Calibrated use of deception in assessing the placebo effect
Alberto JEspay, Principal Investigator, University of Cincinnatiaespay@gmail.com
Alok Dwivedi, El Paso, TX; Anthony E. Lang, Toronto, CA; Michael J. Linke, Cincinnati, OH; Jerzy P. Szaflarski, Birmingham, AL
Submitted February 11, 2015
We thank editorialists Drs. LeWitt and Kim [1] and WriteClick submitter Dr. Kelley for their thoughtful feedback on our clinical trial examining the effect of cost on the placebo response. [2] We welcome the discussion on alternative explanations of our results and embrace the suggestions of exploring "authorized deception" as a tool to incorporate into future trials. We also agree that an independent party should collect post-debriefing data on issues of trust, justification of research, and willingness to participate in future studies from subjects participating in studies involving deception.
Given the number of additional issues raised by Drs. LeWitt, Kim, and Kelley, we have itemized our response here:
(1) Low sample size. We calculated the power needed to show significant differences between groups as 12. With a larger sample size, the magnitude of statistical significance would have increased rather than decreased without adding any more validity to the findings.
(2) Results were confounded by "treatment by period effect."
It is noteworthy that ignoring the second period data also demonstrated a 10% greater improvement in the "expensive" compared to the "cheap" placebo.
Stratified analysis by order showed 14% improvement in "expensive" placebo when this placebo was administered first compared to the "cheap" placebo and 7% improvement when cheap placebo was administered first.
(3) $100 per dose was considered "cheap" relative to the $1,500 dose.
Subjects learned the price of both interventions at the outset. While $100 might never be "cheap" for a single treatment, it is compared to a similar intervention costing 15 times more.
(4) While the "cheap" placebo significantly improved motor function, such improvement was also significantly lower than that of levodopa. On the other hand, the magnitude of benefit of the "expensive" placebo was not significantly lower than that of levodopa (admittedly, a larger sample would have been expected to show a difference). These findings suggest that both are useful to establish therapeutic effect of a treatment in placebo controlled trial but cost-matched placebo may provide a more appropriate efficacy comparator for a treatment.
(5) Our study does not advocate for the use of placebo in general practice. It advocates for the interpretation of these findings as suggestive that there is an untapped opportunity to magnify the magnitude of benefits of standard interventions by enhancing their perception of efficacy.
(6) The implications to the threat to the physician-patient relationship are discussed in the original article. All patients were from the lead author's own clinic (per IRB request) and all have remained his patients after the study. Debriefing them about the nature of the study did not end the patient-physician relationship and did not prevent their engagement in other clinical research opportunities (half of them have enrolled since in other studies). They understood that the scientific question being probed required the calibrated use of deception.
1. LeWitt PA, Kim S. The pharmacodynamics of placebo: Expectation effects
of price as a proxy for efficacy. Neurology 2015;84:1-2.
2. Espay AJ, Norris MM, Eliassen JC, Dwivedi A, Smith MS, Banks C, et al.
Placebo effect of medication cost in parkinson disease: A randomized double-blind study. Neurology 2015; 0: WNL.0000000000001282v1-101212000
For disclosures, contact the editorial office at journal@neurology.org.
We thank editorialists Drs. LeWitt and Kim [1] and WriteClick submitter Dr. Kelley for their thoughtful feedback on our clinical trial examining the effect of cost on the placebo response. [2] We welcome the discussion on alternative explanations of our results and embrace the suggestions of exploring "authorized deception" as a tool to incorporate into future trials. We also agree that an independent party should collect post-debriefing data on issues of trust, justification of research, and willingness to participate in future studies from subjects participating in studies involving deception.
Given the number of additional issues raised by Drs. LeWitt, Kim, and Kelley, we have itemized our response here:
(1) Low sample size. We calculated the power needed to show significant differences between groups as 12. With a larger sample size, the magnitude of statistical significance would have increased rather than decreased without adding any more validity to the findings.
(2) Results were confounded by "treatment by period effect." It is noteworthy that ignoring the second period data also demonstrated a 10% greater improvement in the "expensive" compared to the "cheap" placebo. Stratified analysis by order showed 14% improvement in "expensive" placebo when this placebo was administered first compared to the "cheap" placebo and 7% improvement when cheap placebo was administered first.
(3) $100 per dose was considered "cheap" relative to the $1,500 dose. Subjects learned the price of both interventions at the outset. While $100 might never be "cheap" for a single treatment, it is compared to a similar intervention costing 15 times more.
(4) While the "cheap" placebo significantly improved motor function, such improvement was also significantly lower than that of levodopa. On the other hand, the magnitude of benefit of the "expensive" placebo was not significantly lower than that of levodopa (admittedly, a larger sample would have been expected to show a difference). These findings suggest that both are useful to establish therapeutic effect of a treatment in placebo controlled trial but cost-matched placebo may provide a more appropriate efficacy comparator for a treatment.
(5) Our study does not advocate for the use of placebo in general practice. It advocates for the interpretation of these findings as suggestive that there is an untapped opportunity to magnify the magnitude of benefits of standard interventions by enhancing their perception of efficacy.
(6) The implications to the threat to the physician-patient relationship are discussed in the original article. All patients were from the lead author's own clinic (per IRB request) and all have remained his patients after the study. Debriefing them about the nature of the study did not end the patient-physician relationship and did not prevent their engagement in other clinical research opportunities (half of them have enrolled since in other studies). They understood that the scientific question being probed required the calibrated use of deception.
1. LeWitt PA, Kim S. The pharmacodynamics of placebo: Expectation effects of price as a proxy for efficacy. Neurology 2015;84:1-2.
2. Espay AJ, Norris MM, Eliassen JC, Dwivedi A, Smith MS, Banks C, et al. Placebo effect of medication cost in parkinson disease: A randomized double-blind study. Neurology 2015; 0: WNL.0000000000001282v1-101212000
For disclosures, contact the editorial office at journal@neurology.org.