A Dermatological View—Interpreting Placebo Response in Clinical Trials for Psoriasis

Feb 1, 2013 | Contact Author | By: Howard I. Maibach, MD, University of California San Francisco; and Sonia Lamel, MD, University of Miami
Your message has been sent.
(click to close)
Contact the Author
This item has been saved to your library.
View My Library
(click to close)
Save to My Library
Title: A Dermatological View—Interpreting Placebo Response in Clinical Trials for Psoriasis
psoriasisx placebosx response ratesx random effectsx
  • Article
  • Media
  • Keywords/Abstract
  • Related Material

Keywords: psoriasis | placebos | response rates | random effects

Abstract: By comparing response rates of placebo versus active drug groups in psoriasis RCTs evaluating biologic agents, the authors of this column sought to clarify factors contributing to placebo responses and their implications in improving clinical trial design to determine more accurate drug efficacies.

View citation for this article

HI Maibach and S Lamel, A dermatological view—Interpreting placebo response in clinical trials for psoriasis, Cosm & Toil 128(2) 84-87 (Feb 2013)

* This column was adapted from Lamel et al., Placebo Response in Relation to Clinical Trial Design: Systematic review and meta-analysis of randomized controlled trials for determining biologic efficacy in psoriasis treatment, Arch Dermatol Res Nov 304(9) 707–717(2012).

In clinical medicine, the randomized, placebo-controlled trial (RCT) is the main accepted method to demonstrate the efficacy of investigational topical skin care drugs, and is generally required in order to obtain U.S. Food and Drug Administration (FDA) approval. A similar but generally less rigorous methodology is adopted in the cosmetic industry for skin care trials.1–3 Volunteers in placebo treatment groups within RCTs often report considerable improvement;4 therefore, interpreting the efficacy results for those given the active is complicated by similar results from those given the placebo.

In relation, psoriasis is a long-lasting, immunologically mediated, often severe inflammatory skin disease affecting approximately 3% of the Caucasian population.5 A class of drugs known as the “biologics” is efficacious in treating psoriasis; however, even well-designed psoriasis RCTs can have significant placebo response rates. When response values are present for both the placebo group and the active group, the additive model predicts efficacy by subtracting the response of the placebo group from the active group.6 However, the observed placebo response may reflect factors not implicit in the general definition of placebo response. These indeterminate factors warrant further investigation to optimize RCT design.

The placebo effect has harnessed attention due to widespread belief of its potential health impact. This credence was challenged when reviews of RCTs comparing placebo to no treatment groups showed placebo interventions had no clinically important effects, also suggesting no treatment groups may be necessary to reveal true drug efficacy.7 Additionally, an examination of psoriasis RCTs showed that placebos are more commonly used as comparators in the United States than in other countries and in trials sponsored by pharmaceutical companies.8 This discloses the need to unravel the components impacting placebo responses within RCTs.

By comparing response rates of placebo versus active drug groups in psoriasis RCTs evaluating biologic agents, the authors of this column sought to clarify factors contributing to placebo responses and their implications in improving clinical trial design to determine more accurate drug efficacies.


A systematic review of RCTs evaluating the efficacy of biologics versus placebo was completed. Data was abstracted for a percentage of responders in both active treatment and placebo groups as determined by 75% improvement in psoriasis area severity index (PASI 75) after an initial blinded, placebo-controlled treatment phase. Treatment indication, length of the placebo period, study duration, relevant inclusion criteria, randomization strategy, and the specific medication with route, dosage and schedule of administration information were also recorded.

Meta-analysis was performed by pooling all active treatment and placebo group patients achieving PASI 75. Percentages of responders and averages were tabulated based on the number of patients in each group. Odds ratios comparing active to placebo were calculated overall and for each study. Relationships between study size and effect size were investigated with an augmented funnel plot created by plotting each study by effect size and standard error.

To determine the impact of study characteristics on placebo response, univariate and combined logistic regression models were created using study phase, whether PASI or stability of psoriasis were inclusion criteria, treatment indications, time of outcomes measures documentation, and whether more than 51% of patients were randomized to active drug treatment. Significant interactions were explored using binomial classification and regression tree (CART) models.


The results were culminated from 31 randomized, controlled trials involving 8,285 active treatments and 3,999 placebo treatments where PASI was an outcome measure, and their data were abstracted for analysis. The percentages of placebo responders reaching PASI 75 varied from 0% to 18% with a pooled average of 4.1%. The percentage of active drug responders achieving PASI 75 ranged from 23% to 81% with a pooled average value of 48.4%. The overall odds ratio calculated was 23.94 (p < 0.0001, 95% Confidence Interval (CI) 16.02–35.76), with individual ratios for each study listed in Table 1. The percentage of total variability due to heterogeneity between the studies is 79.5%.

Evaluation of covariates revealed treatment indication to be significant, with psoriatic arthritis studies exhibiting lower placebo responses than psoriasis studies (pooled placebo response of 2.2% vs. 4.5%). The requirement of a higher PASI for study inclusion became more significant in the multivariate model, while longer study duration and more than half of participants randomized to active drug became less so, suggesting possible interactions of these variables. Table 2 shows the contribution of covariates to the rates of placebo responders.

Evaluation via CART model showed studies randomizing less than 51% of participants to active drug exhibited the smallest pooled placebo response (7 studies, pooled response of 2.0%), followed by those randomizing 51% or more and those requiring higher PASI scores for study inclusion (17 studies, pooled placebo response of 3.7%), and studies randomizing greater than 51% to active drug and not requiring a higher PASI (7 studies, pooled placebo response of 7.1%).


Psoriasis RCTs evaluating biologic efficacy have significantly low levels of placebo responders, especially in trials requiring stability of psoriasis for eligibility, longer study duration before outcomes are measured and higher values of PASI for inclusion. The overall percentage of placebo responders of 4.14% is significantly lower than typical placebo response rates in efficacy trials for other psoriasis treatments.9 Acne treatment trials exhibit differences between active drug and placebo responders as small as 14%.10 The authors suspect this indicates observed placebo responses may be secondary to factors inherent in the psoriasis disease course and RCT design and implementation. The observed 0–18% variability in placebo responders also implies that elucidating the responsible factors could lead to optimization of RCT design.

Mechanisms to explain placebo responses have been clarified for different disease processes. A physiologic explanation was first suggested in a post-operative pain trial where placebo responders’ analgesia was antagonized by naloxone.11 Neurobiological mechanisms were again suggested when placebo-treated Parkinson’s disease patients were found to release endogenous dopamine.12 Additionally, analgesia studies in Alzheimer’s patients reveal blunted placebo responses in the cognitively impaired.13 However no physiologic explanation for placebo response in psoriasis exists.

Factors inherent in RCTs also contribute to placebo response. Conditioning is a phenomenon well-known for evoking placebo response.14 Cues associated with medication characteristics, administration frequency and route can facilitate response via conditioning. The expectancy theory, which describes how patient-specific responses are subject to change subsequent formation of associations with treatments and care, implies prior experience with ineffective treatments, prior RCT involvement, stress of being off of regular medications, side effects and input from providers can impact response.15 Biologics are unique as dosing is infrequent and may require administration by study staff, thus minimizing patient non-compliance and reinforcing effects from frequent dosing.

Variation in study design can also alter placebo response. The state of chronic diseases often fluctuates, and patients exhibit spontaneous improvement. Additionally, psoriasis patients are more likely to seek medical care and enroll in RCTs during times of peak severity.16 These factors may be minimized with longer run-in periods, requirement of disease stability, longer placebo treatment duration, extended patient follow-up and averaging PASI scores within study periods. Furthermore, academic centers conducting RCTs frequently have higher recruitment rates, which can escalate rates of placebo response. While databases of willing trial participants help achieve recruitment goals, focusing recruitment effort on volunteers that are new to RCT participation and specific treatments can curtail this influence. Comparator trials and trials with greater randomization to active drug groups exhibit higher placebo responses.17 While comparator trials are effective in determining superiority between drugs of the same class, they should not be used to determine efficacy. When possible, addition of no treatment groups should be implemented to determine true drug efficacy.

Additional volunteer-specific factors may alter placebo responses. Trials have shown that augmenting the patient-practitioner relationship with attention can increase placebo rates incrementally.18 Therefore, providers should foster the same relationship with every patient and blinding providers, patients and outcome assessors is crucial. Determining how to ethically give informed consent while minimizing knowledge of active drug receipt and known side effects may also reduce placebo response.

Response rates vary by outcome measure utilized. Subjective measures typically evoke higher placebo response rates than objective measures.19 PASI is the most widely used efficacy assessment tool in psoriasis RCTs; however, it bears intrinsic subjectivity. Furthermore, PASI exhibits low sensitivity, limited utility (being a composite measure of multiple variables) and low accuracy; in addition, its reliability is dependent on experience.20 Technology may aid assessments of severity gradation and especially accurate affected body surface area assessments.21 Determining how inclusion criteria affect outcomes is also essential. “Eligibility creep” was coined to describe the observed tendency of researchers to exaggerate psoriasis severity to meet recruitment goals and maximize participation, subsequently increasing placebo responses. Placebo responses are higher in RCTs utilizing outcomes measures as inclusion criteria, so different measures should be used for study entry.

The observations described here must be interpreted within the context of the study design. Variability exists between each biologic medication and RCT. Additionally, many components of study design and implementation could not be captured, as they are not reported in study articles. Furthermore, differences exist in patient populations and patient-provider relationships. The low rates of placebo responders in psoriasis RCTs evaluating biologics implies continued scrutiny of study design, and RCT implementation is needed. Further, research and creation of research databases could add to the knowledge gained from this study. The lessons learned from these controlled trials should expedite improvements in the assessment of cosmetic and skin care products.


Send e-mail correspondence to:sonia.lamel@gmail.com.

  1. R Baran and HI Maibach, Textbook of Cosmetic Dermatology, 4th ed, Informa Healthcare USA, Inc., New York (2010)
  2. AO Barel, M Paye and HI Maibach, Handbook of Cosmetic Science and Technology, 3rd ed, Informa Healthcare USA, Inc., New York (2009)
  3. P Elsner and HI Maibach, Cosmeceuticals and Active Cosmetics: Drugs Versus Cosmetics, Taylor & Francis, Boca Raton, FL (2005)
  4. HK Beecher, The powerful placebo, J Amer Med Assoc 159 1602–1606 (1955)
  5. M Icen, CS Crowson, MT McEvoy, FJ Dann, SE Gabriel and H Maradit Kremers, Trends in incidence of adult-onset psoriasis over three decades: A population-based study, J Amer Acad of Dermatol 60 394–401 (2009)
  6. P Enck, S Klosterhalfen, K Weimer, B Horing and S Zipfel, The placebo response in clinical trials: More questions than answers, Philosophical transactions of the Royal Society of London Series B, Biological Sciences 366 1889–1895 (2011)
  7. A Hrobjartsson and PC Gotzsche, Is the placebo powerless? An analysis of clinical trials comparing placebo with no treatment, New England J Med 344 1594–1602 (2001)
  8. KA Katz, JH Karlawish, DS Chiang, RA Bognet, KJ Propert and DJ Margolis, Prevalence and factors associated with use of placebo control groups in randomized controlled trials in psoriasis: A cross-sectional study, J Amer Acad Dermatol 55 814–822 (2006)
  9. AB Gottlieb, RO Ford and MC Spellman, The efficacy and tolerability of clobetasol propionate foam 0.05% in the treatment of mild to moderate plaque-type psoriasis of nonscalp regions, J Cutaneous Med Surg 7 185–192 (2003)
  10. D Thiboutot, DM Pariser, N Egan, J Flores, JH Herndon Jr, NB Kanof et al., Adapalene gel 0.3% for the treatment of acne vulgaris: A multicenter, randomized, double-blind, controlled, phase III trial, J Amer Acad of Dermatol 54 242–250 (2006)
  11. JD Levine, NC Gordon and HL Fields, The mechanism of placebo analgesia, Lancet 2 654–657 (1978)
  12. AP Strafella, JH Ko and O Monchi, Therapeutic application of transcranial magnetic stimulation in Parkinson’s disease: The contribution of expectation, NeuroImage 31 1666–1672 (2006)
  13. U Bingel, L Colloca and L Vase, Mechanisms and clinical implications of the placebo effect: Is there a potential for the elderly? A mini-review, Gerontology 57 354–363 (2011)
  14. L Colloca and FG Miller, How placebo responses are formed: A learning perspective, Philosophical transactions of the Royal Society of London Series B, Biological Sciences 366 1859–1869 (2011)
  15. DA Stone, CE Kerr, E Jacobson, LA Conboy and TJ Kaptchuk, Patient expectations in placebo-controlled randomized clinical trials, J Evaluation In Clinical Practice 11 77–84 (2005)
  16. P Enck, S Klosterhalfen and S Zipfel, Novel study designs to investigate the placebo response, BMC Medical Research Methodology 11 90 (2009)
  17. PI Spuls, L Witkamp, PM Bossuyt and JD Bos, The course of chronic plaque-type psoriasis in placebo groups of randomized controlled studies, Archives of Dermatology 140 338–344 discussion 44 (2004)
  18. TJ Kaptchuk, JM Kelley, LA Conboy, RB Davis, CE Kerr and EE Jacobson et al., Components of placebo effect: Randomised controlled trial in patients with irritable bowel syndrome, BMJ 336 999–1003 (2008)
  19. K Meissner, H Distel and U Mitzdorf, Evidence for placebo effects on physical but not on biochemical outcome parameters: A review of clinical trials, BMC Medicine 5 3 (2007)
  20. PI Spuls, LL Lecluse, ML Poulsen, JD Bos, RS Stern and T Nijsten, How good are clinical severity and outcome measures for psoriasis?: Quantitative evaluation in a systematic review, J Invest Dermatol 130 933–943 (2010)
  21. S Kreft, M Kreft, A Resmam, P Marko and KZ Kreft, Computer-aided measurement of psoriatic lesion area in a multicenter clinical trial—comparison to physician’s estimations, J Dermatol Sci 44 21–27 (2006)


Table 1. Odds ratio of response determined by a random effects model (p < 0.0001)

Table 1. Odds ratio of response determined by a random effects model  (p < 0.0001)

The overall odds ratio calculated was 23.94 (p < 0.0001, 95% Confidence Interval (CI) 16.02–35.76).

Table 2. Multiple variable logistic regression model

Table 2. Multiple variable logistic regression model

The contribution of covariates to the rates of placebo responders

Next image >