Response expectancies in placebo analgesia and their clinical relevance

Pain 93 (2001) 77±84

www.elsevier.nl/locate/pain

Response expectancies in placebo analgesia and their clinical relevance Antonella Pollo a, Martina Amanzio a, Anna Arslanian b, Caterina Casadio b, Giuliano Maggi b, Fabrizio Benedetti a,c,* a

Department of Neuroscience, University of Torino Medical School, 10125 Torino, Italy b Division of Thoracic Surgery, University of Torino, 10126 Torino, Italy c Rita Levi-Montalcini Center for Brain Repair, University of Torino, 10125 Torino, Italy Received 8 November 2000; received in revised form 22 January 2001; accepted 30 January 2001

Abstract Response expectancies have been proposed as the major determinant of placebo effects. Here we report that different expectations produce different analgesic effects which in turn can be harnessed in clinical practice. Thoracotomized patients were treated with buprenorphine on request for 3 consecutive days, together with a basal intravenous infusion of saline solution. However, the symbolic meaning of this basal infusion was changed in three different groups of patients. The ®rst group was told nothing about any analgesic effect (natural history). The second group was told that the basal infusion was either a powerful painkiller or a placebo (classic double-blind administration). The third group was told that the basal infusion was a potent painkiller (deceptive administration). Therefore, whereas the analgesic treatment was exactly the same in the three groups, the verbal instructions about the basal infusion differed. The placebo effect of the saline basal infusion was measured by recording the doses of buprenorphine requested over the three-days treatment. We found that the double-blind group showed a reduction of buprenorphine requests compared to the natural history group. However, this reduction was even larger in the deceptive administration group. Overall, after 3 days of placebo infusion, the ®rst group received 11.55 mg of buprenorphine, the second group 9.15 mg, and the third group 7.65 mg. Despite these dose differences, analgesia was the same in the three groups. These results indicate that different verbal instructions about certain and uncertain expectations of analgesia produce different placebo analgesic effects, which in turn trigger a dramatic change of behaviour leading to a signi®cant reduction of opioid intake. q 2001 Elsevier Science B.V. Published by Elsevier Science B.V. All rights reserved. Keywords: Placebo; Expectancy; Analgesia; Buprenorphine; Postoperative pain

1. Introduction Several theories have been proposed to explain the placebo effect, and many of them are concerned with pain. The anxiety theory states that placebo analgesia is due to a reduction of anxiety (McGlashan et al., 1969; Evans, 1977). The conditioning theory assumes that the placebo effect is a conditioned response due to repeated associations between a conditioned stimulus (e.g. shape and colour of aspirin pills) and an unconditioned stimulus (the active substance of aspirin) (Gleidman et al., 1957; Herrnstein, 1962; Wickramasekera, 1980; Siegel, 1985; Voudouris et al., 1989, 1990; Ader, 1997). The cognitive theory proposes that expectations and beliefs of analgesia play an essential role (Bootzin, 1985; Kirsch, 1985, 1990; Montgomery and Kirsch, 1997; Price and Fields, 1997; Price et al., 1999), and the response* Corresponding author. Tel.: 139-011-6707709; fax: 139-0116707708. E-mail address: [email protected] (F. Benedetti).

appropriate sensation hypothesis states that the global experience of pain results from a complex analysis of different brain states (Wall, 1993). These theories are not necessarily in con¯ict because each of them may represent a different aspect of the same phenomenon (Wall, 1992). In addition, many studies show that different placebo effects can be mediated by different mechanisms, like conditioning or expectation (Amanzio and Benedetti, 1999; Benedetti et al., 1999a,b; Price, 2000). One of the most interesting mechanisms underlying the placebo effect is represented by response expectancies, that is, those expectations which an individual holds about his/ her emotional and physiological responses, such as anxiety and mood (Kirsch, 1985, 1990). These response expectancies appear to be space speci®c for they can be speci®cally obtained in different parts of the body (Montgomery and Kirsch, 1996; Benedetti et al., 1999b; Price et al., 1999). Interestingly, Kirsch and Weixel (1988) found that differences between double-blind and deceptive administration of placebos can produce different outcomes, and concluded

0304-3959/01/$20.00 q 2001 Elsevier Science B.V. Published by Elsevier Science B.V. All rights reserved. PII: S 0304-395 9(01)00296-2

78

A. Pollo et al. / Pain 93 (2001) 77±84

that this is due to the different experimental designs which in turn produce different expectancies. In that study, coffee and decaffeinated coffee were administered following different verbal instructions. In one case, they were given according to the usual double-blind paradigm, in the other case decaffeinated coffee was deceptively presented as real coffee. The authors proposed that the deceptive administration would produce stronger effects than the double-blind protocol because in the double-blind administration expectations are uncertain. In fact, in the double-blind protocol the subject does not know whether the active substance or a placebo is being administered. Taking into account all these considerations, we wanted to study the role of response expectancies in the clinical pharmacological setting, in order to compare placebo analgesia obtained through the classic double-blind design with placebo analgesia produced by the deceptive administration of a placebo. 2. Methods A total of 38 patients participated in the study after they gave their informed consent to receive either a painkiller or a placebo and after approval by our local ethics committee. They underwent thoracic surgery for lung cancer and did not have any health problem other than those tightly related to the neoplasm. All patients with renal, hepatic, cardiovascular and metabolic problems had previously been excluded from the study. The operation was a standard posterolateral thoracotomy with the resection of at least three of the following muscles: latissimus dorsi, serratus anterior, trapezius and rhomboid. Anaesthesia was induced in all cases with fentanyl 150±250 mg i.v. and maintained with a combination of iso¯urane and oxygen. Paralysis was achieved by means of atracurium 30±60 mg and reversed with 1 mg atropine and 2 mg neostigmine i.v. Ondansetron 4 mg was used as an antiemetic drug. All patients who required additional drugs or blood transfusions were excluded from the study. One hour after recovery from anaesthesia, the analgesic treatment was started. A ®rst bolus of 0.1 mg of buprenorphine was given in about 3 min. After 15 min a second bolus of 0.1 mg was given with the same infusion time. Finally, a third bolus of 0.1 mg was administered after 15 min with the same infusion time of 3 min. At this point, a basal intravenous infusion of saline solution (NaCl 0.9%, 20 ml/h) was started and continued for 3 consecutive days. During this period the analgesic treatment was based on buprenorphine on verbal request, in which buprenorphine doses of 0.15 mg were administered when requested by the patients. The drug administrator was completely blind as to the treatment group to which the patient had been assigned. The therapeutic protocol is shown in Fig. 1, together with the subdivision of the patients into three groups. These groups differed only for the verbal instructions about the basal infusion. The ®rst group (Table 1) was used to furnish data about the

natural history of analgesia. These patients were told nothing about any analgesic effect of the basal infusion. They simply knew that the infusion was a rehydrating solution. The second group was studied according to the usual double-blind paradigm. The basal infusion was either buprenorphine (0.01 mg/h in NaCl 0.9%) or placebo (only NaCl 0.9%). According to the double-blind administration, these patients were told that the medication could be either a painkiller or a placebo. The double-blind protocol was run by administering buprenorphine to four patients and placebo to the remaining ten. Only the data of these ten patients will be shown (Table 1), since the other four were mainly used to allow the double-blind administration. The patients of the third group received exactly the same double-blind treatment as the second group. However, the double-blind administration was slightly modi®ed as far as the verbal instructions are concerned. In fact, the patients were told that the basal infusion was a powerful painkiller, even though they actually were receiving either buprenorphine or placebo. Therefore, those patients who actually were receiving the placebo were sure that the basal infusion was a painkiller. In this case also, the double-blind paradigm was run by giving buprenorphine to four patients and placebo to the other ten. Only the data of these ten patients will be shown (Table 1). Thus the second and third groups differed for the verbal instructions. In fact, whereas the patients of Group 2 were uncertain about whether they actually were receiving a painkiller or a placebo (classic double-blind design), the patients of Group 3 were certain that a powerful analgesic drug was being administered (deceptive administration for those ten patients who actually were receiving the placebo). The effects of these different verbal instructions were measured by recording the number of buprenorphine requests during the 3-days analgesic treatment. In addition,

Fig. 1. Experimental paradigm and therapeutic protocol. Initial doses and doses on request of buprenorphine represent the real analgesic treatment of the postoperative patients of the present study (bold line). In addition, a basal infusion of saline solution was administered (broken line) by changing its symbolic meaning in three different groups of patients. The placebo analgesic effect was measured by recording the number of doses of buprenorphine for 3 consecutive days.

a

64 73 54 76 79 62 77 56 67 54 66.2 ^ 9.7

LD,SA,T LD,T,R LD,SA,T LD,SA,T LD,SA,T LD,SA,R LD,SA,R LD,T,R LD,SA,R LD,T,R

A B C D E F G H I J

LD, latissimus dorsi; SA, serratus anterior; T, trapezius; R, rhomboid.

57 58 58 65 66 60 57 69 63 60 61.3 ^ 4.22

m m f m f m m f m f

55 67 54 62 64 60 65 68 57 55 60.7 ^ 5.25

Age (years) 69 78 51 73 58 75 50 70 60 52 63.6 ^ 10.64

Weight (kg)

m f f m m m f m m m

Sex

A B C D E F G H I J Mean ^ SD

Muscles resected

Patient

Weight (kg)

Sex

Patient

Age (years)

Second group (double-blind design)

First group (natural history)

Table 1 Characteristics of the patients of the three groups, including muscle resection during surgery a

LD,SA,R LD,SA,T LD,T,R LD,SA,R LD,SA,R LD,T,R LD,SA,T LD,SA,R LD,SA,T LD,T,R

Muscles resected A B C D E F G H I J

Patient m f m m m m f m f m

Sex

57 65 70 63 65 56 55 59 54 60 60.4 ^ 5.21

Age (years)

76 78 61 63 71 67 64 57 58 53 64.8 ^ 8.22

Weight (kg)

Third group (deceptive administration)

LD,T,R LD,T,R LD,SA,R LD,SA,T LD,SA,T LD,SA,T LD,T,R LD,SA,R LD,SA,T LD,T,R

Muscles resected

A. Pollo et al. / Pain 93 (2001) 77±84 79

80

A. Pollo et al. / Pain 93 (2001) 77±84

the patients were told to report their pain every hour according to a numerical rating scale (NRS) ranging from 0 ˆ no pain to 10 ˆ unbearable pain. This pain rating was assessed either by a doctor or by a relative during the night (when the patient was not sleeping) for 3 consecutive days. Sometimes, on the 2nd and 3rd day the patient him/herself reported on a diary the NRS. The results were analysed by means of the analysis of variance (ANOVA) followed by the Newman±Keuls test for multiple comparisons. In addition, the data of each patient will be shown.

3. Results The three groups of patients did not differ for sex, age (F…2; 27† ˆ 0:09, P ˆ 0:917), weight (F…2; 27† ˆ 0:18, P ˆ 0:832) and type of surgery, as shown in Table 1. The number of buprenorphine doses received by the patients of the three groups during the 3-days treatment differed considerably, as shown in Fig. 2. It can be seen that there was a reduction of the doses of buprenorphine with the doubleblind administration of the basal infusion compared to the natural history. This reduction was even larger with the deceptive administration of the basal infusion. Overall, the total dose of buprenorphine received at the end of the 3-days treatment was 11.55 mg in the natural history group, 9.15 mg in the double-blind group, which is a reduction of 20.8%, and 7.65 mg in the deceptive administration group, which represents a decrease of 33.8% with respect to natural history and of 16.4% compared with double-blind administration (Fig. 3).

Fig. 2. Number of patients who requested different doses of buprenorphine. (A) Patients who did not receive any particular information about the saline basal infusion. They only knew it was a rehydrating solution. (B) Patients who underwent the classic double-blind administration of the saline basal infusion. They knew it could be either a painkiller or a placebo (C) Patients who received the deceptive administration of the saline basal infusion. They believed it was a painkiller.

Fig. 3. Total dose of buprenorphine received at the end of the 3-days analgesic treatment in the three groups of patients. The three different verbal instructions about the saline basal infusion produced different buprenorphine intake.

A. Pollo et al. / Pain 93 (2001) 77±84

81

Table 2 Number of doses and total dose of buprenorphine received at the end of the analgesic treatment for each patient First group (natural history)

Second group (double-blind)

Third group (deceptive administration)

Patient

Number of doses

Total dose (mg)

Patient

Number of doses

Total dose (mg)

Patient

Number of doses

Total dose (mg)

A B C D E F G H I J

7 6 6 7 8 7 7 6 5 8

1.2 1.05 1.05 1.2 1.35 1.2 1.2 1.05 0.9 1.35

A B C D E F G H I J

6 6 5 5 6 5 4 5 5 4

1.05 1.05 0.9 0.9 1.05 0.9 0.75 0.9 0.9 0.75

A B C D E F G H I J

6 5 4 4 3 3 4 3 4 5

1.05 0.9 0.75 0.75 0.6 0.6 0.75 0.6 0.75 0.9

Mean ^ SD

6.7 ^ 0.95

1.15 ^ 0.14

5.1 ^ 0.74

0.91 ^ 0.11

4.1 ^ 0.99

0.76 ^ 0.15

These differences are statistically signi®cant, as shown by the data of each patient in Table 2. Whereas the mean dose of buprenorphine received by the ®rst group at the end of the 3 days was 1.15 ^ 0.14 mg, the second group received a mean dose of 0.91 ^ 0.11 mg and the third group 0.76 mg ^0.15 (F…2; 27† ˆ 21:42, P , 0:001). The Newman±Keuls test showed that the ®rst group differed from the second (q ˆ 5:646, P , 0:01), the second from the third (q ˆ 3:529, P , 0:01), and the ®rst from the third (q ˆ 9:175, P , 0:01). Table 2 also shows the number of buprenorphine doses received by each patient. Overall, the reduction of buprenorphine requests in the second group with respect to the ®rst shows that a placebo analgesic effect of the saline basal infusion occurred. Likewise, the larger reduction of buprenorphine requests in the third group relative to the second indicates that the placebo analgesic response was stronger in the former. This is also shown in Fig. 4, where it can be seen that the time course of pain was the same in the three groups of patients (A) in spite of the different doses of buprenorphine administered (B). In other words, the same analgesic effect was obtained with signi®cantly different doses of buprenorphine, indicating that the larger the reduction of buprenorphine requests, the stronger the placebo analgesic effect. We also considered the possibility that the patients of the second and third groups could have requested less doses of buprenorphine because they felt reassured by the supposed long-lasting effect of the continuous basal infusion. By contrast, the patients of the ®rst group could have requested more buprenorphine doses simply because they were afraid of remaining without any painkiller, for example, during the night. If this was true, there should be differences between the three groups in pain rating at the time of the buprenorphine requests. However, that this was not the case is shown in detail in Table 3. It can be seen that there were no differences between the patients of the three groups (above), since in each case buprenorphine was requested when the NRS was around 4.5±5 (no statistically signi®cant differences are

Fig. 4. Time course of pain (A) and total amount of buprenorphine received (B) in the three groups of patients. Empty squares: natural history. Empty circles: patients who received the placebo with the classic double-blind design. Black circles: patients who received the deceptive administration of the placebo. In (A) the missing circles mean that NRS could not be recorded because most of the patients were sleeping. In (B) each measure represents the total dose from time 0; therefore, the last measures on the right indicate the total doses at the end of the treatment. Note that the same analgesic effect (A) was obtained with different doses of buprenorphine (B).

82

A. Pollo et al. / Pain 93 (2001) 77±84

Table 3 Mean pain intensity (NRS) ^ standard deviation at which buprenorphine was requested, in each patient (above) and for each dose (below) a First group (natural history)

Second group (double-blind)

Third group (deceptive administration)

Patient

Pain intensity at which buprenorphine was requested (NRS)

Patient

Pain intensity at which buprenorphine was requested (NRS)

Patient

Pain intensity at which buprenorphine was requested (NRS)

A B C D E F G H I J

5.4 ^ 0.65 4.7 ^ 0.67 4.7 ^ 0.67 4.42 ^ 0.97 5.33 ^ 1.47 5.08 ^ 0.58 5.5 ^ 1.46 5.17 ^ 0.88 5.25 ^ 1.04 4.9 ^ 0.65

A B C D E F G H I J

5.1 ^ 0.48 4.72 ^ 1.02 5.45 ^ 0.88 5.28 ^ 1.05 5.17 ^ 0.96 4.9 ^ 0.75 5.1 ^ 0.9 5.3 ^ 1.1 4.62 ^ 1.12 5.23 ^ 0.87

A B C D E F G H I J

5.38 ^ 1.25 5.5 ^ 0.71 5.17 ^ 1.04 5.1 ^ 0.42 5.17 ^ 1.04 5.5 ^ 0.71 4.5 ^ 0.71 4.9 ^ 1.24 5.1 ^ 0.74 5.1 ^ 0.82

First group (natural history)

Second group (double-blind)

Third group (deceptive administration)

Dose

Pain intensity at which buprenorphine was requested (NRS)

Dose

Pain intensity at which buprenorphine was requested (NRS)

Dose

Pain intensity at which buprenorphine was requested (NRS)

2nd 3rd 4th 5th 6th 7th 8th

5.65 ^ 0.67 (10 patients) 4.8 ^ 0.63 (10 patients) 5.55 ^ 1.34 (10 patients) 4.8 ^ 0.95 (10 patients) 4.56 ^ 0.58 (9 patients) 4.5 ^ 0.78 (6 patients) 4.3 ^ 0.56 (2 patients)

2nd 3rd 4th 5th 6th

5.35 ^ 0.55 (10 patients) 5.1 ^ 0.7 (10 patients) 5.28 ^ 1.12 (10 patients) 4.81 ^ 0.96 (8 patients) 4.65 ^ 0.73 (3 patients)

2nd 3rd 4th 5th 6th

5.65 ^ 0.47 (10 patients) 5.4 ^ 0.88 (10 patients) 5.13 ^ 0.95 (7 patients) 4.5 ^ 0.45 (3 patients) 4.25 ^ 0.29 (1 patient)

a

Therefore, the value of each patient is the mean of different doses, whereas the value for each dose is the mean of different patients.

present). Likewise, we found no signi®cant differences when pain intensity was analysed for each buprenorphine dose request (Table 3, below). In addition, there were no differences in evening requests in the three groups. Thus the three groups showed exactly the same pain threshold for buprenorphine request, indicating that the request was triggered by the intensity of pain.

4. Discussion The present study shows that different verbal instructions produced different outcomes which, in turn, led to a signi®cant change of behaviour and a signi®cant reduction of opioid intake. In fact, the reduction of buprenorphine requests in the double-blind group was as large as 20.8% compared to the natural history group, and the reduction in the deceptive administration group was even larger, reaching 33.8%. Despite this dose reduction of buprenorphine, the time course of pain was the same in the three groups over the 3-days period of treatment. This indicates that a strong placebo effect occurred, and that this was stronger in the deceptive administration group compared to the doubleblind group. We decided to measure the placebo effect in this way for ethical reasons. In fact, a procedure of this sort is not unethi-

cal for at least three reasons. First, the analgesic treatment with buprenorphine on request was exactly the same in the three groups of patients, and this therapeutic protocol is common practice in our ward. Second, the time course of pain was the same in all the three groups, so that analgesia was not reduced in any way. Third, we ran double-blind administrations after informed consent was obtained, so that the patients knew that they could receive either a painkiller or a placebo. It is important to stress that the reduction of the requests of buprenorphine was due to a real analgesic placebo effect. In fact, the two placebo treated groups could have requested less doses of buprenorphine because of factors which have nothing to do with pain intensity. We considered this possibility at the beginning of the study by recording the pain intensity every hour in order to see whether a true analgesic effect was present. For example, the placebo-treated patients were certainly more reassured than the natural history group, simply because the basal infusion made them con®dent of a continuous and constant analgesia. However, we ruled out the possibility that differences of this sort were present in the three groups. As shown in Fig. 4A, the analgesic effect was the same in the three groups and the pain thresholds for buprenorphine request were alike as well (Table 3). This indicates that the real trigger for the request of buprenorphine was pain intensity and not other factors, such as

A. Pollo et al. / Pain 93 (2001) 77±84

anxiety. All these data suggest that analgesia was the same in the three groups of patients. The most interesting ®nding of the present study is concerned with the different outcomes obtained with the different experimental protocols. In a demonstration of the importance of response expectancies, Kirsch and Weixel (1988) compared the usual double-blind administration with a deceptive administration. These authors predicted that the deceptive administration would produce stronger effects than the double-blind administration because the latter induces less certain expectations about the outcome. In the present study we con®rm these ®ndings in the clinical pharmacological setting. In fact, our results indicate that the different placebo analgesic responses were due to the different verbal instructions which, in turn, induced different expectations. To be more speci®c, in the double-blind administration the patients were less certain about whether they actually were receiving a painkilling basal infusion, whereas in the deceptive administration the patients strongly believed that they were receiving an analgesic long-lasting basal infusion. These ®ndings are in agreement with those reported by the expectancy theorists, in particular by Kirsch and colleagues (Kirsch, 1985; Kirsch and Weixel, 1988; Wickless and Kirsch, 1989; Kirsch, 1990; Montgomery and Kirsch, 1997). Many other studies con®rm that expectations about the outcome play a very important role in placebo analgesia (Bootzin, 1985; Gracely et al., 1985; Fields and Price, 1997; Price and Fields, 1997; Amanzio and Benedetti, 1999; Benedetti et al., 1999b; Price et al., 1999; Gracely, 2000; Price, 2000). However, it is worth pointing out that the explanatory mechanisms of response expectancies and conditioning are not mutually exclusive (Price, 2000). The modern interpretation of classical conditioning phenomena considers that conditioning itself leads to the acquisition of expectancies that certain events will follow other events (Rescorla, 1988). In addition, it should be stressed that either one mechanism or the other can be involved in different types of placebo effects. For example, some placebo responses can be explained on the basis of a pure conditioning mechanism (e.g. Siegel, 1985; Ader, 1997; Benedetti et al., 1999a). It is worth emphasizing once again that the placebo protocol of the present study has important clinical implications. In fact, the use of the placebo basal infusion, and in particular the strong expectations induced by the deceptive administration, produced an important reduction of opioid intake. Therefore, the understanding of the placebo phenomenon and of its conditioning and expectancy mechanisms can also be harnessed to the patient's advantage.

Acknowledgements This work was supported by grants from MURST and CNR `Coordinate Project on Trigeminal Pain'.

83

References Ader R. The role of conditioning in pharmacotherapy. In: Harrington A, editor. The placebo effect: an interdisciplinary exploration, Cambridge, Mass: Harvard University Press, 1997, pp. 138±165. Amanzio M, Benedetti F. Neuropharmacological dissection of placebo analgesia: expectation-activated opioid systems versus conditioningactivated speci®c subsystems. J Neurosci 1999;19:484±494. Benedetti F, Amanzio M, Baldi S, Casadio C, Maggi G. Inducing placebo respiratory depressant responses in humans via opioid receptors. Eur J Neurosci 1999a;11:625±631. Benedetti F, Arduino C, Amanzio M. Somatotopic activation of opioid systems by target-directed expectations of analgesia. J Neurosci 1999b;19:3639±3648. Bootzin RR. The role of expectancy in behaviour change. In: White L, Tursky B, Schwartz GE, editors. Placebo: theory, research, and mechanisms, New York: Guilford, 1985, pp. 196±210. Evans FJ. The placebo control of pain: a paradigm for investigating nonspeci®c effects in psychotherapy. In: Brady JP, Mendels J, Reiger WR, Orne MT, editors. Psychiatry: areas of promise and advancement, New York: Plenum, 1977, pp. 249±271. Fields HL, Price DD. Toward a neurobiology of placebo analgesia. In: Harrington A, editor. The placebo effect: an interdisciplinary exploration, Cambridge, Mass: Harvard University Press, 1997, pp. 93±116. Gleidman LH, Grantt WH, Teitelbaum HA. Some implications of conditional re¯ex studies for placebo research. Am J Psychiat 1957;113:1103± 1107. Gracely RH. Charisma and the art of healing: can non-speci®c factors be enough? In: Devor M, Rowbotham MC, Wiesenfeld-Hallin Z, editors. Proceedings of the 9th World Congress of Pain, Seattle: IASP Press, 2000, pp. 1045±1067. Gracely RH, Dubner R, Deeter WR, Wolskee PJ. Clinician's expectations in¯uence placebo analgesia. Lancet 1985;1:8419±8423. Herrnstein RJ. Placebo effect in the rat. Science 1962;138:677±678. Kirsch I. Response expectancy as a determinant of experience and behaviour. Am Psychol 1985;40:1189±1202. Kirsch I. Changing expectations: a key to effective psychotherapy. Paci®c Grove, CA: Brooks-Cole, 1990. Kirsch I, Weixel LJ. Double-blind versus deceptive administration of a placebo. Behav Neurosci 1988;102:319±323. McGlashan TH, Evans FJ, Orne MT. The nature of hypnotic analgesia and placebo response to experimental pain. Psychosom Med 1969;31:227± 246. Montgomery GH, Kirsch I. Mechanisms of placebo pain reduction: an empirical investigation. Psychol Sci 1996;7:174±176. Montgomery GH, Kirsch I. Classical conditioning and the placebo effect. Pain 1997;72:107±113. Price DD. Factors that determine the magnitude and presence of placebo analgesia. In: Devor M, Rowbotham MC, Wiesenfeld-Hallin Z, editors. Proceedings of the 9th World Congress of Pain, Seattle: IASP Press, 2000, pp. 1085±1095. Price DD, Fields HL. The contribution of desire and expectation to placebo analgesia: implications for new research strategies. In: Harrington A, editor. The placebo effect: an interdisciplinary exploration, Cambridge, Mass: Harvard University Press, 1997, pp. 117±137. Price DD, Milling LS, Kirsch I, Duff A, Montgomery GH, Nicholls SS. An analysis of factors that contribute to the magnitude of placebo analgesia in an experimental paradigm. Pain 1999;83:147±156. Rescorla RA. Pavlovian conditioning: it is not what you think it is. Am Psychol 1988;43:151±160. Siegel S. Drug-anticipatory responses in animals. In: White L, Tursky B, Schwartz GE, editors. Placebo: theory, research, and mechanisms, New York: Guilford, 1985, pp. 288±305. Voudouris NJ, Peck CL, Coleman G. Conditioned response models of placebo phenomena: further support. Pain 1989;38:109±116.

84

A. Pollo et al. / Pain 93 (2001) 77±84

Voudouris NJ, Peck CL, Coleman G. The role of conditioning and verbal expectancy in the placebo response. Pain 1990;43:121±128. Wall PD. The placebo effect: an unpopular topic. Pain 1992;51:1±3. Wall PD. Pain and the placebo response. Ciba Foundation Symposium 174. Experimental and theoretical studies of consciousness, New York: Wiley, 1993, pp. 187±216.

Wickless C, Kirsch I. The effects of verbal and experiential expectancy manipulations on hypnotic susceptibility. J Pers Soc Psychol 1989;57:762±768. Wickramasekera I. A conditioned response model of the placebo effect: predictions from the model. Biofeedback Self-Reg 1980;5:5±18.