- Email: [email protected]
Assessing placebo effects without placebo groups: an untapped possibility?
Pain 90 (2001) 201±203
www.elsevier.nl/locate/pain
Editorial
Assessing placebo effects without placebo groups: an untapped possibility? The study by Amanzio et al. in this issue of Pain makes several contributions, all of which suggest possibilities for alternative ways of assessing the contributions of placebo effects in analgesic treatments and for maximizing such effects among patients. Using an experimental approach similar to that ®rst developed by Gracely and colleagues (1983), two important ®ndings were found in both their experimental and clinical pain analyses. The ®rst was that analgesic effects of drugs were reduced by the psychological intervention of administering the drugs in a hidden as compared to an open infusion or by combining an open administration with an infusion of naloxone. The magnitudes of these reductions were very similar across both psychological and pharmacological interventions, supporting the hypothesis that the psychological intervention activated an endogenous opioid mechanism. This possibility is strengthened by recent studies from this same laboratory as well as by those conducted much earlier (Amanzio and Benedetti, 1999; Benedetti et al., 1999; Grevert et al., 1983; Levine et al., 1978). The second important ®nding was that much of the variability in analgesic response was eliminated by these same psychological and pharmacological interventions, suggesting the strong possibility that placebo mechanisms contribute a lot to the variability in analgesic effectiveness. The study is unique in that no placebo group is used in their analyses. Instead, they use a strategy of subtracting differences in pain intensities under conditions of hidden versus open injection of analgesics. Moreover, they apply this strategy to both clinical postoperative pain and to experimentally induced ischemic arm pain, a form of pain that closely models aspects of clinical pain. The two types of studies are complementary and their combination provides a consistent estimate of the magnitude of the placebo analgesic effect and its contribution to the variability in analgesic effectiveness. This study has several implications for new ways to conduct research on placebo analgesia and for optimizing placebo effects in patient care. I will elaborate on three implications not discussed in detail by the authors. The ®rst is that their results provide an alternative way of conceptualizing the magnitude of the placebo effect. A major component of the placebo effect measured in their study may be characterized as a difference score in contrast to a percentage change. As can be seen in Fig. 3 of their paper, placebo effects found for four different analgesics
appear as parallel shifts in the time-pain intensity curves, comparing open and hidden infusion conditions. These differences were about 1.0±1.5 pain-scale units and were similar at each time point for a given drug/condition. This makes the placebo effect larger percentage-wise as time elapses and pain declines (e.g. from about 17% at 15 min to about 32% at 60 min in the case of metamizol). This parallel shift in pain intensity over time is consistent with parallel reductions in nociceptive stimulus-response curves after administration of exogenous opioids, an unsurprising similarity considering the likely relationship of placebo analgesia to endogenous opioid mechanisms (Price et al., 1985, 1986; Fields and Price, 1997). A somewhat unexpected result of their study was that the magnitudes of placebo effects, measured as difference scores, were as large for experimental pain as they were for clinical pain (c.f. compare open and hidden ketorolac in Figs. 3 and 7), about 1.0±1.5 pain intensity scale units. These sizes of effects are also similar to those of other recent studies (Benedetti et al., 1999; Montgomery and Kirsch, 1997; Price et al., 1999). If magnitudes of placebo effects can be described as differences in pain intensity rather than as percentages, then these effects appear more consistent across different baseline pains and analgesics than percentage changes. For example, although difference scores were similar across the four conditions illustrated in Fig. 3 of their paper, percentage scores would be different. Buprenorphine would have a smaller percentage placebo effect (about 11± 15%) than metamizol (17±32%). How large and how consistent are the magnitudes of placebo analgesic effects when intravenous analgesics are given in postoperative pain? Without knowledge of baseline pain or natural history, the answer may be complex and/or meaningless if percentages are used, particularly in the case of only asking patients about percent relief (Beecher, 1955, 1959; Fields and Price, 1997; Turner et al., 1994; Wall, 1994). On the other hand, the answer may be simpler and more meaningful when expressed as differences in pain intensities between open and hidden injection. Another major implication of their study is that their approach suggests an alternative way of assessing the in¯uence of placebo in studies that do not have placebo conditions. Unlike experimental pain, whose intensity can be controlled to some extent by stimulus parameters, investigators must rely on the natural history of pain intensity in
0304-3959/01/$20.00 q 2001 International Association for the Study of Pain. Published by Elsevier Science B.V. All rights reserved. PII: S 0304-395 9(01)00265-2
202
D.D. Price / Pain 90 (2001) 201±203
clinical studies in order to assess the magnitude of the placebo effect (Fields and Price, 1997). As a result of several sources of variation, active and placebo treatments may be followed by reductions in pain as a consequence of natural history rather than treatment. Without a natural history group, having a placebo group in a study does not provide a measure of the placebo effect. For this reason, nearly all studies of clinical analgesia do not measure the placebo effect. They measure the difference between the placebo condition and the active treatment condition. Since both pharmacological and non-pharmacological studies of analgesia rarely include an untreated comparison group or condition (in cross-over studies), little knowledge exists with regard to the magnitude and time course of placebo effects or frequency of occurrence of placebo responses among individuals. The study by Amanzio et al. shows that a major component of the placebo effect can be measured without either a placebo or natural history group. Thus, both differences between results of placebo and natural history conditions and between results of open and hidden injections can provide an estimate of the placebo effect or at least its major component. This component is that which results from the patient's perception of the administration of the agent. This component is not completely identical with the placebo effect because there may be additional non-speci®c effects related to physical and psychological contexts, including interactions with healthcare professionals and perception of side effects. In addition, it is possible that there may be more than an additive interaction between the placebo and the active drug just as there are sometimes synergistic interactions between two or more active drugs. Many of the conditions necessary to achieve adequate control for the placebo effect are not satis®ed in studies of pain treatments wherein the active treatment can be subjectively distinguished from the control treatment. Examples of this problem abound in the literature on acupuncture analgesia. However, even most pharmacological studies of analgesic agents that have perceptible side effects have this same problem. What is needed are alternative ways of assessing and controlling for placebo effects and these potential alternatives may be integrally related to increased understanding of mechanisms of placebo analgesia. For example, recent studies have identi®ed factors that contribute to placebo analgesia, including past experience with effective analgesic agents (e.g. conditioning), expectations of analgesia, memory distortion, and possibly the desire for symptom change (Montgomery and Kirsch, 1997; Price and Fields, 1997; Price et al., 1999; Jensen and Karoly, 1991). Of these factors, that of expected pain intensities has recently been shown to account for the largest proportions in the variance in pain ratings during placebo conditions, about 49% in a study by Montgomery and Kirsch (1997) and between 25 and 36% in a study by Price et al. (1999). It is possible that measuring additional factors, such as desire for pain reduction, may improve the capacity to account for placebo-
related variance in future clinical studies of analgesia. If so, then adding measures of expectation of and desire for pain reduction to the kind of study presented by Amanzio et al. might offer an alternative method of evaluating the in¯uence of placebo in treatments. The feasibility of this approach is suggested by Amanzio and Benedetti's method of ascertaining the variability in analgesia that results from conditions of open injection. Under some circumstances, this alternative approach might even substitute for the conventional use of a placebo group. The clear advantage would be that no patient would have to be given a placebo treatment. This alternative approach may be considered heresy by traditionalists but could become more appropriate in light of the World Medical Organization's increased concerns about using placebo groups in clinical trials. Finally, the characterization of factors that mediate placebo effects and the ability to relate these factors to the variability of placebo analgesia suggests the possibility of enhancing placebo responses among patients. A hypothetical example would be simply telling the patient that the agent to be injected is known to produce powerful reductions in pain in some people. It would not be a lie and it could enhance the placebo response. The study by Amanzio et al. suggests that generally similar statements combined with other aspects of open injection of ketololac extend the analgesic effect by 1.0±1.5 pain intensity scale units (see Fig. 7B of their paper). Although this increase may not be clinically signi®cant in itself, both pain research scientists and the pharmaceutical industry go to the ends of the earth to make improvements of this magnitude. Adding one or two sentences to each pain treatment might help to produce them.
References Amanzio M, Benedetti F. Neuropharmacological dissection of placebo analgesia: expectation-activated opioid systems versus conditioningactivated speci®c subsystems. J Neurosci 1999;19:484±494. Beecher HK. The powerful placebo. J Am Med Assoc 1955;159:1602± 1606. Beecher HK. Measurement of subjective responses: quantitative effects of drugs, New York: Oxford University Press, 1959. Benedetti F, Arduino C, Amanzio M. Somatotopic activation of opioid systems by target-directed expectations of analgesia. J Neurosci 1999;19(9):3639±3648. Fields HL, Price DD. Toward a neurobiology of placebo analgesia. In: Harrington A, editor. Placebo: probing the self-healing brain, Boston, MA: Harvard University Press, 1997. Gracely RH, Dubner R, Wolskee PJ, Deeter WR. Placebo and naloxone can alter postsurgical pain by separate mechanisms. Nature 1983;306:264± 265. Grevert P, Albert LH, Goldstein A. Partial antagonism of placebo analgesia by naloxone. Pain 1983;16:129±143. Jensen MP, Karoly P. Motivation and expectancy factors in symptom perception: a laboratory study of the placebo effect. Psychosomat Med 1991;53:144±152. Levine JD, Gordon NC, Fields HL. The mechanism of placebo analgesia. Lancet 1978;2(8091):654±657. Montgomery GH, Kirsch I. Classical conditioning and the placebo effect. Pain 1997;72:103±113.
D.D. Price / Pain 90 (2001) 201±203 Price DD, Fields HL. Where are the causes of placebo analgesia? An experiential behavioral analysis. Pain Forum 1997;6(1):44±52. Price DD, von der Gruen A, Miller J, Ra®i A, Price C. A psychophysical analysis of morphine analgesia. Pain 1985;22:320±330. Price DD, Harkins SW, Ra®i A, Price C. A simultaneous comparison of fentanyl's analgesic effects on experimental and clinical pain. Pain 1986;24:197±203. Price DD, Milling LS, Kirsch I, Duff A, Montgomery GH, Nicholls SS. An analysis of factors that contribute to the magnitude of placebo analgesia. Pain 1999;83(2):147±156. Turner JA, Deyo RA, Loeser JD, Von Korff M, Fordyce WE. The importance of placebo effects in pain treatment and research. J Am Med Assoc 1994;271(20):1609±1614. Wall PD. The placebo and the placebo response. In: Wall PD, Melzack R, editors. Textbook of pain, New York: Churchill Livingstone, 1994.
203
Donald D. Price a,b,* Department of Oral and Maxillofacial Surgery, University of Florida, Gainesville, FL, 32605, USA b Department of Neuroscience, University of Florida, Gainesville, FL, 32605, USA a
* Oral and Maxillofacial Surgery, JHMHC, Box 100416, 1600 Southwest Archer Road, Gainesville, FL 32610, USA. E-mail address: [email protected]¯.edu
www.elsevier.nl/locate/pain
Editorial
Assessing placebo effects without placebo groups: an untapped possibility? The study by Amanzio et al. in this issue of Pain makes several contributions, all of which suggest possibilities for alternative ways of assessing the contributions of placebo effects in analgesic treatments and for maximizing such effects among patients. Using an experimental approach similar to that ®rst developed by Gracely and colleagues (1983), two important ®ndings were found in both their experimental and clinical pain analyses. The ®rst was that analgesic effects of drugs were reduced by the psychological intervention of administering the drugs in a hidden as compared to an open infusion or by combining an open administration with an infusion of naloxone. The magnitudes of these reductions were very similar across both psychological and pharmacological interventions, supporting the hypothesis that the psychological intervention activated an endogenous opioid mechanism. This possibility is strengthened by recent studies from this same laboratory as well as by those conducted much earlier (Amanzio and Benedetti, 1999; Benedetti et al., 1999; Grevert et al., 1983; Levine et al., 1978). The second important ®nding was that much of the variability in analgesic response was eliminated by these same psychological and pharmacological interventions, suggesting the strong possibility that placebo mechanisms contribute a lot to the variability in analgesic effectiveness. The study is unique in that no placebo group is used in their analyses. Instead, they use a strategy of subtracting differences in pain intensities under conditions of hidden versus open injection of analgesics. Moreover, they apply this strategy to both clinical postoperative pain and to experimentally induced ischemic arm pain, a form of pain that closely models aspects of clinical pain. The two types of studies are complementary and their combination provides a consistent estimate of the magnitude of the placebo analgesic effect and its contribution to the variability in analgesic effectiveness. This study has several implications for new ways to conduct research on placebo analgesia and for optimizing placebo effects in patient care. I will elaborate on three implications not discussed in detail by the authors. The ®rst is that their results provide an alternative way of conceptualizing the magnitude of the placebo effect. A major component of the placebo effect measured in their study may be characterized as a difference score in contrast to a percentage change. As can be seen in Fig. 3 of their paper, placebo effects found for four different analgesics
appear as parallel shifts in the time-pain intensity curves, comparing open and hidden infusion conditions. These differences were about 1.0±1.5 pain-scale units and were similar at each time point for a given drug/condition. This makes the placebo effect larger percentage-wise as time elapses and pain declines (e.g. from about 17% at 15 min to about 32% at 60 min in the case of metamizol). This parallel shift in pain intensity over time is consistent with parallel reductions in nociceptive stimulus-response curves after administration of exogenous opioids, an unsurprising similarity considering the likely relationship of placebo analgesia to endogenous opioid mechanisms (Price et al., 1985, 1986; Fields and Price, 1997). A somewhat unexpected result of their study was that the magnitudes of placebo effects, measured as difference scores, were as large for experimental pain as they were for clinical pain (c.f. compare open and hidden ketorolac in Figs. 3 and 7), about 1.0±1.5 pain intensity scale units. These sizes of effects are also similar to those of other recent studies (Benedetti et al., 1999; Montgomery and Kirsch, 1997; Price et al., 1999). If magnitudes of placebo effects can be described as differences in pain intensity rather than as percentages, then these effects appear more consistent across different baseline pains and analgesics than percentage changes. For example, although difference scores were similar across the four conditions illustrated in Fig. 3 of their paper, percentage scores would be different. Buprenorphine would have a smaller percentage placebo effect (about 11± 15%) than metamizol (17±32%). How large and how consistent are the magnitudes of placebo analgesic effects when intravenous analgesics are given in postoperative pain? Without knowledge of baseline pain or natural history, the answer may be complex and/or meaningless if percentages are used, particularly in the case of only asking patients about percent relief (Beecher, 1955, 1959; Fields and Price, 1997; Turner et al., 1994; Wall, 1994). On the other hand, the answer may be simpler and more meaningful when expressed as differences in pain intensities between open and hidden injection. Another major implication of their study is that their approach suggests an alternative way of assessing the in¯uence of placebo in studies that do not have placebo conditions. Unlike experimental pain, whose intensity can be controlled to some extent by stimulus parameters, investigators must rely on the natural history of pain intensity in
0304-3959/01/$20.00 q 2001 International Association for the Study of Pain. Published by Elsevier Science B.V. All rights reserved. PII: S 0304-395 9(01)00265-2
202
D.D. Price / Pain 90 (2001) 201±203
clinical studies in order to assess the magnitude of the placebo effect (Fields and Price, 1997). As a result of several sources of variation, active and placebo treatments may be followed by reductions in pain as a consequence of natural history rather than treatment. Without a natural history group, having a placebo group in a study does not provide a measure of the placebo effect. For this reason, nearly all studies of clinical analgesia do not measure the placebo effect. They measure the difference between the placebo condition and the active treatment condition. Since both pharmacological and non-pharmacological studies of analgesia rarely include an untreated comparison group or condition (in cross-over studies), little knowledge exists with regard to the magnitude and time course of placebo effects or frequency of occurrence of placebo responses among individuals. The study by Amanzio et al. shows that a major component of the placebo effect can be measured without either a placebo or natural history group. Thus, both differences between results of placebo and natural history conditions and between results of open and hidden injections can provide an estimate of the placebo effect or at least its major component. This component is that which results from the patient's perception of the administration of the agent. This component is not completely identical with the placebo effect because there may be additional non-speci®c effects related to physical and psychological contexts, including interactions with healthcare professionals and perception of side effects. In addition, it is possible that there may be more than an additive interaction between the placebo and the active drug just as there are sometimes synergistic interactions between two or more active drugs. Many of the conditions necessary to achieve adequate control for the placebo effect are not satis®ed in studies of pain treatments wherein the active treatment can be subjectively distinguished from the control treatment. Examples of this problem abound in the literature on acupuncture analgesia. However, even most pharmacological studies of analgesic agents that have perceptible side effects have this same problem. What is needed are alternative ways of assessing and controlling for placebo effects and these potential alternatives may be integrally related to increased understanding of mechanisms of placebo analgesia. For example, recent studies have identi®ed factors that contribute to placebo analgesia, including past experience with effective analgesic agents (e.g. conditioning), expectations of analgesia, memory distortion, and possibly the desire for symptom change (Montgomery and Kirsch, 1997; Price and Fields, 1997; Price et al., 1999; Jensen and Karoly, 1991). Of these factors, that of expected pain intensities has recently been shown to account for the largest proportions in the variance in pain ratings during placebo conditions, about 49% in a study by Montgomery and Kirsch (1997) and between 25 and 36% in a study by Price et al. (1999). It is possible that measuring additional factors, such as desire for pain reduction, may improve the capacity to account for placebo-
related variance in future clinical studies of analgesia. If so, then adding measures of expectation of and desire for pain reduction to the kind of study presented by Amanzio et al. might offer an alternative method of evaluating the in¯uence of placebo in treatments. The feasibility of this approach is suggested by Amanzio and Benedetti's method of ascertaining the variability in analgesia that results from conditions of open injection. Under some circumstances, this alternative approach might even substitute for the conventional use of a placebo group. The clear advantage would be that no patient would have to be given a placebo treatment. This alternative approach may be considered heresy by traditionalists but could become more appropriate in light of the World Medical Organization's increased concerns about using placebo groups in clinical trials. Finally, the characterization of factors that mediate placebo effects and the ability to relate these factors to the variability of placebo analgesia suggests the possibility of enhancing placebo responses among patients. A hypothetical example would be simply telling the patient that the agent to be injected is known to produce powerful reductions in pain in some people. It would not be a lie and it could enhance the placebo response. The study by Amanzio et al. suggests that generally similar statements combined with other aspects of open injection of ketololac extend the analgesic effect by 1.0±1.5 pain intensity scale units (see Fig. 7B of their paper). Although this increase may not be clinically signi®cant in itself, both pain research scientists and the pharmaceutical industry go to the ends of the earth to make improvements of this magnitude. Adding one or two sentences to each pain treatment might help to produce them.
References Amanzio M, Benedetti F. Neuropharmacological dissection of placebo analgesia: expectation-activated opioid systems versus conditioningactivated speci®c subsystems. J Neurosci 1999;19:484±494. Beecher HK. The powerful placebo. J Am Med Assoc 1955;159:1602± 1606. Beecher HK. Measurement of subjective responses: quantitative effects of drugs, New York: Oxford University Press, 1959. Benedetti F, Arduino C, Amanzio M. Somatotopic activation of opioid systems by target-directed expectations of analgesia. J Neurosci 1999;19(9):3639±3648. Fields HL, Price DD. Toward a neurobiology of placebo analgesia. In: Harrington A, editor. Placebo: probing the self-healing brain, Boston, MA: Harvard University Press, 1997. Gracely RH, Dubner R, Wolskee PJ, Deeter WR. Placebo and naloxone can alter postsurgical pain by separate mechanisms. Nature 1983;306:264± 265. Grevert P, Albert LH, Goldstein A. Partial antagonism of placebo analgesia by naloxone. Pain 1983;16:129±143. Jensen MP, Karoly P. Motivation and expectancy factors in symptom perception: a laboratory study of the placebo effect. Psychosomat Med 1991;53:144±152. Levine JD, Gordon NC, Fields HL. The mechanism of placebo analgesia. Lancet 1978;2(8091):654±657. Montgomery GH, Kirsch I. Classical conditioning and the placebo effect. Pain 1997;72:103±113.
D.D. Price / Pain 90 (2001) 201±203 Price DD, Fields HL. Where are the causes of placebo analgesia? An experiential behavioral analysis. Pain Forum 1997;6(1):44±52. Price DD, von der Gruen A, Miller J, Ra®i A, Price C. A psychophysical analysis of morphine analgesia. Pain 1985;22:320±330. Price DD, Harkins SW, Ra®i A, Price C. A simultaneous comparison of fentanyl's analgesic effects on experimental and clinical pain. Pain 1986;24:197±203. Price DD, Milling LS, Kirsch I, Duff A, Montgomery GH, Nicholls SS. An analysis of factors that contribute to the magnitude of placebo analgesia. Pain 1999;83(2):147±156. Turner JA, Deyo RA, Loeser JD, Von Korff M, Fordyce WE. The importance of placebo effects in pain treatment and research. J Am Med Assoc 1994;271(20):1609±1614. Wall PD. The placebo and the placebo response. In: Wall PD, Melzack R, editors. Textbook of pain, New York: Churchill Livingstone, 1994.
203
Donald D. Price a,b,* Department of Oral and Maxillofacial Surgery, University of Florida, Gainesville, FL, 32605, USA b Department of Neuroscience, University of Florida, Gainesville, FL, 32605, USA a
* Oral and Maxillofacial Surgery, JHMHC, Box 100416, 1600 Southwest Archer Road, Gainesville, FL 32610, USA. E-mail address: [email protected]¯.edu