refreshing to read such a clear and well written report. Clinically oriented.

    Can I make an appeal to researchers to shift focus a bit? In the study of placebo, I think we’d find much more value in research design which targets specific therapists who are known to ‘significanty outperform’. There’s a small group of people around the world who rank above the 99th percentile and they hold the clues.

    I’ve had a shoulder issue treated by one such ‘natural’, many years ago (a Physio). On the surface level, if you were to watch a video of the interaction, nothing much happened. He just spent a few minutes moving my arm above my head. But 20 minutes later I realized that something very significant had changed. No point going into detail, but it felt like a 6-sigma event! Later, I quizzed the hell out of him, and it was fairly clear he didn’t have any idea how it all worked, nor was he interested. He did however say one thing which stuck with me. He said: “when I’m working really well, I go into a sort of trance state”. This is from a guy whom most would consider very and ‘down-to-earth’. He almost winced as he said the word ‘trance’ because he never uses words like that. And the public know what’s up – they vote with their feet. This Physio in his prime, once saw over 100 patients in a day.

    Knowledge is great but it doesn’t heal. One might argue that it’s the willingness to let go of knowing that allows healing to occur.

  4. Perhaps a very big problem is bias. We know that lack of empathy/ therapeutic alliance reduces positive outcomes for acupuncture.It must be very difficult for researchers to dose empathy and perhaps skew it unwittingly in favour of the group they want to show the best response.
    May I recommend The Brain Science Poscasts with Francisco Benedetti for a really good overview(as well as other gems).