Placebo effects on cutaneous pain and itch: a systematic review and meta-analysis of experimental results and methodology

1. Introduction

Placebo effects, positive treatment outcomes for sensations such as pain and itch that arise through psychobiological mechanisms independent of an actual treatment,53,56 are routinely observed in clinical trials and practice.85,142 Their prevalence and magnitude likely vary across conditions and contexts, but these effects are thought to occur in many clinical trial participants receiving placebos,46 and magnitudes can vary extensively, from no effect to large effects.141Placebo effects are routinely studied in healthy participants,15,32 allowing for better-controlled investigation of the underlying mechanisms compared with research in clinical settings. Although these effects are most often studied in pain, itch is a similar but distinct sensation with overlapping neurobiological mechanisms,122 highly susceptible to psychological influence.2,100,126Placebo effects on itch routinely occur in the treatment of dermatological conditions,139 but their relation to placebo effects on pain is not well understood. A deeper investigation of the factors that shape the magnitude of both placebo effects on pain and itch will further our understanding of when and how these effects occur, and the mechanisms that underlie them.

In mechanistic placebo research, positive treatment expectations are typically induced using classical conditioning, verbal suggestions, observational learning, or a combination of these learning processes.28 Classical conditioning induces placebo effects by forming associations between an (inert) treatment and a decrease in sensation8,13; initially reinforced with a genuine reduction in sensation, the effect of which becomes associated with the inert treatment. For example, if one experiences pain relief every time they take a given medicine, they may come to expect pain relief from this medicine. Those expectations alone may be enough to foment some pain relief, such that if this person ingested a pill that they believed to be their analgesic medicine but was in fact a placebo, they would still experience some degree of pain reduction. Verbal suggestions explicitly provide positive information regarding the pain-relieving or itch-relieving effects of a treatment.140 This could come in the form of a doctor telling you that a new medicine will reduce your itch symptoms, inducing expectations for this outcome, which propagate some degree of itch relief on top of any biological effects of the treatment. Placebo effects can also be formed by observing the effects of a pain-relieving or itch-relieving treatment in another person.10,131Observational learning could, for example, form expectations for pain relief by seeing a friend's pain symptoms improve after trying a different physical therapy exercise. There seems to be an additive benefit to combining multiple learning processes when inducing these effects,12,28 although this has not been systematically reviewed.

One goal of experimental research into placebo effects on pain and itch has been to identify factors that influence these effects.98,143 Methods used in experimental placebo research are heterogeneous, varying factors such as the type of sensation (eg, thermal pain, electrical pain), the type of placebo intervention (eg, sham electrodes, gels, or pills), the number of acquisition and evocation trials used in a conditioning design,33,124 and the difference in intensity of pain stimulations between placebo and control trials.63 Demographic characteristics of study populations like sex and age may also potentially impact resulting placebo effects. Although some studies investigating sex differences in placebo responses have found that men are more responsive to verbal suggestions for placebo effects on pain,52,138 findings are mixed for classical conditioning and remain unexplored for itch. Age differences across the adult lifespan similarly have not been investigated for placebo effects on pain or itch. Systematic review and meta-analysis allows us to study what influence these methodological and demographic factors may have across studies. Previous meta-analyses of placebo effects on pain and itch have documented their widespread prevalence in clinical trials,139,142 and for pain, they demonstrated that mechanistic research tends to find larger placebo effects than those seen in clinical research. The use of longer pain stimuli was also associated with larger placebo effects.141 Since the most recent meta-analysis of mechanistic research into placebo effects on pain over a decade ago,141 numerous new studies have been published, particularly studies with healthy samples. To date, no meta-analyses have sought to quantify the magnitude of experimentally induced placebo effects on itch, nor have methodological factors been studied systematically as a potential source of heterogeneity in placebo effect magnitudes for pain or itch.

Given the growing body of research into placebo effects in cutaneous sensations, a systematic review and meta-analysis is warranted to provide insights into the distinct contributions of experimental components. Examining placebo effects across the literature may provide a better understanding of how these effects can be enhanced and potentially used in clinical settings, creating research avenues for novel therapies or informing doctor–patient communication. In pursuit of this aim and building on previous meta-analyses of similar scope,110,111,141,142 we conducted a systematic review and meta-analysis on the magnitude of placebo effects for inert treatments, in experiments on pain and itch, in healthy participants. First, we assessed the magnitude of placebo effects (defined as the decrease in pain or itch intensity after an inert treatment compared with a within-subject or between-subject control) by learning process (verbal suggestion, classical conditioning with verbal suggestion, and observational learning). To investigate the role of methodological and demographic factors, we then conducted subgroup analyses assessing the effect of the type of cutaneous sensation and the type of placebo intervention, and meta-regression to assess the impact of the number of learning and evocation trials used in classical conditioning models, the difference in calibrated intensity of placebo and control stimulations, sex distribution, and mean age of the participants on placebo-effect magnitudes.

2. Methods 2.1. Protocol and registration

The protocol for this study was preregistered on ClinicalTrials.gov (ID: NCT04387851) and was conducted following PRISMA guidelines106 (Supplementary digital content, available at https://links.lww.com/PAIN/B752). We registered a single search strategy for placebo and nocebo studies, the results of which have been divided into 2 articles after evaluating the amount of articles yielded by the search, facilitating a more clear and nuanced discussion for each set of findings. Here, we report on the placebo studies.

2.2. Databases and selection criteria

PubMed, PsycINFO, EMBASE, and the Cochrane CENTRAL Methodology Library were searched to identify studies. Languages were limited to English, Dutch, and German, and the publication period was not restricted. Searches were initially conducted on March 18, 2019, and subsequently updated on April 10, 2020, and July 15, 2021. The complete key-worded search strategy for each database is available in the supplementary digital content (available at https://links.lww.com/PAIN/B752).

We searched for original, controlled experimental studies on healthy participants that aimed to experimentally induce placebo or nocebo effects on cutaneous sensations (ie, pain or itch stimulations that were administered on the skin); of which, the results of studies on placebo effects are reported here. Patient samples were not included because the current review focuses on learning mechanisms and methodological factors, which can be studied with better experimental control in healthy samples. For better homogeneity of study designs, we focused only on cutaneous sensations, and excluded, eg, studies on visceral or ischemic sensation. For the purposes of inclusion and exclusion, studies were considered to have induced a placebo effect if a learning mechanism (eg, conditioning, verbal suggestion, observational learning) was used to induce positive expectations about an inert treatment and not to purely ambiguous stimuli (eg, colored shapes). This was done to focus the scope of this review on experimental studies, which induced expectations around treatments as opposed to abstract stimuli, thereby improving the clinical relevance of the meta-analyses. We only included studies that featured some form of control comparator, whether that was within or between subjects, so that the placebo effect could be calculated as the difference between placebo and control. Studies that excluded nonresponders from the analyses were excluded. Studies that did not fulfill one or more of the criteria mentioned above were excluded from further review and meta-analysis. Our search terms did not include words specifically intended to collect observational learning studies because we did not originally plan to investigate this learning mechanism in our preregistration. Still, our search identified observational learning studies, and we decided to include them in our review because we likely identified all relevant observational learning studies with reference list and Web of Science searches.

2.3. Study selection

Titles and abstracts of articles retrieved using the above search strategy were independently screened by 2 authors (J.S.B. and M.M.E.V.S). The full text of articles to be included and articles about which doubts existed were then retrieved and assessed for eligibility by 2 authors independently (J.S.B. and M.A.T.). The reference lists of all included articles were also screened for study inclusion by one author (J.S.B.) and a student assistant, and included articles were also entered in Web of Science to identify articles that have cited them and should potentially be included in the meta-analysis in April, 2020. When full texts were not available online, authors were contracted through email to request access. Disagreements concerning study inclusion decisions were resolved by a third author (K.J.P).

2.4. Data extraction

One author (J.S.B.) used a standardized form to independently extract data from the included studies to derive study characteristics and data for analyses. Another author (M.A.T.) checked 25% of extracted values for accuracy. Extracted information included details of the experimental induction (ie, learning mechanism used), control condition, study population, placebo treatment, sensation type, pain/itch outcome data, how sensations were measured (eg, 0-10 numeric rating scale, visual analogue scale, 0-20 Gracely scale, etc), type of cutaneous stimulation (eg, heat pain, pressure pain, histamine-evoked itch), information for quality and bias assessment, and outcome data for meta-analysis (eg, sample size, pain/itch rating means and standard deviations). Doubts regarding data extraction were resolved through discussion with a third author (K.J.P.). Missing data were requested from the authors of included studies. If the authors did not respond, but data could be extracted from published figures, this was done with the software WebPlotDigitizer version 4.4 (Rohatgi, 2020).

2.5. Risk of bias 2.5.1. Risk of bias assessment within studies

Risk of bias was assessed by a student assistant and one of the authors (M.A.T.), independently from one another, using the method developed by Marcuzzi et al.,91 specifically for quantitative sensory testing studies. This method assesses (1) whether the inclusion criteria were clearly described (3 items), (2) whether the sample is clearly described and representative of the population (5 items), (3) whether the recruitment process was clearly described (3 items) (4) whether the somatosensory assessment methods are standardized, validated, and well described (6 items), (5) adequate blinding if relevant (1 item), and (6) whether potential confounders were considered (2 items). Items were scored as satisfied (0 points), not satisfied (2 points), partially satisfied or unclear (1 point), or not applicable. Studies receive a score ranging from 0 to 40 based on these criteria, with higher scores indicating a greater risk of bias. Meta-regression was used to test for a relationship between risk of bias score and the magnitude of the placebo effect. An example of the risk of bias tool can be found in the supplementary digital content (available at https://links.lww.com/PAIN/B752).

2.5.2. Risk of publication bias across studies

Risk of publication bias across studies was assessed visually with funnel plots. Studies lying outside the funnel of expected results were included in subsequent analyses, but their outlier status was noted in the study characteristics table (Tables 1–4). Publication bias was assessed with Duval and Tweedie's trim and fill method,47 a nonparametric technique for estimating the number of missing studies in a meta-analysis and the impact these studies would likely have on the overall effect size.

Table 1 - Characteristics of studies included in the classical conditioning with verbal suggestion (pain) meta-analysis. Author Year N Sample age Percent female Sensation induction method Placebo manipulation Rating scale Acquisition trials (Placebo/Control) Evocation trials (Placebo/Control) Calibrated stimulus intensity difference (0-100)* First or mean outcome measure Risk of bias score Comparison Outlier based on funnel plot Au Yeung7 2014 20 19.8 59% Electrical Sham TENS 0-100 VAS 32 (16P/16C) 32 (16P/16C) NA First 3 W Barnes11 2021 62 19.4 52% Electrical Sham TENS 0-10 pain intensity 30 (15P/15C) 20 (10P/10C) 45 Mean 3 W Case21 2019 28 NR 53% Thermal Inert gel 0-80 pain intensity 8 (4P/4C) 16 (8P/8C) NA Mean 7 W − Choi23 2011 15 25.3 0% Electrical Sham IV 0-100 NRS Unknown 10 (5P/5C) NA Mean 3 W Chouchou24 2015 26 23.4 46% Thermal Inert gel 0-100 VAS 16 (8P/8C) 10 (5P/5C) 35 Mean 4 W Colagiuri27 2018 21 20.2 71% Electrical Sham TENS 0-100 VAS 32 (16P/16C) 32 (16P/16C) NA First 5 W Colloca30 2006 10 22.7 83% Electrical Sham electrode 0-10 NRS 36 (18P/18C) 12 (6P/6C) NA Mean 5 W + Colloca34 2008 15 22.5 100% Electrical Sham electrode 0-10 VAS 24 (12P/12C) 12 (6P/6C) NA Mean 3 W Colloca35 2008a 16 32.0 66% Laser Inert gel 0-10 NRS 30 (15P/15C) 30 (15P/15C) NA Mean 5 W Colloca31 2009 16 22.6 100% Electrical Sham electrode 0-10 NRS 24 (12P/12C) 12 (6P/6C) NA Mean 5 W Colloca33 2010 46 22.8 65% Electrical Sham electrode 0-10 VAS 20 (10P/10C) 40 (20P/20C) 30 Mean 3 W + Colloca33 2010 46 22.8 65% Electrical Sham electrode 0-10 VAS 80 (40P/40C) 40 (20P/20C) 30 Mean 3 W + Colloca36 2019 53 28.1 64% Electrical Sham electrode 0-10 NRS 18 (9P/9C) 36 (18P/18C) 60 Mean 3 W Colloca29 2020 400 29.4 59% Thermal Sham electrode 0-100 VAS 24 (12P/12C) 12 (6P/6C) NA Mean 5 W Corsi37 2017 46 27.4 52% Thermal Sham electrode 0-100 VAS 12 (6P/6C) 6 (3P/3C) NA Mean 3 W de Jong39 1996 36 21.3 100% Electrical Inert gel 0-100 VAS 20 (10P/10C) 10 (5P/5C) 25 Mean 5 B De Pascalis40 2002 36 25.4 65% Electrical Inert gel 0-10 VAS 12 (6P/6C) 30 (15P/15C) NA Mean 2 W De Pascalis43 2021 56 23.3 100% Cold cup Inert gel 0-100 NRS 2 (1P/1C) 2 (1P/1C) NA Mean 4 W Egorova48 2020 24 NR 50% Thermal Inert gel 0-20 Gracely scale 48 (24P/24C) 24 (12P/12C) 25 Mean 5 W Eippert49 2009 19 25.0 0% Thermal Inert gel 0-100 VAS 12 (6P/6C) 30 (15P/15C) 40 Mean 6 W Eippert50 2009a 13 25.0 0% Thermal Inert gel 0-100 VAS 12 (6P/6C) 30 (15P/15C) 40 Mean 0 W Feldhaus55 2021 624 24.6 60% Thermal Inert gel 0-100 VAS 16 (8P/8C) 16 (8P/8C) 40 Mean 3 W Flaten57 2018 25 21.9 56% Thermal Inert pill 0-10 NRS 3 (2P/1C) 2 (1P/1C) NA Mean 0 W Frangos58 2021 46 39.7 85% Thermal Inert gel 0-200 VAS 24 (12P/12C) 20 (10P/10C) 45 Mean 3 W Freeman59 2015 24 NR 50% Thermal Inert gel 0-20 Gracely scale 18 (9P/9C) Unknown 25 Mean 5 W Gaab60 2019 81 25.2 60% Thermal Inert gel 0-10 VAS 16 (8P/8C) 4 (2P/2C) 30 Mean 3 W Geisler62 2020 33 27.4 0% Thermal Inert gel 0-100 VAS 16 (8P/8C) 8 (4P/4C) 40 Mean 3 W Geuter63 2013 40 26.0 0% Thermal Inert gel 0-100 VAS 24 (12P/12C) 30 (15P/15C) 50 Mean 3 W Geuter63 2013 40 26.0 0% Thermal Inert gel 0-100 VAS 24 (12P/12C) 30 (15P/15C) 30 Mean 3 W Grahl66 2018 23 24.6 0% Thermal Sham TENS 0-100 VAS 24 (12P/12C) 24 (12P/12C) 40 Mean 0 W Hartmann67 2021 45 23.8 51% Electrical Inert gel 0-8 pain intensity Unknown 32 (16P/16C) 30 Mean 1 W Huneke74 2013 73 37.6 66% Laser Inert gel 0-10 NRS 20 (10P/10C) 20 (10P/10C) 40 Mean 3 B Jarcho77 2016 15 24.3 100% Thermal Inert gel 0-100 VAS 2 (1P/1C) 2 (1P/1C) NA Mean 4 W Kirsch79 2014 48 26.4 50% Thermal Sham acupuncture 0-20 Gracely scale Unknown Unknown 40 Mean 5 B Klinger80 2007 12 26.1 50% Electrical Inert gel 0-8 pain intensity 10 (5P/5C) 10 (5P/5C) 25 Mean 8 W Kong82 2006 16 28.4 44% Thermal Sham acupuncture 0-20 Gracely scale 48 (24P/24C) 24 (12P/12C) 40 Mean 5 W Laverdure-Dupont85 2009 38 23.4 58% Thermal Inert gel 0-100 VAS 16 (8P/8C) 10 (5P/5C) 20 Mean 3 W Lee86 2020 21 23.6 43% Pressure Inert gel 0-100 VAS 12 (6P/6C) 12 (6P/6C) NA Mean 5 W Lui89 2010 31 23.5 58% Laser Sham electrode 0-100 VAS 24 (12P/12C) 12 (6P/6C) NA Mean 7 W Martin95 2010 40 21.2 70% Thermal Inert gel 0-10 NRS 16 (8P/8C) 2 (1P/1C) 30 Mean 5 W

留言 (0)

沒有登入
gif