Effects of open-label placebos across outcomes and populations: An updated systematic review and meta-analysis of randomized controlled trials

doi:10.21203/rs.3.rs-5216072/v1

Download PDF

Article

Effects of open-label placebos across outcomes and populations: An updated systematic review and meta-analysis of randomized controlled trials

https://doi.org/10.21203/rs.3.rs-5216072/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

This work synthesizes and updates findings from four previous systematic-reviews and meta-analyses on open-label placebos (OLPs). For the first time, it directly tests whether OLPs have different effects on self-report versus objective outcomes and on clinical versus non-clinical samples. We searched eight data-bases up to November 9, 2023, and included 58 randomized controlled trials (RCTs), compromising 61 separate comparisons. OLPs yielded a small positive effect across various health-related conditions (k = 61, n = 4569, SMD = 0.36, 95% CI = 0.26; 0.45, p < 0.0001; I² = 56%). The OLP effect differed between self-report outcomes (k = 55, n = 3171, SMD = 0.40) and objective outcomes (k = 14, n = 1176, SMD = 0.02, Q = 9.81, p < 0.01), and between clinical samples (k = 22, n = 1398, SMD = 0.48) and non-clinical samples (k = 39, n = 3,171, SMD = 0.29, Q = 4.69, p < 0.05). Neithter the level of suggestiveness nor the type of control moderated the effect. These findings confirm on a large dataset that OLPs are effective for self-report but not for objective outcomes. They also point to potential differences in effectiveness and mechanisms in comparsion to deceptive placebos.

Biological sciences/Psychology

Health sciences/Medical research

Health sciences/Medical research/Outcomes research

open-label placebo

objective outcome

meta-analysis

systematic-review

The placebo effect is a well-known psychological phenomenon that can lead to significant improvements in both clinical and non-clinical populations for self-report outcomes (e.g., self-rated questionnaires) and objective outcomes (e.g., physiological, or behavioral variables). Placebos have been shown to have beneficial effects in conditions such as allergies, anxiety, Alzheimer’s disease, depression, fatigue, pain, physical performance, and various physiological systems including cardiovascular, respiratory, immune, and endocrine systems^1–3. However, traditional placebos, which involve deceiving recipients about their treatment, raise ethical concerns and limit their applicability in clinical and general practice². An alternative to deceptive placebos is the use of open-label placebos (OLPs), where recipients are informed transparently that they are receiving a placebo with no pharmacologically active ingredients⁴. Since the pioneering OLP trial by Sandler and Bodfish of 2008⁵, numerous studies have explored various aspects of OLPs⁶.

The meta-analysis by Spille et al. (2023)⁷ found a moderate effect of OLP interventions on self-report outcomes in non-clinical samples (k = 17; standardized mean difference [SMD] = 0.43), but no effect on objective outcomes (SMD = 0.02). In clinical samples, earlier meta-analyses by Charlesworth et al. (2017)⁸ (k = 5; SMD = 0.88) and von Wernsdorff et al. (2021)⁹ (k = 12; SMD = 0.72) reported moderate to large effects of OLPs compared to control groups. Buergler et al. (2023)¹⁰ conducted two network meta-analyses, one on clinical and one on non-clinical samples. The non-clinical one identified a moderate effect of OLPs administered via nasal sprays (SMD = 0.43), while other forms of interventions (e.g. dermal creams, pills) did not show signifanct effects. In the clinical part conditioned OLP pills (SMD = 0.89) and standard OLP pills (SMD = 0.46) showed significant effect sizes, while injections and suspensions did not.

These findings highlight the need to consider both the form of outcome (self-report vs. objective) as well as the population type (clinical vs. non-clinical) when evaluating the effectiveness of OLPs. However, none of the existing meta-analyses directly compared self-report and objective outcomes or synthesized clinical and non-clinical samples. No meta-analysis has yet quantified the overall effect of OLPs across both self-report and objective outcomes as well as across clinical and non-clinical populations.

A key concept in OLP research is the induction of treatment expectations through verbal suggestions when administering the OLP treatment¹¹. In particular, OLPs accompanied by a rationale highlighting the treatment’s positive effects are believed to be more effective than those without suggestive elements¹¹. Most OLP studies use a rationale inspired by Kaptchuk et al. (2010)¹², which included the following points: (1) the placebo effect is powerful, (2) the body is automatically responding to placebos, (3) a positive attitude towards palcebos is not necessary, and (4) adherence to the placebo regime is important. Buergler et al. (2023)¹⁰ found that OLP interventions with a suggestive rationale were more effective than those without such a rationale¹⁰. However, this network meta-analysis examined this question on a limited number of trials with minimal expectation induction (two non-clinical and three clinical). Additionally, Spille et al. (2023) observed an effect of suggestiveness only on objective outcomes, but not on self-report outcomes in non-clinical populations⁷. Thus, further research with larger number of trials is needed to clarify whether the level of suggestiveness influences the OLP effectiveness across different outcome types. Moreover, the impact of suggestive treatment rationales in OLPs has been explored in only a few studies.

All meta-analyses mentioned previously examined the effect of OLPs across various control conditions. However, research in depression, for instance, has shown that treatment effect sizes vary depending on the type of control condition used. Psychotherapy interventions, when compared to waiting list (WL) control groups, tend to show greater effectiveness than when compared to no treatment (NT) control groups¹³. Buergler et al. (2023) found that OLP interventions were superior to NT and WL but not to treatment as usual (TAU) in clinical populations¹⁰. In non-clinical populations, Spille et al. (2023) proposed a novel control group called covert placebo (CP)⁷. In a CP condition, participants receive the same physical treatment as in the OLP condition, but the rationale for the treatment is designed to divert attention from the placebo’s mechanism (e.g., participants are given a technical explanation for receiving the placebo treatment). Unlike deceptive placebos, CPs aime to avoid creating specific expectations in recipients about the treatment’s outcome. Spille et al. (2023) suggested that CPs could serve as an alternative to NT for assessing OLP effects in non-clinical samples⁷. However, the limited data available from Spille et al. (2023) necessitates a reevaluation of their findings⁷.

In summary, to address these unresolved questions, we conducted a systematic review and meta-analysis, updating and synthesizing the research by Buergler et al. (2023), Spille et al. (2023) and von Wernsdorff et al. (2021)^7,9,10. We derived the following research questions (RQs): First, do OLP interventions offer greater benefits compared to control conditions across various outcomes and populations (RQ1)? Second, do these benefits differ between self-report and objective outcomes (RQ2)? Third, do these benefits differ between clinical and non-clinical populations (RQ3)? Fourth, do the level of expectation induction through suggestion influence the OLP effect (RQ4)? Fifth, do effect sizes vary based on the type of control group used (RQ5)?

This systematic review and meta-analysis utilized the same databases and search terms as Spille et al. (2023)⁷ and von Wernsdorff et al. (2021)⁹. The study adhered to the revised Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement¹⁴ (Supplemental digital appendix A1). We pre-registered the study with PROSPERO (#ANONYMIZED) prior to data collection.

Eligibility criteria

We included randomized controlled trials (RCTs) that investigated the effects of OLPs on various health-related outcomes in both clinical and non-clinical populations. The OLP intervention could include any form of intert substance without pharmacological activity, such as placebo pills, capsules, nasal sprays, dermatological creams, injections, acupuncture, or verbal suggestions¹⁵. It was essential that the OLP was administered transparently, with recipients fully informed that they were receiving a placebo^6,15. OLPs had to be compared with one of the following control conditions: CP, NT, TAU, or WL control condition. We included only trials reporting outcomes on a continuous scale, whether self-report (e.g., self-rated questionnaire) or objective measure (i.e., physiological, or behavioral variables). In accordance with the Cochrane Handbook for Systematic Reviews¹⁶, we included crossover trials only if data from the initial phase of the trial (i.e., prior to the crossover) were available. We contacted authors, if this data was not reported. If the authors could not provide the data or did not respond, we excluded the trial from the analyses.

Information sources and search strategy

On November 9, 2023, we conducted a comprehensive systematic review in multiple databases: EMBASE via Elsevier, MEDLINE via PubMed, APA PsycINFO and PSYNDEX Literature with PSYNDEX Tests via EBSCO, the Web of Science Core Collection, and the most recent edition of the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library, Wiley). We used the search terms identical to those employed by von Wernsdorff et al. (2021)⁹ and Spille et al. (2023)⁷, focusing on variants of ‘open-label placebo’ (e.g., ‘placebo’, ‘open-label’, ‘non-blind’, and ‘without deception’). The number of hits for each search term in these databases is provided in Supplemental digital appendix A2. We restricted our search to publications from January 2020 onward (i.e., the search date of von Wernsdorff et al., 2021). For non-clinical samples, we included trials published from April 15, 2021 (i.e., the search data of Spille et al., 2023). Additionally, we screened all entries in the Journal of Interdisciplinary placebo Studies DATABASE (JIPS, https://jips.online/) from January 2020, as well as the complete database in the Program in Placebo Studies & Therapeutic Encounter (PiPS, http://programinplacebostudies.org/), using keywords in the publication titles (e.g. ‘analgesia’, ‘expectation’, ‘non-deceptive’, ‘open-label’, ‘placebo’, ‘suggestion’). No restrictions were applied regarding the language of publication or the age of participants.

Selection process

Results from the literature databases, including hits from the JIPS and PiPS databases, were exported into the systematic review management software Rayyan¹⁷. Duplicates were removed using Rayyan’s semi-automatized duplicate-detection feature. Two researchers (#ANONYMIZED and #ANONYMIZED) independently assessed study eligility. First, they screened all titles and abstracts, followed by a full-text assessment of reports deemed potentially eligible in the first stage by one of the investigators. Disagreements were resolved through discussion, with the supervision of two additional researchers (#ANONYMIZED and #ANONYMIZED).

Data items and collection process

Two researchers (#ANONYMIZED and #ANONYMIZED) independently extracted data from the included studies in a standardized Excel form, which was piloted with three records. Data extraction covered the following areas: study details (i.e., author, year, title), sample characteristics (i.e., type of population, population size, distribution of participants), intervention and control conditions (e.g., pill, cream, spray, duration), outcomes (i.e., baseline, post-intervention or change scores), and outcome form (i.e., self-report and objective). We extracted data for the outcome defined as primary outcome in the respective report. If a primary outcome was not specified, we extracted all outcomes related to the OLP intervention to minimize bias from selective outcome choice based on effect size and hypothesis fit¹⁸. In trials where baseline outcome scores were unavailable prior to experimental exposure, we extracted post-intervention values only.

Participants were classified into clinical populations if they met a medical condition (e.g., allergic rhinitis, cancer-related fatigue, chronic low back pain, menopausal hot flashes) or a mental disorder (e.g., major depressive disorder), as diagnosed by a clinician or psychologist¹⁹. Those classified as non-clinical populations were generally healthy individuals. Subclinical traits, such as test anxiety or low levels of well-being, did not qualify participants for the clinical population and were therefore categorized as non-clinical. The degree of suggestiveness of the treatment rationale was independently evaluated by two researchers (#ANONYMIZED and #ANONYMIZED) based on the description of the OLP administration. Rationals of OLP interventions featuring one or more elements from Kaptchuk et al. (2010, see Introduction)¹² were rated as having a ‘high’ degree of suggestiveness, while those lacking elements of suggestive expectation induction were rated as having a ‘low’ degree of suggestiveness. In studies with different treatment rationales across separate intervention groups, both groups were extracted as distinct trials and coded accordingly as OLP+ (‘high’ degree of suggestiveness) and OLP- (‘low’ degree of suggestiveness). To ensure comparability across clinical trials, all control conditions (i.e., NT, TAU, WL, and CP) were checked for compliance with given definitions of comparators²⁰. Thus, NT refers to a condition in which no alternative treatment is provided, while TAU included access to standard treatment practices for the condition. If participants were offered an OLP intervention following the OLP group, the control condition was labeled as a WL condition. For non-clinical trials, NT refers no intervention, and CP involved the same physical treatment as the OLP condition, but with a rationale designed to avoid creating specific expectations regarding the outcome⁷. Missing values were addressed by contacting the authors. If the authors did not responed or could not provide the data, the study was excluded. All extracted data were cross-checked using Excel's data validation feature. Discrepancies were resolved through discussion and consensus, supervised by (#ANONYMIZED and #ANONYMIZED).

The study at hand is an extended update of the reviews by von Wernsdorff et al.⁹ and by Spille et al (2021)⁷. Therefore, one reviewer (#ANONYMIZED) re-evaluated and extracted data form the studies included in these reviews. In cases of discrepancies regarding selection and extraction due to methodological differences (e.g., von Wernsdorff et al.⁹ extracted consistently post-intervention values only), a second reviewer (#ANONYMIZED ) independently re-assessed these reports. In cases where changes were necessary, both reviewers (#ANONYMIZED and #ANONYMIZED) independently extracted any additional data. The same data cross-checking process, as illustrated above, was applied to ensure accuracy.

Risk of bias assessment

We assessed the risk of bias of the included trials using the Cochrane risk of bias tool (RoB 2.0)²¹. The RoB 2 assesses bias arising from the randomization process (domain 1), deviations from intended interventions (domain 2), missing outcome data (domain 3), measurement of the outcome (domain 4), and selection of the reported result (domain 5)²¹. The results of the five domains are aggregated into an overall risk of bias rating, which is equivalent to the worst rating in any of the domains. In line with the previous meta-analyses on OLPs^7,9,10, we applied the same special rules to the RoB 2 to account for the unblinded nature of OLPs. When a rating of a ‘high’ risk of bias in domain 4 resulted only due to the signaling question 4.5 (‘Is it likely that the assessment of the outcome was influenced by knowledge of the intervention received?’), we overrode the suggestion of the algorithm for this domain and labeled it with ‘some concerns’. Since eliciting treatment expectation is a crucial mode of action of OLPs, the effect of knowing about the group allocation cannot be separated from the placebo or nocebo effect (i.e., excitement or disappointment respectively)⁹. Moreover, we would have lost all variance in the assessment, as all studies would have received a ‘high’ overall rating as consequence. The RoB 2 for the newly included studies was assessed by two researchers independently (#ANONYMIZED and #ANONYMIZED), with discrepancies resolved through discussion and consensus. The results from assessments of the studies included in previous versions of the review by Spille et al. (2023)⁷ and von Wernsdorff et al.(2021) ⁹ are presented in the Supplemental digital appendix A4.

Data synthesis and analysis

Statistical analyses were performed using R, version 4.3.2²². To evaluate the effects of the OLP interventions, we calculated standardized mean differences (SMDs) by subtracting the mean pre-post change in the intervention group from the mean pre-post change in the control group and dividing the result by the pooled pre-intervention standard deviation²³. For studies that reported only post scores, SMDs were calculated based on these values. Standard errors of the means were converted into SDs following the guidelines outlined in the Cochrane Handbook¹⁶. We used Hedges’ g which corrects for bias due to small sample sizes²⁴. We interpreted values of 0.20, 0.50, and 0.80 as small, moderate, and large effect, respectively²⁵. In studies without defined primary outcome^26–50, we initially computed SMDs for each of the outcomes independently, which were then averaged to obtain an overall SMD estimate for the respective trial⁵¹. This was done by the aggregate function in R from the metafor package⁵², under the assumption of an intra-study correlation coefficient of ρ = 0.6 ⁵³. In studies where the type of administration of the OLP intervention varied (e.g., one group received OLP nasal spray and another intervention group received OLP pills)^{28,37,42,54,55}, the mean pre- and post-values along with their SD were aggregated prior to data analysis, in accordance with the guidelines outlined in the Cochrane Handbook¹⁶.We aggregated the interventions groups of these trials as follows: for Barnes et al. (2019)⁵⁵, we combined the ‘Semi-Open Label’ and ‘Fully-Open Label’ intervention groups; for El Brihi et al. (2019)³³, we combined the groups receiving one and four OLP pills per day; for Kube et al. (2020)³⁷, we aggregated the ‘OLP-H’ (hope) and ‘OLP-E’ (expectation) groups; for Olliges et al. (2022)⁴², we combined the ‘OLP pain’ and ‘OLP mood’ groups; and for Winkler et al. (2023)²⁸, we merged the ‘OLP nasal active’, ‘OLP nasal passive’, and ‘OLP capsule’ groups In studies where the OLP rationale was manipulated accross two independent samples^31,41,56, the SMDs were calculated separately for each intervention group (OLP + and OLP- respectively) and treated as two distinct trials. To prevent unity-of-analysis error due to double counting, in these cases the control group data was divided in half for each intervention group¹⁶. Similarly, if a trial contributed to more than one comparison (i.e., assessing both self-report and objective outcomes)^{26,30,35,37,41,48,57}, again, the sample size was divided by the number of comparisons to prevent unit-of-analysis error due to double-counting. To ensure consistent interpretation of the direction of effects where a positive SMD value indicates a beneficial effect of the OLP intervention for recipients, the means of some studies were multiplied by -1, as outlined in the Cochrane Handbook¹⁶.

After aggregating all effect sizes within studies, we conducted a meta-analysis using the meta package in R⁵⁸. Anticipating heterogeneity among trials, we employed a random-effects model with the inverse-variance weighting method⁵⁹. We weighted the trials based on post-intervention sample sizes, which are more conservative than pre-intervention samples sizes. All tests were two-tailed. Heterogeneity among studies was assessed using the Q statistic and quantified with the I² index as well as prediction intervals. The I² values indicate the percentage of total variance between effects that is due to true effect variation and are interpreted as follows: 0 to 40% might not be important; 30 to 60% may represent moderate heterogeneity; 50 to 90% may represent substantial heterogeneity; and 75 to 100% is considered to constitute considerable heterogeneity¹⁶. The prediction intervals indicate true effect variation and represent the range into which the true effect size of all populations will fall⁶⁰. If the prediction interval lies entirely on the positive side (i.e. does no include the zero), favoring the OLP intervention, it suggests that, despite variations in effect sizes, the OLP is likely to be beneficial; however, is important to note that broad prediction intervals are relatively common and reflect inherent variability in the data⁵³.

Sensitivity analysis

We conducted four sensitivity analyses to examine the robustness of results. First, we excluded outliers utilizing the “non-overlapping confidence intervals” approach, which regards a comparison as outlier, if its 95% confidence interval of the effect size did not overlap with the 95% confidence interval of the pooled effect size⁵³. Second, we excluded studies assessed with a ‘high’ risk of bias according to the Rob 2 assessment. We chose the criterion ‘high’, because almost all studies had at least a risk to ‘some concerns’ due to the lack of blinding of the OLP interventions and to self-reported outcomes. Third, we considered potential publication bias using Duval and Tweedie’s trim and fill procedure, which provides an estimate of the pooled effect size after adjusting for asymmetry in the funnel plot⁶¹. Fourth, we estimated the overall OLP effect using a hierarchical three-level meta-analytic model, with effect sizes nested in studies⁵³.

Reporting bias assessment and certainty assessment

We assessed publication bias by creating a funnel plot, which plots effect estimates (SMDs) from individual studies against their standard error (SE). We visually inspected the funnel plot for asymmetry and conducted Egger’s regression test, which regresses the SMDs against their SE⁶².

We included 58 RCTs with 61 seperate comparisons. Of these, 10 trials were from von Wernsdorff et al. (2021)⁹, 17 from Spille et al. (2023)⁷, and 31 were identified in our updated search (for a flow chart see Fig. 1). The chance-corrected agreement between raters after the full-text screening in the updated search was substantial (κ = 0.78)⁶³. We included two individual studies reported in one article meeting the eligibility criteria⁵⁵, which were excluded in Spille et al. (2023)⁷ due to missing values. Here, we obtained the necessary data from Buergler et al. (2023)¹⁰. One trial identified through PiPS⁶⁴ had not been included in any prior systematic reviews. Four records reported the results of two separate experiments with independent samples, both of which were included separately in the analysis^34,43,55,65. We excluded several eligible studies from the meta-analysis for specific reasons: two authors did not respond on questions on missing outcome data^66,67; data from two crossover studies were unavailable^68,69; and two studies were follow-ups of included trials^70,71, where we used the parent trial instead. The follow-up studies were excluded because they only re-investigated the intervention group⁷⁰, or included participants who switched to an OLP intervention after the initial trial (both studies). Other exclusions involved studies lacking randomization^72,73; differences in the treatment dosage between the OLP and the control group condition⁷⁴; and the absence of a suitable control group⁷⁵. Moreover, we excluded a treatment arm in Rathschlag & Klatt (2021)⁴³ because the author could provide only raw outcome data. Furthermore, we excluded the previously included study by Sandler and Bobfish (2008)⁷⁶ in von Wernsdorff et al. (2021)⁹ because the control group did not meet the TAU definition implemented in the present analysis.

Trial characteristics

Detailed qualitative characteristics of included trials are presented in Supplemental digital appendix A6, and detailed numerical characteristics are presented in Supplemental digital appendix A7. The trials were published in English between 2001 and 2023. The 58 RCTs (37 non-clinical and 21 clinical trials) involved a total of 4,677 participants, with 2,498 individuals receiving an OLP intervention and 2,279 individuals in the control conditions. The mean age of participants weighted for sample size was 30.7 years (SD = 8.5) with an age range from 18 to 70 years. The mean percentage of females weighted for sample size was 73%. Sample sizes ranged from 20 to 177 individuals. The duration of the trials was from one to 90 days, with an average duration of 13.5 days (SD = 17.3). Twenty-five studies were conducted in Germany^{28,30,37–40,42–46,48,56,64,65,77–84}, 11 studies in the USA^{12,34,36,85–92}, six studies each in Australia^{31,33,50,55,93} and Austria^{27,47,57,94–96}, four in Switzerland^41,49,54,97 and one in Israel⁹⁸, Japan²⁶, the Netherlands⁸⁹, New Zealand²⁹, Portugal³², and the United Kingdom³⁵, respectively.

Twelve non-clinical trials investigated the protective effects of OLPs on experimentally induced emotional distress of various types: six used distressing pictures^34,57,65,94, three experimentally induced sadness^77–79, one guilt⁹⁷, one intrusive memorie³⁹, and one distress related to social exclusion⁴⁹. Four trials investigated the impact of OLPs on experimentally induced pain^37,41,83,90. Four trials focused on the effects of OLPs on physical and mental well-being^31,33,43 two trials on arousal and well-being^30,48, and one on effects on psychological distress²⁸. Seven trials investigated physiological processes: two trials focused on nausea⁵⁵, one on wound healing²⁹, one on experimentally induced itch⁸⁹, one on sleep quality²⁷, one on caffeine withdrawal symptoms⁹³, and one on physiological recovery⁵⁰. Additionally, two trials investigated the impact of OLPs on cognitive performance^35,81, and two on test anxiety^44,54. The impact of OLPs on promoting beneficial behaviors such as Progressive Muscle Relaxation⁹⁵ and Acts of Kindnes ⁹⁶ was investigated in two trials and one experiment explored effects on experimentally induced acute stress⁴⁵.

In clinical samples, five trials focused on allergic rhinitis^{38,40,46,56,84}, four trials on chronic low-back pain^26,32,80,85, and three on major depressive disorder^47,87,98. Additionally, three trials focused on cancer-related fatigue^36,91,92, two on irritable bowel syndrome^12,88, and one on opioid use disorder⁸⁶. One trial investigated the impact of OLPs on knee osteoarthritis pain⁴², one on menopausal hot flushes⁸², and one on the reduction of experimentally induced emotional distress in women diagnosed with major depressive disorder⁶⁴.

The type of OLP administration varied across the primary trials: 32 trials used OLP pills, 11 trials used active or passive nasal sprays (i.e., with or without a prickling sensation), four trials administered the OLP in the form of drops, three as a dermal cream, three as a decaffeinated coffee, and one each in the form of oral spray, oral vapor, saline injection, sham acupuncture, sham deep brain stimulation, verbal suggestion, and imaginary pills. Seven trials used CP as control condition, 39 trials NT, four trials TAU and eight trials WL. Regarding the OLP rationale, 46 trials used a treatment rationale with ‘high’ levels of suggestiveness, six trials used a treatment rationale without suggestive elements and three trials manipulated the treatment rationale, resulting in one intervention group with suggestion (OLP+) and one without (OLP-). In total, 91 self-report and 62 objective outcomes were extracted from non-clinical trials, and 34 self-report and two objective outcomes from clinical trials. The chance-corrected agreement between raters regarding the outcomes to be extracted was substantial (κ = 0.61)⁶³.

Risk of bias in studies

Supplemental digital appendix A4c presents the risk of bias assessment for the newly included trials in the updated review. In the updated review, eight trials^{26–28,85,86,91,93,95} (26%) were rated as having a high risk of bias. Seven of these trials had missing outcome data^{26,27,85,86,91,93,95}, while one trial²⁸ had concerns about the randomization process and one trial failed to analyze the data according to intention-to-treat⁹⁵. Twenty-three trials (74%) were rated as having a moderate risk of bias or ‘some concerns’. This was primarily due to the fact, that the majority of the trials collected self-report data, which could be influenced by participants’ the knowledge of their group allocation. In the overall sample, 11 primary trials (19%) were rated as having a ‘high’ risk of bias, 45 trials (78%) had ‘some concerns’ and two trials (3%) had a ‘low’ risk of bias.

Effects of OLPs

Table 1 summarizes the results of the main, sensitivity, and subgroup analyses. Figure 2 displays the forest plot for the main analysis of OLPs. We analyzed 61 comparisons across 58 trials, including three trials that manipulated treatment rationales in two distinct intervention groups (OLP + and OLP-)^31,41,56. The meta-analysis revealed a small but significant positive effect of OLP interventions compared to control conditions (k = 61, n = 4569, SMD = 0.36, 95% CI = 0.26; 0.45, p < 0.0001; I² = 56%), with moderate heterogeneity between trials (RQ1). The sensitivity analyses produced similar results, demonstrating the robustness of the results.

Table 1

*Effects of open-label placebos (OLPs).Main, sensitivity analyses, and subgroup analyses*
Analyses	k	n	g	CI_lb	CI_ub	I²	I²_lb	I²_ub	𝜏	PI_lb	PI_ub	p	Q	p_sg
Main and sensitivity analyses
All comparisons	61	4,569	0.36	0.26	0.45	55.66	40.88	66.74	0.27	-0.18	0.89	****
Outliers excluded	53	3,873	0.39	0.33	0.45	0.55	0.00	32.54	0.02	0.31	0.47	****
High risk excluded	50	3,701	0.33	0.23	0.44	56.65	40.55	68.39	0.27	-0.22	0.89	****
Adjusted for publication bias	69	5,027	0.27	0.18	0.37	64.84	54.65	72.74	0.33	-0.39	0.93	****
3-Level Model	177	4,569	0.33	0.24	0.43	56.60	-	-	0.33	-0.41	1.08	****
Subgroup analyses
Outcome
Self-reports	55	3,974	0.40	0.31	0.50	52.67	35.60	65.21	0.25	-0.12	0.92	****	9.81	**
Objective data	14	1,176	0.02	-0.19	0.24	69.87	47.88	82.59	0.34	-0.75	0.80
Population
Clinical	22	1,398	0.48	0.36	0.59	9.90	0.00	44.33	0.09	0.26	0.69	****	4.69	*
Non-clinical	39	3,171	0.29	0.17	0.42	63.41	48.46	74.02	0.30	-0.33	0.92	****
Suggestiveness
High	53	4,146	0.39	0.30	0.48	46.63	26.29	61.36	0.22	-0.06	0.84	****	2.29
Low	8	423	0.11	-0.25	0.46	70.31	38.40	85.69	0.42	-1.01	1.23
Type control
NT	42	3,067	0.30	0.18	0.42	62.66	47.96	73.20	0.31	-0.34	0.94	****	4.54
TAU	4	349	0.50	0.18	0.82	49.34	0.00	83.23	0.23	-0.70	1.70	**
CP	7	666	0.50	0.34	0.65	0.00	0.00	70.81	0.00	0.29	0.70	****
WL	8	487	0.43	0.25	0.61	0.00	0.00	67.58	0.00	0.20	0.66	****
Note. ^a The inferential statistics for the subgroup analysis used adjusted sample sizes to correct for unit-of-analysis errors due to double counting. These adjusted sample sizes differ from those shown in Table 1, which include participants who provided both self-reports and objective data.
Abbreviations. k = number of studies, CI = confidence interval, lb = lower bound of 95% CI, ub = upper bound of 95% CI, sg = subgroup, NT = no treatment, TAU = treatment as usual, CP = covert placebo, WL = waiting list.
p < 0.05; p < 0.01; p < 0.001; **p < 0.0001

OLP interventions significantly improved self-report outcomes (k = 55, n = 3974, SMD = 0.40, CI 95% = 0.31; 0.50, p < 0.0001; I² = 53%) but showed no significant effect on objective outcomes (k = 14, n = 1176, SMD = 0.02, CI 95% = -0.19; 0.24, p = 0.83; I² = 70%), with moderate to substantial heterogeneity. The difference in effectiveness between self-report and objective outcomes was significant (Q = 9.81, p < 0.01, RQ2).

OLP interventions had a small but significant effect in both clinical samples (k = 22, n = 1398, SMD = 0.48, CI 95% = 0.36; 0.59, p < 0.0001; I² = 10%) and non-clinical samples (k = 39, n = 3171, SMD = 0.29, CI 95% = 0.17; 0.42, p < 0.0001; I² = 63%), with lower heterogeneity in clinical samples. The effectiveness differed significantly between clinical and non-clinical samples (Q = 4.69, p < 0.05, RQ3).

Trials with a suggestive treatment rationale showed a small significant OLP effect (k = 53, n = 4146, SMD = 0.39, CI 95% = 0.30; 0.48, p < 0.0001; I² = 47%), whereas trials without suggestiveness did not (k = 8, n = 423, SMD = 0.11, CI 95% = -0.25; 0.46, p = 9.55; I² = 70%). However, the difference between these subgroups based on suggestiveness was not significant (Q = 2.29, p = 0.13, RQ4).

Finally, small to medium significant effects of OLPs were observed across all types of control conditions (Table 3). The difference between control subgroups was not significant (Q = 4.54, p = 0.21, RQ5).

Reporting bias

The visual inspection of the funnel plot (Fig. 3) and the Egger’s regression showed no evidence of publication bias for the main meta-analysis across outcomes and samples (t = 0.40, p = 0.69).

This systematic review and meta-analysis updates and synthesizes findings from four previous meta-analyses of randomized controlled trials (RCTs) on open-label placebos (OLPs)^7–10. With the inclusion of 21 additional trials, our review includes a total of 58 trials with 4,677 participants and is substantially larger than previous meta-analyses. This expansion allowed for novel statistical comparisons, offering deeper insights into OLP effects. For the first time, the review facilitated direct comparisons between self-report and objective outcomes and synthesized data from clinical and non-clinical samples. The review encompasses various OLP interventions, evaluates different outcomes, and compares these interventions across diverse control groups and sample types.

The results indicate that OLP interventions produce a small but clinically relevant positive effect across a wide range of conditions compared to control conditions (SMD = 0.36; RQ1). In addition, OLPs improved self-report outcomes but had no effect on objective outcomes, with a significant difference between the two (RQ2). OLPs had a small effect in both clinical and non-clinical samples, with a significant difference in effectiveness between these populations (RQ3). OLPs were effective when the treatment rationale was suggestive (i.e., inducing treatment expectations) but not when it was not, though the difference between these groups was not significant (RQ4). Lastly, OLPs exhibited small to medium significant effects against all kinds of control groups, with no significant differences between these control subgroups (RQ5). The overall heterogeneity was moderate in the primary analysis, but decreased substantially when excluding outliers, focusing solely on clinical samples, or comparing OLPs to covert placebos or waiting list controls.

To our knowledge, this systematic review and meta-analysis is the most comprehensive synopsis of OLP research to date. The overall positive effect of OLPs (RQ1) aligns with previous meta-analyses^7–10, confirming with extensive data that placebos are effective even when recipients are aware they are receiving a treatment, which is not pharmacologically active⁴. This finding suggests that OLPs can serve as a viable alternative to deceiving participants at least in certain clinical and research settings².

OLPs appear to affect self-report and objective outcomes differently (RQ2). While OLPs have a beneficial effect on self-report outcomes across both clinical and non-clinical populations, they show no effect on objective outcomes. Despite increasing the number of trials (from 17 to 55 comparisons for self-report outcomes and from eight to 14 comparisons for objective outcomes), the effect sizes remain consistent with those reported by Spille et al. (2023) (self-report SMDs: 0.43 vs. 0.40; objective SMDs: -0.02 vs. 0.02), highlighting consistency in the findings⁷. The results suggest that OLPs positively impacts individuals’ perceptions of their health but do not alter biological markers. This finding contrasts with meta-analytical findings on deceptive placebos in clinical trials, where patient-reported outcomes also show higher effect sizes (SMD = 0.26) than observer-rated outcomes (SMD = 0.13), but where the latter are significant⁹⁹. This difference could be interpreted in the sense that deceptive and non-deceptive placebo effects are based on different mechanisms. In the history of placebo research, deceptive placebos were initially thought to be merely suggestive to the patient, with no corresponding biological effect. This picture changed when the endogenous opoid system was identified as the biological mechanism of placebo analgesia^100,101. Today, there is a large body of literature that clearly demonstrates the biological mechanisms and effects of deceptive placebos¹⁰². For OLPs the situation is different. To date there exists only one experimental study demonstrating a biological mechanisms of OLP analgesia, but only in so-called OLP responders, who represent just 40% of the total sample¹⁰³. Our finding that OLPs, unlike deceptive placebos, have no effect on objective measures highlights the need to further investigate the psychological and biological mechanisms behind the reported OLP effects. Of particular interest are the variables suggestion and expectation, also with regard to interpersonal differences. Another fruitful focus could be the perspective to understand OLPs as a healing ritual and in particular addressing its embodied aspects^104–106.

Although OLP interventions benefit both clinical and non-clinical samples, they affect these groups differently (RQ3). Our meta-analysis shows a larger effect in clinical samples (SMD = 0.48) compared to non-clinical samples (SMD = 0.29). This difference likely arises from the greater need for relief in clinical samples due to disease-related impairments, which provides a larger potential for improvement^107,108. For example, Barnes et al. (2023)³¹ found in a non-clinical sample larger OLP effects on wellbeing only for participants whose baseline scores were below the average, suggesting that higher distress levels might enhance OLP effects. Additionally, clinical trials lasted longer than to non-clinical trials, with a mean duration of 26.7 days (SD = 21.4) compard to 5.2 days (SD = 6.3) for non-clinical samples. This extended duration may allow patients to develop a stronger connection to the healing ritual aspects of OLPs compared to non-clinical samples⁶. It may also leave more time for the intervention to unfold. The results for clinical samples align with findings of Buergler et al. (2023)¹⁰, who reported an almost identical effect for OLP pills compared to NT (SMD = 0.46). This effect is somewhat smaller than those suggested by earlier meta-analyses on OLPs in clinical samples (SMD = 0.88⁸ and SMD = 0.72⁹). However, the earlier meta-analyses were based on a relatively small body of evidence with small sample sizes. The reduction in magnitude of the effect sizes may indicate a time-lag bias, where significant results were more likely to be reported early in OLP research, while later studies with insignificant results were likely to be published over time¹⁰. Similar earlier studies may have lower methodological quality. A sensitivity analysis excluding trials with a high risk of bias in Wernsdorff (2021) revealed a reduced effects size of SMD = 0.49 being very close to our finding⁹. The beneficial effects of OLP interventions in non-clinical samples (SMD = 0.29) are smaller than in clinical samples. They are also smaller that in the prior meta-analysis by Spille et al. (2023)⁷ as well as the significant findings for nasal spray in non-clinical samples for from Buergler et al. (2023)¹⁰. This may be due to the combination of studies with self-reported and objective outcome which were separated in Spille et al. (2023). With respect to Buergler et al. (2023) also methodological differences may account for these variations. For example, they excluded trials with balanced-placebo designs as well as those with CP as a control condition¹⁰, while our analysis included such trials (see eligibility criteria). In addition, they included non-randomized trials^72,73, which we excluded. Moreover, Buergler et al. (2023) used a different classification for clinical and non-clinical samples¹⁰, based on the investigated states/conditions rather than on the health status of the participants/patients. Therefore they classified some trials as clinical, which were considered to be non-clinical by our approach^33,44,81,95. Lastly, Buergler et al. (2023) included studies with deceptive placebo control group¹⁰, which were excluded in our analysis. These methodological differences resulted in fewer trials in the non-clinical network compared to our analysis (12 trials in Buergler et al., 2023¹⁰ vs. 37 in the present analysis).

The degree of suggestiveness in the treatment rationale accompanying the administration of OLPs in our review had a major impact on their effectiveness (RQ4). Trials with a suggestive rationale that aimed to build treatment expectations showed a small positive effect (SMD = 0.39), while those without such a rationale did not demonstrate significant effects (SMD = 0.11). These results align with Buergler at al. (2023), emphasizing that building expectancy is crucial for OLP effectiveness and that merely administering an inert treatment is not sufficient¹⁰. This supports the prevailing view that OLP effects are based on the elicitation of treatment expectations through a suggestive rationale^6,109.

Regarding the impact of different types of control groups on OLP effectiveness (RQ5), we compared OLP interventions against four types of comparators: seven used a covert placebo (CP), 42 used no treatment controls (NT), four used treatment as usual (TAU), and eight used weight list (WL) controls. The inferential analysis revealed no significant differences among these control groups, indicating that OLPs yield beneficial effects across all control types. However, as “absence of evidence is not evidence of absence”¹¹⁰, these findings should be interpreted carefully. Moreover, this findings somewhat contrast with Buergler et al. (2023), who concluded that OLPs performed better than ‘nothing’ but only slightly better than ‘something’ and that comparator groups should be chosen carefully¹⁰. As for population types, methodological differences between our study and Buergler et al. (2023)¹⁰ may account for these discrepancies in results (see above).

The presented systematic review and meta-analysis has several strengths. First, it adhered to rigorous methodology, including a pre-registered search protocol and data synthesis, with two independent researchers assessing eligibility, extracting data, and evaluating risk of bias. Second, it incorportated 21 additional RCTs compared to the most recent previous meta-analysis of OLPs¹⁰, thereby strengthening the robustness of the results. Third, we directly compared differences in the effectiveness of OLPs between self-report and objective outcomes as well as between clinical and non-clinical samples and these differences yielded valuable insights. Finally, we conducted multiple sensitivity analyses, which confirmed the robustness of the main findings and reinforced observed trends in OLP effects.

Several limitations should be considered when interpreting the results. First, most of the trials included in the review had relatively small sample sizes (i.e., fewer than 100 participants). As already denoted by Buergler et al. (2023), this limitation raised the possibility of a so-called ‘small study effect’¹⁰, where smaller trials often report larger treatment effects than larger trials^62,111,112. Second, we combined effect sizes across different samples (e.g., across various clinical conditions) and outcomes (e.g., self-report vs. objective), which could bias the observed OLP effect, as various conditions and outcomes may respond differently to OLP administration. In a similar vein, we independently aggregated multiple treatment arms with varying degree of suggestiveness in the treatment rationale (i.e. OLP + and OLP-) and analyzed these arms separately in the analysis. This approach may potentially skew effect sizes due to over-representation and inter-correlation of outcomes and treatment arm in question. However, we tried to avoid over-representation in terms of unit-of-analysis error due to double counting by splitting the sample sizes of the respective groups. Moreover, we tried to model the potential dependency of multiple effect size estimates for one sample by conducting a sensitivity analysis using a hierarchical three-level meta-analytic model, with effect sizes nested in studies, leading to comparable results for the main analysis. Third, we did not peform a GRADE rating. This is because, we consider the GRADE system to be suboptimal for the use in our OLP review. We see especially the inability to blind patients/participants to the intervention, affecting the risk of bias rating, and the inclusion of a huge varieties of different outcomes in our review as problematic with respect to GRADE.

Our systematic review has identified several research gaps. First, trials with larger sample sizes are essential to enhance the robustness of the findings. Second, longer study durations are required, as most trials in non-clinical populations lasted only one day, and as long-term follow-up trials on OLP interventions in clinical samples have yielded mixed results^70,71. This raises questions about the durability of OLP effets and underscores the need for further research on long-term effects in both clinical and non-clinical settings. Second, to increase the validity of the results, future trials should include more representative samples. Non-clinical trials in particular, often included younger and predominantly female participants compared to clinical populations. Third, as OLPs affect self-report and objective outcomes differently, future research should put a stronger focus on objective outcomes, particularly in clinical settings. Finally, this systematic review and meta-analysis included several trials published by the same research groups, indicating the need for research teams with less allegiance to OLPs to mitigate potential bias.

In summary, the results of the presented systematic review and meta-analysis indicate several key findings: (1) OLP interventions demonstrate small but significant positive overall benefit across various conditions, samples, and outcomes in a larger sample of studies. (2) OLPs are effective only on self-report measures but not for objective measures, and this finding points to important research questions about their mechanisms, as well as how they differ from deceptive placebos. (3) OLPs are more effective in clinical than in non-clinical samples. (4) The suggestiveness of the treatment rationale in OLP interventions is crucial, as trials lacking suggestive elements did not yield beneficial effects. (5) Different comparator groups produced similar estimates of OLP effectiveness.

Acknowledgements

We thank all the authors of the studies included in this systematic review and meta-analysis for their valuable contributions.

Author contributions

JF contributed to the conceptualization of the study and shared project administration responsibilities with SS. He supervised the study's execution, developed methodological considerations, and conducted the final formal analysis. Additionally, he wrote the original draft based on CT's master's thesis.

CT co-conceptualized the study, was responsible for the literature search, screening and data extraction, curated the data, and contributed to the methodological development. He provided the foundation for the original draft with his master's thesis, performed the initial data analysis, played a key role in the formal analysis, and was responsible for reviewing and editing the manuscript.

PS served as a co-reviewer and, together with CT, conducted the screening process of the studies and assisted in data curation. He contributed to the methodology and edited the final version of the paper.

JG supervised the project throughout its various phases and reviewed and edited the manuscript.

SS conceived the idea for the study and shared project administration with JF. He was responsible for the study's conceptualization, contributed to the methodology, supervised the various phases, and reviewed and edited the final version of the draft.

Data availability statement

All data generated or analysed during this study are either included in this published article and its supplementary information file or have been deposited at https://osf.io/pxzcg/

Competing interests statement

The authors declare they have no competing interests.

Benedetti, F. Mechanisms of placebo and placebo-related effects across diseases and treatments. Annu. Rev. Pharmacol. Toxicol. 48, 33–60 (2008).
Evers, A. W. M. et al. Implications of placebo and nocebo effects for clinical practice: Expert consensus. Psychother Psychosom 87, 204–210 (2018).
Finniss, D. G., Kaptchuk, T. J., Miller, F. & Benedetti, F. Biological, clinical, and ethical advances of placebo effects. The Lancet 375, 686–695 (2010).
Evers, A. W. M. et al. What should clinicians tell patients about placebo and nocebo effects? Practical considerations based on expert consensus. Psychother Psychosom 90, 49–56 (2021).
Sandler, A. D. & Bodfish, J. W. Open-label use of placebos in the treatment of ADHD: A pilot study. Child Care Health Dev 34, 104–110 (2008).
Kaptchuk, T. J. Open-label placebo: Reflections on a research agenda. Perspectives in Biology and Medicine 61, 311–334 (2018).
Spille, L., Fendel, J. C., Seuling, P. D., Göritz, A. S. & Schmidt, S. Open-label placebos—a systematic review and meta-analysis of experimental studies with non-clinical samples. Sci Rep 13, 3640 (2023).
Charlesworth, J. E. G. et al. Effects of placebos without deception compared with no treatment: A systematic review and meta-analysis. J Evid Based Med 10, 97–107 (2017).
von Wernsdorff, M., Loef, M., Tuschen-Caffier, B. & Schmidt, S. Effects of open-label placebos in clinical trials: A systematic review and meta-analysis. Sci Rep 11, 3855 (2021).
Buergler, S., Sezer, D., Gaab, J. & Locher, C. The roles of expectation, comparator, administration route, and population in open-label placebo effects: A network meta-analysis. Sci Rep 13, 11827 (2023).
Colloca, L. & Miller, F. G. How placebo responses are formed: A learning perspective. Phil. Trans. R. Soc. B 366, 1859–1869 (2011).
Kaptchuk, T. J. et al. Placebos without deception: A randomized controlled trial in irritable bowel syndrome. PLOS ONE 5, e15591 (2010).
Michopoulos, I. et al. Different control conditions can produce different effect estimates in psychotherapy trials for depression. Journal of Clinical Epidemiology 132, 59–70 (2021).
Page, M. J. et al. The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. International Journal of Surgery 88, 105906 (2021).
Blease, C., Colloca, L. & Kaptchuk, T. J. Are open-label placebos ethical? Informed consent and ethical equivocations. Bioethics 30, 407–414 (2016).
Higgins, J. P. T. et al. Cochrane Handbook for Systematic Reviews of Interventions. (John Wiley & Sons, 2019).
Ouzzani, M., Hammady, H., Fedorowicz, Z. & Elmagarmid, A. Rayyan—a web and mobile app for systematic reviews. Syst Rev 5, 210 (2016).
López‐López, J. A., Page, M. J., Lipsey, M. W. & Higgins, J. P. T. Dealing with effect size multiplicity in systematic reviews and meta‐analyses. Research Synthesis Methods 9, 336–351 (2018).
Rothman, K. J., Greenland, S. & Lash, T. L. Modern Epidemiology. (Wolters Kluwer health - Lippincott Williams & Wilkins, Philadelphia, 2008).
Mohr, D. C. et al. The selection and design of control conditions for randomized controlled trials of psychological interventions. Psychotherapy and Psychosomatics 78, 275–284 (2009).
Sterne, J. A. C. et al. RoB 2: A revised tool for assessing risk of bias in randomised trials. BMJ 366, l4898 (2019).
R Core Team. R: A language and environment for statistical computing. https://www.R-project.org/ (2021).
Morris, S. B. Estimating effect sizes from pretest-posttest-control group designs. Organizational Research Methods 11, 364–386 (2008).
Borenstein, M. Effect Sizes for Continuous Data. in The handbook of research synthesis and meta-analysis, 2nd ed 221–235 (Russell Sage Foundation, New York, NY, US, 2009).
Cohen, J. Statistical power analysis. Curr Dir Psychol Sci 1, 98–101 (1992).
Ikemoto, T. et al. Open-Label Placebo Trial among Japanese Patients with Chronic Low Back Pain. Pain Research and Management 2020, e6636979 (2020).
Potthoff, J. & Schienle, A. Effects of (non)deceptive placebos on reported sleep quality and food cue reactivity. J. Sleep Res. (2023) doi:10.1111/jsr.13947.
Winkler, A., Hahn, A. & Hermann, C. The impact of pharmaceutical form and simulated side effects in an open-label-placebo rct for improving psychological distress in highly stressed students. Sci Rep 13, 6367 (2023).
Mathur, A., Jarrett, P., Broadbent, E. & Petrie, K. J. Open-label placebos for wound healing: A randomized controlled trial. Annals of Behavioral Medicine 52, 902–908 (2018).
Walach, H., Schmidt, S., Bihr, Y.-M. & Wiesch, S. The effects of a caffeine placebo and experimenter expectation on blood pressure, heart rate, well-being, and cognitive performance. European Psychologist 6, 15–25 (2001).
Barnes, K., Babbage, E., Barker, J., Jain, N. & Faasse, K. The role of positive information provision in open-label placebo effects. Appl Psychol Health Well Being (2023) doi:10.1111/aphw.12444.
Carvalho, C. et al. Open-label placebo treatment in chronic low back pain: A randomized controlled trial. Pain 157, 2766–2772 (2016).
El Brihi, J., Horne, R. & Faasse, K. Prescribing placebos: An experimental examination of the role of dose, expectancies, and adherence in open-label placebo effects. Ann Behav Med 53, 16–28 (2019).
Guevarra, D. A., Moser, J. S., Wager, T. D. & Kross, E. Placebos without deception reduce self-report and neural measures of emotional distress. Nat Commun 11, 3785 (2020).
Heller, M. K., Chapman, S. C. E. & Horne, R. Beliefs about medicines predict side-effects of placebo modafinil. Ann Behav Med 56, 989–1001 (2022).
Hoenemeyer, T. W., Kaptchuk, T. J., Mehta, T. S. & Fontaine, K. R. Open-label placebo treatment for cancer-related fatigue: A randomized-controlled clinical trial. Sci Rep 8, 2784 (2018).
Kube, T. et al. Deceptive and nondeceptive placebos to reduce pain: An experimental study in healthy individuals. The Clinical Journal of Pain 36, 68–79 (2020).
Kube, T., Hofmann, V. E., Glombiewski, J. A. & Kirsch, I. Providing open-label placebos remotely - a randomized controlled trial in allergic rhinitis. PLoS ONE 16, (2021).
Kube, T., Kirsch, I., Glombiewski, J. A. & Herzog, P. Can placebos reduce intrusive memories? Behav. Res. Ther. 158, (2022).
Kube, T., Kirsch, I., Glombiewski, J. A., Witthöft, M. & Bräscher, A.-K. Remotely provided open-label placebo reduces frequency of and impairment by allergic symptoms. Psychosom. Med. 84, 997–1005 (2022).
Locher, C. et al. Is the rationale more important than deception? A randomized controlled trial of open-label placebo analgesia. Pain 158, 2320–2328 (2017).
Olliges, E. et al. Open-label placebo administration decreases pain in elderly patients with symptomatic knee osteoarthritis – a randomized controlled trial. Front. Psychiatry 13, (2022).
Rathschlag, M. & Klatt, S. Open-label placebo interventions with drinking water and their influence on perceived physical and mental well-being. Front Psychol 12, 658275 (2021).
Schaefer, M. et al. Open-label placebos reduce test anxiety and improve self-management skills: A randomized-controlled trial. Sci Rep 9, 13317 (2019).
Schaefer, M., Hellmann-Regen, J. & Enge, S. Effects of open-label placebos on state anxiety and glucocorticoid stress responses. Brain Sciences 11, 508 (2021).
Schaefer, M., Zimmermann, K. & Enck, P. A randomized controlled trial of effects of open-label placebo compared to double-blind placebo and treatment-as-usual on symptoms of allergic rhinitis. Sci Rep 13, 8372 (2023).
Schienle, A. & Jurinec, N. Open-label placebos as adjunctive therapy for patients with depression. Contemp. Clin. Trials Comm. 28, (2022).
Schneider, R. et al. Effects of expectation and caffeine on arousal, well-being, and reaction time. Int. J. Behav. Med. 13, 330–339 (2006).
Stumpp, L., Jauch, M., Sezer, D., Gaab, J. & Greifeneder, R. Effects of an open-label placebo intervention on reactions to social exclusion in healthy adults: A randomized controlled trial. Sci Rep 13, 15369 (2023).
Urroz, P., Colagiuri, B., Smith, C. A., Yeung, A. & Cheema, B. S. Effect of acupuncture and instruction on physiological recovery from maximal exercise: A balanced-placebo controlled trial. BMC Complementary and Alternative Medicine 16, 227 (2016).
Marín-Martínez, F. & Sánchez-Meca, J. Averaging dependent effect sizes in meta-analysis: A cautionary note about procedures. The Spanish Journal of Psychology 2, 32–38 (1999).
Viechtbauer, W. Conducting meta-analyses in R with the metafor package. Journal of Statistical Software 36, 1–48 (2010).
Harrer, M., Cuijpers, P., Furukawa, T. & Ebert, D. Doing Meta-Analysis with R: A Hands-on Guide. (Chapman and Hall/CRC, New York, 2021). doi:10.1201/9781003107347.
Buergler, S. et al. Imaginary pills and open-label placebos can reduce test anxiety by means of placebo mechanisms. Sci Rep 13, 2624 (2023).
Barnes, K., Yu, A., Josupeit, J. & Colagiuri, B. Deceptive but not open label placebos attenuate motion-induced nausea. Journal of Psychosomatic Research 125, 109808 (2019).
Schaefer, M., Sahin, T. & Berstecher, B. Why do open-label placebos work? A randomized controlled trial of an open-label placebo induction with and without extended information about the placebo effect in allergic rhinitis. PLoS One 13, e0192758 (2018).
Schienle, A., Unger, I. & Schwab, D. Changes in neural processing and evaluation of negative facial expressions after administration of an open-label placebo. Sci Rep 12, 6577 (2022).
Balduzzi, S., Rücker, G. & Schwarzer, G. How to perform a meta-analysis with R: A practical tutorial. BMJ Ment Health 22, 153–160 (2019).
Marín-Martínez, F. & Sánchez-Meca, J. Weighting by inverse variance or by sample size in random-effects meta-analysis. Educational and Psychological Measurement 70, 56–73 (2010).
Borenstein, M., Higgins, J. P. T., Hedges, L. V. & Rothstein, H. R. Basics of meta-analysis: I2 is not an absolute measure of heterogeneity. Research Synthesis Methods 8, 5–18 (2017).
Duval, S. & Tweedie, R. Trim and Fill: A Simple Funnel-Plot–Based Method of Testing and Adjusting for Publication Bias in Meta-Analysis. Biometrics 56, 455–463 (2000).
Page, M. J., Sterne, J. A. C., Higgins, J. P. T. & Egger, M. Investigating and dealing with publication bias and other reporting biases in meta-analyses of health research: A review. Research Synthesis Methods 12, 248–259 (2021).
Cohen, J. A coefficient of agreement for nominal scales. Educational and Psychological Measurement 20, 37–46 (1960).
Haas, J. W., Rief, W., Glombiewski, J. A., Winkler, A. & Doering, B. K. Expectation-induced placebo effect on acute sadness in women with major depression: An experimental investigation. Journal of Affective Disorders 274, 920–928 (2020).
Schaefer, M., Kühnel, A., Schweitzer, F., Enge, S. & Gärtner, M. Neural underpinnings of open-label placebo effects in emotional distress. Neuropsychopharmacology 48, 560–566 (2023).
Flowers, K. M. et al. Conditioned open-label placebo for opioid reduction after spine surgery: A randomized controlled trial. Pain 162, 1828–1839 (2021).
Nurko, S. et al. Effect of open-label placebo on children and adolescents with functional abdominal pain or irritable bowel syndrome: A randomized clinical trial. JAMA Pediatr. 176, 349–356 (2022).
Braescher, A.-K., Ferti, I.-E. & Witthöft, M. Open-label placebo effects on psychological and physical well-being: A conceptual replication study. (2022) doi:10.32872/cpe.7679.
Lee, S., Choi, D.-H., Hong, M., Lee, I.-S. & Chae, Y. Open-label placebo treatment for experimental pain: A randomized-controlled trial with placebo acupuncture and placebo pills. J. Integr. Complement. Med. 28, 136–145 (2022).
Carvalho, C. et al. Open-label placebo for chronic low back pain: A 5-year follow-up. Pain 162, 1521–1527 (2021).
Kleine-Borgmann, J., Dietz, T.-N., Schmidt, K. & Bingel, U. No long-term effects after a 3-week open-label placebo treatment for chronic low back pain: A 3-year follow-up of a randomized controlled trial. Pain 164, 645–652 (2023).
Disley, N., Kola-Palmer, S. & Retzler, C. A comparison of open-label and deceptive placebo analgesia in a healthy sample. JOURNAL OF PSYCHOSOMATIC RESEARCH 140, (2021).
Haas, J. W., Winkler, A., Rheker, J., Doering, B. K. & Rief, W. No open-label placebo effect in insomnia? Lessons learned from an experimental trial. J. Psychosom. Res. 158, (2022).
Morales-Quezada, L. et al. Conditioning open-label placebo: a pilot pharmacobehavioral approach for opioid dose reduction and pain control. Pain Rep 5, e828 (2020).
De Vita, M. J. et al. The effects of cannabidiol and analgesic expectancies on experimental pain reactivity in healthy adults: A balanced placebo design trial. Exp Clin Psychopharmacol 30, 536–546 (2022).
Sandler, A. D. & Bodfish, J. W. Open-label use of placebos in the treatment of ADHD: A pilot study. Child Care Health Dev 34, 104–110 (2008).
Friehs, T., Rief, W., Glombiewski, J. A., Haas, J. & Kube, T. Deceptive and non-deceptive placebos to reduce sadness: A five-armed experimental study. Journal of Affective Disorders Reports 9, 100349 (2022).
Glombiewski, J. A., Rheker, J., Wittkowski, J., Rebstock, L. & Rief, W. Placebo mechanisms in depression: An experimental investigation of the impact of expectations on sadness in female participants. Journal of Affective Disorders 256, 658–667 (2019).
Hahn, A., Göhler, A. C., Hermann, C. & Winkler, A. Even when you know it is a placebo, you experience less sadness: First evidence from an experimental open-label placebo investigation. J. Affective Disord. 304, 159–166 (2022).
Kleine-Borgmann, J., Schmidt, K., Hellmann, A. & Bingel, U. Effects of open-label placebo on pain, functional disability, and spine mobility in patients with chronic back pain: A randomized controlled trial. PAIN 160, 2891 (2019).
Kleine-Borgmann, J. et al. Effects of open-label placebos on test performance and psychological well-being in healthy medical students: A randomized controlled trial. Sci Rep 11, 2130 (2021).
Pan, Y. et al. Open-label placebos for menopausal hot flushes: A randomized controlled trial. Sci Rep 10, 20090 (2020).
Rief, W. & Glombiewski, J. A. The hidden effects of blinded, placebo-controlled randomized trials: An experimental investigation. PAIN® 153, 2473–2477 (2012).
Schaefer, M., Harke, R. & Denke, C. Open-label placebos improve symptoms in allergic rhinitis: A randomized controlled trial. Psychotherapy and Psychosomatics 85, 373–374 (2016).
Ashar, Y. K. et al. Effect of pain reprocessing therapy vs placebo and usual care for patients with chronic back pain: A randomized clinical trial. JAMA Psychiatry 79, 13–23 (2022).
Belcher, A. M. et al. Effectiveness of conditioned open-label placebo with methadone in treatment of opioid use disorder: A randomized clinical trial. JAMA Netw. Open 6, E237099 (2023).
Kelley, J. M., Kaptchuk, T. J., Cusin, C., Lipkin, S. & Fava, M. Open-label placebo for major depressive disorder: A pilot randomized controlled trial. Psychotherapy and Psychosomatics 81, 312–314 (2012).
Lembo, A. et al. Open-label placebo vs double-blind placebo for irritable bowel syndrome: A randomized clinical trial. Pain 162, 2428–2435 (2021).
Meeuwis, S. et al. Placebo effects of open-label verbal suggestions on itch. Acta Derm Venerol 98, 268–274 (2018).
Mundt, J. M., Roditi, D. & Robinson, M. E. A comparison of deceptive and non-deceptive placebo analgesia: Efficacy and ethical consequences. Annals of Behavioral Medicine 51, 307–315 (2017).
Yennurajalingam, S. et al. Open-label placebo for the treatment of cancer-related fatigue in patients with advanced cancer: A randomized controlled trial. Oncologist 27, 1081–1089 (2022).
Zhou, E. S. et al. Open-label placebo reduces fatigue in cancer survivors: A randomized trial. Support Care Cancer 27, 2179–2187 (2019).
Mills, L., Lee, J. C., Boakes, R. & Colagiuri, B. Reduction in caffeine withdrawal after open-label decaffeinated coffee. J. Psychopharmacol. 37, 181–191 (2023).
Schienle, A., Kogler, W. & Wabnegger, A. A randomized trial that compared brain activity, efficacy and plausibility of open-label placebo treatment and cognitive repappraisal for reducing emotional distress. Sci Rep 13, 13998 (2023).
Schienle, A. & Unger, I. Open-label placebo treatment to improve relaxation training effects in healthy psychology students: A randomized controlled trial. Sci Rep 11, 13073 (2021).
Schienle, A. & Unger, I. Non-deceptive placebos can promote Acts of Kindness: A randomized controlled trial. 13, (2023).
Sezer, D., Locher, C. & Gaab, J. Deceptive and open-label placebo effects in experimentally induced guilt: A randomized controlled trial in healthy subjects. Sci Rep 12, 21219 (2022).
Nitzan, U. et al. Open-label placebo for the treatment of unipolar depression: Results from a randomized controlled trial. Journal of Affective Disorders 276, 707–710 (2020).
Hróbjartsson, A. & Gøtzsche, P. C. Placebo interventions for all clinical conditions. Cochrane Database of Systematic Reviews (2010) doi:10.1002/14651858.CD003974.pub3.
Levine, J. D., Gordon, N. C., Jones, R. T. & Fields, H. L. The narcotic antagonist naloxone enhances clinical pain. Nature 272, 826–827 (1978).
Benedetti, F. The opposite effects of the opiate antagonist naloxone and the cholecystokinin antagonist proglumide on placebo analgesia. Pain 64, 535–543 (1996).
Benedetti, F., Carlino, E. & Pollo, A. How placebos change the patient’s brain. Neuropsychopharmacology 36, 339–354 (2011).
Benedetti, F., Shaibani, A., Arduino, C. & Thoen, W. Open-label nondeceptive placebo analgesia is blocked by the opioid antagonist naloxone. Pain 164, 984–990 (2023).
Wilhelm, M. et al. Working with patients’ treatment expectations – what we can learn from homeopathy. Front. Psychol. 15, (2024).
Kaptchuk, T. J. The placebo effect in alternative medicine: Can the performance of a healing ritual have clinical significance? Ann Intern Med 136, 817–825 (2002).
Schmidt, S. & Walach, H. Making Sense in the Medical System: Placebo, Biosemiotics, and the Pseudomachine. in vol. 5 195–215 (2016).
Hyland, M. E. Motivation and placebos: Do different mechanisms occur in different contexts? Philosophical Transactions of the Royal Society B: Biological Sciences 366, 1828–1837 (2011).
Kaptchuk, T. J. & Miller, F. G. Open label placebo: Can honestly prescribed placebos evoke meaningful therapeutic benefits? BMJ 363, k3889 (2018).
Blease, C. R., Bernstein, M. H. & Locher, C. Open-label placebo clinical trials: Is it the rationale, the interaction or the pill? BMJ Evid Based Med 25, 159–165 (2020).
Feres, M. & Feres, M. F. N. Absence of evidence is not evidence of absence. J. Appl. Oral Sci. 31, ed001 (2023).
Sterling, T. D., Rosenbaum, W. L. & Weinkam, J. J. Publication decisions revisited: The effect of the outcome of statistical tests on the decision to publish and vice versa. The American Statistician 49, 108–112 (1995).
Sterne, J. A. C., Gavaghan, D. & Egger, M. Publication and related bias in meta-analysis: Power of statistical tests and prevalence in the literature. Journal of Clinical Epidemiology 53, 1119–1129 (2000).

No competing interests reported.

OLPsupplementaldigitalappendix004.docx

Download PDF

Reviewers invited by journal
16 Oct, 2024
Editor assigned by journal
16 Oct, 2024
Editor invited by journal
09 Oct, 2024
Submission checks completed at journal
08 Oct, 2024
First submitted to journal
07 Oct, 2024

You are reading this latest preprint version

Effects of open-label placebos across outcomes and populations: An updated systematic review and meta-analysis of randomized controlled trials

Status:

Version 1

Abstract

Figures

Introduction

Methods

Eligibility criteria

Information sources and search strategy

Selection process

Data items and collection process

Risk of bias assessment

Data synthesis and analysis

Sensitivity analysis

Reporting bias assessment and certainty assessment

Results

Trial characteristics

Risk of bias in studies

Effects of OLPs

Reporting bias

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1