ORIGINAL ARTICLE
Direct Comparison of Placebo Effects on Clinical and Experimental Pain Julie Charron, PhD,* Pierre Rainville, PhD,† and Serge Marchand, PhD‡§
Objectives: Placebo effects have been suggested to be more potent on clinical than experimental pain. However, this proposition is based on the comparison of the magnitude of placebo analgesia between studies using different methodologies or between different groups of subjects within the same study. The authors sought to provide a more direct test of this hypothesis using a within-subject design and to investigate the potential mediating effect of expectancy.
Method: Sixteen patients with low back pain rated the intensity and the unpleasantness of their clinical pain and underwent two cold pressor tests, both before and after a saline injection presented either as a potent painkiller (placebo treatment) in one session or as an inactive substance in a control session. Results: The placebo treatment produced comparable increases in expected relief for clinical and experimental pain. However, ratings of pain intensity, pain unpleasantness, and perceived relief confirmed the larger placebo effect in low back pain than cold pressor pain. Retrospective ratings of perceived relief in low back pain generally showed the largest placebo effect compared with concurrent pain ratings. Furthermore, when the placebo session was performed after the control session, the placebo effect on low back pain was substantially reduced and observed only in perceived relief. Variations in expectation could not account for the large difference in placebo analgesia between clinical and experimental pain. Conclusions: The important reduction in placebo analgesia in low back pain after the single pre-exposure to the ineffective control treatment suggests the additional involvement of highly flexible mechanisms that may counteract the pro-analgesic effects of expectations. Key Words: placebo, cold pressor pain, low back pain, expectation, conditioning (Clin J Pain 2006;22:204–211)
Received for publication September 12, 2004; revised January 29, 2005; accepted February 11, 2005. From the *Department of Psychology, University of Montre´al, Montre´al, Que´bec, Canada; †Faculte´ de Me´decine Dentaire, Universite´ of Montre´al, Montre´ al, Que´ bec, Canada; ‡Faculty of Medicine, University of Sherbrooke, Sherbrooke, Que´bec, Canada; and §Department of Sciences and Health, University of Que´bec en Abitibi-Te´miscamingue, RouynNoranda, Que´bec, Canada. Supported by the Fonds de la recherche en sante´ du Que´bec, the Canadian Foundation for Innovation, and the CIHR. Reprints: Serge Marchand, PhD, Universite´ de Sherbrooke Faculte´ de me´decine, Neurochirurgie, 3001, 12e Avenue nord, Sherbrooke, Que´bec J1H 5N4, Canada (e-mail:
[email protected]). Copyright Ó 2006 by Lippincott Williams & Wilkins
204
P
lacebo-induced change in pain is a fascinating phenomenon that underlines the impact of cognitive states on physical health. In recent years, interest in this phenomenon has grown from a clinical concern to a methodologic and research subject that has important clinical implications.1,2 Indeed, a better understanding of the mechanisms of placebo analgesia brings promise to applications in all analgesia settings, with both active and placebo treatments. Experimental pain studies contribute to the understanding of basic processes involved in pain and analgesia as they provide well-controlled stimulation conditions and allow the assessment of treatment-related effects in healthy individuals. These studies have largely contributed to the assessment of the relative importance of conditioning, expectations, and the endogenous opioid system in placebo analgesia.3 However, one concern regarding experimental studies is the generalizability of findings to clinical settings. This problem can be addressed by a direct comparison of clinical and experimental studies to establish their similarities and differences and identify the factors that may contribute to those differences. In contrast, as Price emphasized,4 few studies of clinical analgesia directly measure the placebo effect. They instead use the difference between the placebo condition and the active condition to assess the efficacy of the active treatment of interest. Recent studies on the mechanisms underlying placebo analgesia used either clinical1,5 or experimental pain paradigms,5–11 but no study has directly compared the two types of pain within individuals. It has been previously suggested that the magnitude of placebo analgesic effects is larger in clinical pain compared with experimental pain studies.12 Forms of experimental pain that are more stressful are associated with a greater placebo effect,13 and this may explain the hypothesized stronger placebo effect in clinical than experimental pain. However, in a recent meta-analysis, Vase et al14 reported that the effect sizes observed in clinical and experimental studies examining placebo mechanisms were comparable (mean effect size, clinical = 0.87 and experimental = 0.93). A meta-analysis of 187 studies15 did find a larger placebo response for acute pain as compared with chronic, experimental, and postoperative pain, but the size of the placebo effect was similar for experimental and chronic pain. It is difficult to interpret the comparison between clinical and experimental studies, considering that the analysis was based on a comparison of studies performed on different subjects and in different contexts. Furthermore, the large range in effect size for both clinical (20.64 to 2.29) and experimental studies (0.44–2.10) reported by Vase et al14 attests to the great variability in the magnitude of placebo effects observed between studies. Clin J Pain Volume 22, Number 2, February 2006
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Clin J Pain Volume 22, Number 2, February 2006
Placebo Effects on Clinical and Experimental Pain
One possible explanation for the large differences between studies relies on the expectations of relief held by patients and healthy volunteers who undergo a placebo procedure. Expectation of relief has been shown to mediate placebo analgesia, at least in part, in both clinical and experimental contexts.1,6,10,16,17 This mediating variable may contribute to the hypothesized difference between placebo effects on clinical and experimental pain to the extent that the expectations associated with any analgesic treatment may be linked to specific outcomes (eg, reduction of specific types of pain). Although this hypothesis may run against the old notion that placebo effects are nonspecific and should equally affect clinical and experimental pain, there is unequivocal evidence that very specific expectations may be induced and lead to restricted placebo effects (eg, analgesia restricted to a specific body area7). This implies that the mechanisms underlying expectationmediated placebo effects are highly flexible and could differentially affect different types of pain. To our knowledge, no study has directly compared clinical and experimental pain for the effect of placebo analgesia within the same subjects. The present study had two main goals: to compare the effect of placebo analgesia for clinical and experimental pain in the same subjects and to investigate the effect of expectation of relief as a potential explanatory factor for the hypothesized difference between clinical and experimental pain in placebo analgesia. To do so, the effect of placebo analgesia on clinical and experimental pain was measured in patients with low back pain. Pain intensity, pain unpleasantness, and perceived relief were measured separately for clinical pain (low back pain) and experimental pain (cold pressor test) in the same subjects and during the same sessions. To verify the mediating effect of expectancy, we also measured the expected outcome of the placebo treatment separately for clinical and experimental pain. A preliminary report of this study has been presented in abstract form.18
METHOD Participants Following Ethics Committee approval and informed consent, 16 patients with chronic low back pain (10 men, 6 women) completed the study. One additional subject dropped out of the study after the first session and is not included in the analyses. The participants were 18 to 60 years of age (mean = 39.8, SD = 13.2), and the persistent pain duration was 10 months to 25 years (average = 8.4 years, SD = 6.9). Patients volunteered to come to the laboratory for two 3-hour sessions performed on two different days. They were
told that they would receive an analgesic in one session and a control saline solution in the other. They also knew that the order was randomized. Exclusion criteria included pregnancy, the use of opioid, analgesic, antidepressant, or antiepileptic drugs, and the presence of cardiovascular or neurologic disease.
Low Back Pain In each of the two sessions, patients rated the intensity and unpleasantness of their current low back pain every 2 minutes for 20 minutes. Subjects were free to move during the session and to change from sitting to standing positions between tests. The evaluation was done with 0-to-100 numeric scales,19,20 and the data were collected by an experimenter unaware of the treatment condition (placebo versus control instructions). We stressed the difference between intensity and unpleasantness using the instructions from Price et al.21
Experimental Pain In each of the two sessions, circulating cold water was used to produce cold pressor pain. At the beginning of the first session, subjects immersed their right hand in the water bath for a 1-minute pretest. Water temperature was adjusted individually to produce moderate pain (10–13°C). Then, two experimental tests of 2 minutes each were performed at the selected temperature, both before and after the treatment. Subjects rated the intensity and unpleasantness of cold pain every 15 seconds using the same scales used during the low back pain measures. The two successive tests were separated by a 5-minute pause.22
Experimental Design and Procedure A fully factorial, within-subject design was applied to test for the effects of the placebo treatment condition (placebo versus control) and the different types of pain (clinical versus experimental). The control and placebo treatments were administered to the same subjects in separate sessions performed on separate days. This design was used to control for the natural history of pain during the testing session and to allow for a more powerful within-subject comparison, with each subject being his or her own control, to evaluate the effect of the treatment of both clinical and experimental pain. In each testing session, subjects first rated their low back pain and then underwent the first pair of cold pressor tests (Table 1). The placebo or the control treatment was administered next. Both treatments consisted of one intravenous injections of 1 mL saline and differed only in the instructions given to the patients immediately before the injection. One of two trained nurses pseudo-randomized the order of the control
TABLE 1. Study Design Baseline Pain Low back pain
Cold pressor test (2 tests)
20 min
10 min
Treatment (saline injection: suggestions for analgesia [placebo] or no analgesia [control]) 10 min
q 2006 Lippincott Williams & Wilkins
Expected Relief for LBP and CPT
1 min
Post-Treatment Pain Cold pressor test (2 tests)
Low back pain
10 min
20 min
Perceived Relief for LBP and CPT
1 min
205
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Charron et al
and placebo sessions, performed the injection, and gave the placebo or control instructions in both testing sessions. The experimenter who administered the cold pressor test and collected the subjects’ ratings was out of the testing room during that time and was not informed of the treatment condition. Cold pressor tests were repeated after the treatment, followed by another series of low back pain ratings.
Placebo and Control Instructions On one test day, subjects were told they received a potent analgesic (placebo condition), and on the other day, they were told they received a saline injection as a control (control condition). The nurses’ instructions to the subjects were standardized and kept as similar as possible between conditions, except for the analgesic properties of the solution suggested only in the placebo condition. During the placebo condition, the instructions were as follows (translated from French):
n We will now administer the analgesic solution. It is a substance known to give rapid and effective relief for many types of pain. We will wait 10 minutes before we take other measures. We ask you to stay seated during this time. We will inject 1 mL of solution in the vein of your arm. You might feel some prickling or a temporary burning sensation at the injection site. The dose we used very rarely causes side effects like drowsiness or cutaneous reactions, but we ask you to warn us of any discomfort you might feel. During the control condition, the instructions were as follows (translated from French):
n We will now administer the control solution. It is a substance that should have little impact on your pain. We will wait 10 minutes before we take other measures. We ask you to stay seated during this time. We will inject 1 mL of salt-water solution in the vein of your arm. You might feel some prickling or a temporary burning sensation at the injection site. This substance very rarely causes side effects like drowsiness or cutaneous reactions, but we ask you to warn us of any discomfort you might feel.
Expected and Perceived Relief Immediately after the injection, subjects were asked to rate their expectations of change for the post-treatment evaluation. They were specifically asked ‘‘Do you think the treatment we administered will change your pain?’’ and they had to give a numeric rating for the expected level of change for both types of pain (low back pain and cold pressor test pain). The scale had a –100 anchor point (maximal increase of my pain) on the left side, a 0 (no change) in the middle, and a +100 (total relief) at the right extremity. Finally, after the end of the tests, subjects were asked to report their perceived changes in pain after the treatment. They were asked ‘‘Do you think the treatment we administered changed your pain during the last test?’’ and had to give a rating for the perceived change for both clinical and experimental pain. The same scale was used for ratings of expectation and perceived changes.
206
Clin J Pain Volume 22, Number 2, February 2006
Statistical Analysis Pain ratings were averaged within each condition for each subject and transformed into a difference pain score by calculating the decrease in pain following the treatment (mean pretreatment rating minus mean post-treatment rating). Differences between the placebo treatment condition and the control condition were further calculated for ratings of expectation, perceived relief, and difference pain score to obtain global indices of expected, observed, and retrospectively perceived placebo analgesia. A 2 3 2 repeated-measures ANOVA (treatment [placebo vs. control] 3 type of pain [clinical vs. experimental]) was first performed using SPSS 9.0.0 for Windows. An alpha of 0.05 was chosen. The session in which the placebo was administered (placebo in the first vs. second session) was included as a third variable to examine the possible moderating effect of this factor. Then, additional 2 3 2 ANOVAs were performed separately for the two sessions when the three-way interaction was significant. Data were further explored using bivariate correlations and multiple regression analyses performed between the dependent variables.
RESULTS Expectation of Relief The placebo procedure was effective in inducing considerable expectations of relief (Fig. 1A). Ratings of expected relief increased significantly in the placebo treatment compared with the control condition for both low back pain (mean 33.75%) and cold pressor pain (mean 30.94%; main effect of treatment: F = 12.980, P = 0.003). However, when the session order entered the analysis, a three-way interaction was significant (treatment 3 pain 3 order; F = 7.194, P = 0.018). When the placebo was given in the first session (see Fig. 1B), only the main effect of treatment approached significance (F = 4.528, P = 0.066; main effect of type of pain: F = 0.190, P = 0.674; interaction F = 1.694, P = 0.229). In contrast, when the placebo was given in the second session (see Fig. 1C), the main effect of the treatment was highly significant (F = 12.633, P = 0.012), and this effect was larger with the low back pain than with the cold pressor pain (interaction F = 6.729, P = 0.041; main effect of type of pain: F = 2.588, P = 0.159). This indicated that patients expected relief of both clinical and experimental pain following the placebo procedure, and that patients who received the placebo instructions in the second session expected slightly more relief of low back pain than cold pain. Further examination of the ratings of expectations indicated that all but four subjects expected more pain relief in the placebo than the control condition. These four subjects expected no pain relief in any condition (n = 2), equal pain relief in both the placebo and control conditions (n = 1), or some increase in clinical pain in the placebo condition and no change in the other conditions (n = 1). Excluding these subjects from the analyses of expectation logically led to more robust statistical results. The main effect of treatment reached significance when the placebo session was performed first (F = 10.09, P = 0.025) or second (F = 18.92, P = 0.007), and the interaction between treatment and type of pain was again q 2006 Lippincott Williams & Wilkins
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Clin J Pain Volume 22, Number 2, February 2006
Placebo Effects on Clinical and Experimental Pain
(pretreatment vs. post-treatment; main effects and interactions, P . 0.05 for all). Further analyses were therefore conducted on treatment-related changes in pain intensity and unpleasantness.
Placebo Effects on Pain
FIGURE 1. Expectation of relief in low back pain and cold pressor pain immediately after the injection of saline presented either as a potent analgesic (Placebo) or as a control inert substance (Control). Results are shown for the whole sample (A) and for subjects who received the analgesic suggestions before the first (B) or second session (C).
significant only when the placebo session was performed second (F = 8.05, P = 0.036). Since an increase in expectation was posited a priori as a mediating factor for the placebo effect, we excluded those four subjects from the analysis of pain intensity, unpleasantness, and perceived changes in pain. However, these subjects entered the correlation and regression analyses examining the relations between the dependent variables.
Pain Intensity and Unpleasantness Baseline Pain and Effects of the Control Treatment Pretreatment ratings of cold pressor pain intensity and unpleasantness were higher than those of low back pain (intensity: F = 15.222, P = 0.003; unpleasantness: F = 11.423, P = 0.007), but no generalized difference was observed between the placebo and control condition for either type of pain (main effects of treatment and interactions between pain and treatment, P . 0.20 for all). The control treatment did not produce significant changes in pain intensity or unpleasantness of cold pressor pain or low back pain in any condition q 2006 Lippincott Williams & Wilkins
The treatment-related effects on pain ratings are illustrated in Figure 2. Compared with the control condition, the placebo treatment produced significant decreases in pain unpleasantness (main effect of treatment: F = 5.740, P = 0.038) and decreases in pain intensity approaching significance (F = 3.707, P = 0.083). However, this effect interacted with the type of pain and the session order for both pain intensity (F = 6.790, P = 0.026) and unpleasantness (F = 11.586, P = 0.007) (see Fig. 2A, B). When the placebo was administered in the first session (see Fig. 2C, D), there was a larger placebo effect for low back pain than experimental pain (interaction treatment 3 pain) for both pain intensity (F = 9.38, P = 0.028) and unpleasantness (F = 12.806, P = 0.016). The main effect of treatment (placebo . control) also reached significance on unpleasantness (F = 6.946, P = 0.046) but not intensity (F = 3.033, P = 0.142). In contrast, no main effect or interaction approached significance when the placebo instructions were given in the second session (P . 0.10 for all; see Fig. 2E, F). These results confirmed that the placebo procedure produced a larger decrease in low back pain than experimental pain. However, this effect reached significance only when the placebo instructions were given in the first session. The placebo effect observed in session 1 was directly contrasted to changes in pain observed in the control condition in session 1 using a between-group t tests (subgroup placebo first vs. subgroup control first). This approach is less powerful (because of a small n in each subgroup) but is not affected by potential carryover effects.23 In these more conservative analyses, the placebo effect was highly significant on perceived relief of low back pain (P = 0.004) and significant on low back pain intensity (P = 0.027) and unpleasantness (P = 0.033). Between-group effects examining changes in pain and perceived relief did not reach significance for experimental pain (P . 0.2 for all). A similar comparison between subgroups was performed on data acquired in session 2 (subgroup placebo second vs. subgroup control second). In this case, the placebo effect approached significance only on perceived relief of low back pain (P = 0.053).
Placebo Effects on Perceived Relief Post-treatment ratings of the perceived changes in pain (Fig. 3) were generally consistent with changes in concurrent ratings of pain intensity and unpleasantness. There was a highly significant main effect of treatment (placebo . control; F = 20.661, P , 0.001), a main effect of pain (F = 6.273, P = 0.031), and an interaction between treatment and pain (F = 14.112, P = 0.004), confirming the superior placebo effect on low back pain (see Fig. 3A). The treatment 3 pain 3 order interaction also approached significance (F = 3.872, P = 0.077), so we examined the data again according to the order of sessions. Pain relief increased with placebo treatment, but more so when the placebo was given in the first session (see Fig. 3B; main effect of treatment: F = 18.462, P = 0.008) as opposed to
207
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Charron et al
Clin J Pain Volume 22, Number 2, February 2006
FIGURE 2. Treatment-induced decreases in pain sensation intensity and unpleasantness ratings of low back pain and cold pressor pain as a result of the injection of saline presented either as a potent analgesic (Placebo) or as a control inert substance (Control). Results are shown for the whole sample (A, B) and for subjects who received the analgesic suggestions before the first (C, D) or second session (E, F).
the second session (see Fig. 3C; F = 3.858, P = 0.107). Similarly, the superior placebo effect observed for low back pain (interaction treatment 3 pain) was confirmed only when the placebo was administered in the first session (placebo first: F = 18.618, P = 0.008; placebo second: F = 1.429, P = 0.286). Examination of the means indicated that the perceived relief in low back pain increased considerably in the placebo condition mainly when the placebo procedure was administered first (see Fig. 3B). In contrast, ratings of perceived relief in cold pressor pain indicated a modest but significantly stronger relief of 7.5 in the placebo condition compared with 4.2 in the control condition (F = 5.50, onetailed P = 0.019). These effects confirmed the stronger placebo effect in low back pain than cold pressor pain and the stronger placebo effect on low back pain when the placebo treatment was administered first.
Standardized Estimates of the Placebo Effect The overall placebo effects were larger for low back pain (effect size range 1.13–2.15) than cold pressor pain (0.29–0.71), consistent with the absolute changes in ratings (see Figs. 2A, B and 3A). Retrospective evaluations of the perceived relief in low back pain (effect size = 2.15) showed larger effect sizes than ratings of pain intensity (1.20) and unpleasantness (1.13). Placebo effects were very large in low back pain when the placebo condition was administered first
208
(effect size 2.23–3.28), while only ratings of perceived relief showed a placebo effect when the placebo was administered second (effect size = 1.22).
Relation Between Expectation, Changes in Pain, and Perceived Relief in Low Back Pain The placebo effect observed in low back pain was further examined using bivariate (Pearson r) and partial correlation analyses to document the relation between the dependent variables and to consider the possible mechanisms involved. The expectation of relief predicted the changes in pain intensity and perceived relief induced by the placebo most consistently when the placebo was administered in the first session (intensity: placebo first, r = 0.70, P , 0.05; placebo second, r = 0.36, P = NS; perceived relief: placebo first, r = 0.69, P , 0.05; placebo second, r = 0.68, P , 0.05). However, the expectation of relief did not significantly predict changes in pain unpleasantness induced by the placebo (P = NS for all). Changes in unpleasantness were largely explained by changes in pain intensity in all conditions (r = 0.72–0.93, P , 0.05 for all), independent of expectation (partial r = 0.63–0.96, P , 0.05 for all). In turn, placebo-induced changes in perceived relief showed the strongest correlation with changes in pain unpleasantness (placebo first, r = 0.79, P , 0.01). Partial correlation analyses indicated that both changes in expected relief and in pain unpleasantness contributed to changes in q 2006 Lippincott Williams & Wilkins
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Clin J Pain Volume 22, Number 2, February 2006
Placebo Effects on Clinical and Experimental Pain
No model significantly predicted the differences in perceived changes in cold pressor pain between the placebo and control session (P . 0.2 for all).
DISCUSSION Efficacy of the Placebo Induction Previous authors have clearly shown the critical role of expectation in placebo analgesia.3,24 Here, patients reported relatively high expectations of pain relief in the placebo condition, confirming that our experimental procedure was adequate to generate expectation-induced placebo analgesia. The placebo effect was largely confirmed by the significant decreases in pain intensity and unpleasantness and by the increase in perceived relief observed in the placebo condition compared with the control condition. Standardized estimates confirmed the moderate to large placebo effects consistent with those reported previously.14
Clinical Pain Is More Sensitive Than Experimental Pain to Placebo Analgesia
FIGURE 3. Perceived relief in low back pain and cold pressor pain reported at the end of each session in which the injection of saline was presented either as a potent analgesic (Placebo) or as a control inert substance (Control). Results are shown for the whole sample (A) and for subjects who received the analgesic suggestions before the first (B) or second session (C).
perceived relief, independently of the other variables (P , 0.05 for all).
Predictors of the Placebo Response in Perceived Relief Linear regressions were performed on the increase in perceived relief of low back pain induced by the placebo as the dependent variable (perceived relief after the placebo minus perceived relief after the control). Placebo-induced changes in expectation of relief and in pain intensity and unpleasantness were entered as independent variables. Using a stepwise approach, only the changes in unpleasantness contributed significantly to predict the perceived relief induced by the placebo (R2 = 0.576), while expectation was marginally significant, and changes in intensity did not contribute to the model. Forcing all variables into the model slightly improved the fit of the model (R2 = 0.658) and revealed a significant contribution of expectation. These models were confirmed when the placebo was administered in the first but not the second session. q 2006 Lippincott Williams & Wilkins
As previously suggested,12,13 clinical pain was more sensitive to the placebo treatment than experimental pain. However, we did not confirm the hypothesis that differences in expectations account for this difference in placebo effects in low back pain and cold pain. The procedure used here to produce placebo effects was very similar to those used in other experimental studies,6,8,17 and the expectations of relief induced for clinical and experimental pain were robust and very similar. For these reasons, the smaller placebo effect observed in experimental pain is unlikely to be caused by specific aspects of the placebo instructions and procedure used here. This difference in placebo effect between clinical and experimental pain may be explained by pain-related factors or subject-related factors. First, the nature of clinical and experimental pain and their partly distinct underlying mechanisms may contribute to the differential effect observed here. Persistent clinical pain is often felt to be a meaningful threat that interferes with many aspects of one’s life. Furthermore, the causal agent of persistent low back pain is internal and often undetermined. In contrast, experimental pain has a definite and short duration, subjects are typically reassured about the safety of the procedure used, and the causal agent (the nociceptive stimulus) is external and clearly identified. Each of these factors may contribute to render clinical pain more susceptible to the effects of expectation of analgesia. However, this explanation is incomplete in view of the studies showing robust placebo analgesic effects on experimental pain in healthy individuals14 and in patients.25 In this latter study, irritable bowel syndrome patients showed significant analgesic effects of a rectal placebo on pain produced by both rectal distention and cutaneous heat. However, in this specific case, the placebo effect may have affected a physiologic mechanism underlying both the visceral and cutaneous hyperalgesia characterizing this population. If the placebo effect acted on a common underlying mechanism, one would expect placebo-related effects to generalize to perceptual abnormalities (here cutaneous hyperalgesia) related to the underlying physiopathology. However, in the present study, the placebo effect acted on
209
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Charron et al
back pain-related mechanisms independent of the processes involved in cutaneous pain perception. Experimental pain perception was probably subserved by physiologic processes at least partly segregated from clinical pain process, and only the latter were affected by the placebo treatment. The second group of factors that may contribute to explain this difference between the susceptibility of clinical and experimental pain to expectation-induced placebo analgesia relates to the characteristics of the subjects. Since clinical pain studies are, by necessity, conducted in patients, and experimental studies are generally conducted in healthy volunteers, a constellation of nonspecific individual differences may contribute to the differences between placebo analgesia observed in experimental and clinical studies (eg, emotional state, personality variables). However, the present results do restrict the range of potential explanatory variables. The differential placebo effect observed here using a within-subject design allowed us to control for the involvement of a nonspecific factor, because such a factor would have affected both types of pain. The explanation may reside in the interaction between subject-related factors and the specific context of a study on pain and analgesia. It is also possible that the uncontrolled heterogeneity of the pathophysiology in our patient population may have precluded us from showing a generalization of the placebo effect to experimental pain. This generalization may occur in some low back pain patients with hypersensitivity that generalizes to cutaneous pain. A rating of medical use of analgesics was available for only six subjects, but the correlation with back pain relief was nonsignificant (r = 0.508, P . 0.3). Moreover, an analysis of the correlation between the change of back pain after the placebo and pain duration or age for all subjects was nonsignificant (r = 20.351, P . 0.2 for pain duration, r = 20.351, P . 0.1 for age). Patients may respond differently to comparable levels of expectation of analgesia for clinical and experimental pain. Price et al26 showed that individual differences in the specific goal adopted by participants in a pain study (eg, avoid pain altogether or experience less pain) interact with expectations to modulate pain perception. This effect was shown most consistently on pain affect, the dimension most readily modulated by the placebo in the present study. Motivation or the desire of relief was also shown to contribute to placebo responses in a study investigating sedative or stimulant effects.27 In a study of placebo analgesia in healthy volunteers, however, the desire of relief was not associated with changes in pain.6 In that study, the magnitude of the changes induced in desire of relief was fairly small, and the authors suggested that the desire of relief might be of importance mostly for clinical pain. In a more recent study, some evidence for the contribution of desire for relief to placebo analgesia was reported in irritable bowel syndrome patients.25 We did not measure the desire for relief in the present study, but it appears plausible that the patients experienced more significant goals related to their clinical condition and variable levels of desire for relief for their experimental pain. Expectations might have affected most effectively the experience that was most significant for the patients. As suggested previously by Price,28 the interaction between the expectations of relief and goal-directed desire for relief may be critical to produce robust placebo analgesia.
210
Clin J Pain Volume 22, Number 2, February 2006
Placebo Effects in Retrospective Ratings of Relief Changes in the retrospective evaluation of relief were more sensitive to placebo effects than changes in concurrent ratings of pain intensity and unpleasantness. This effect is consistent with previous studies investigating placebo analgesia6 and pain memory.29 Here, perceived relief was predicted by changes in pain unpleasantness and secondarily by expectations, even after the variance associated with pain intensity was accounted for. This result strongly emphasizes the primary impact of affective processes on retrospective evaluation of relief based on pain memory. This pivotal role of affect on the retrospective evaluation of pain and pain relief likely goes beyond the immediate effect of the pain unpleasantness experienced during a treatment and may encompass the effects of the patient’s global affective state before and during a treatment.30 This role of affect in pain memory further underscores the importance of assessing both pain sensation and pain affect. The distorting effect of memory processes on pain-related information raises some concern about studies relying solely on retrospective ratings of pain to assess analgesic treatment or placebo effect, as they may not reflect specifically changes in the experience of pain.6,31 On the other hand, the memory of pain and relief may better reflect how the patient generally feels about the treatment, and this may affect subjective well-being, compliance to the treatment, and the development of beliefs and expectations about future treatment. These effects may contribute significantly to predict future outcomes. These effects may also explain the persistent differences found between actual pain reduction and estimated pain relief in some clinical conditions, even when the analgesic procedure has been used for years.32
Does Pre-Exposure to an Ineffective Treatment Block Placebo Analgesia? One last finding of this study requires attention. The robust placebo effects observed here in low back pain were much stronger and exclusively observed when the placebo session was performed before the control session. Taken from a different angle, placebo effects in low back pain were largely blocked when the placebo session followed the control session. This unexpected but large difference should be considered with caution as it is based on the comparison of two small subgroups of subjects. However, this effect may reflect another puzzling dissociation between expectation and the placebo effect. Indeed, expectations of relief in low back pain were stronger when the placebo suggestions were given in the second session, yet placebo analgesia was smaller and observed only in ratings of relief in that condition. One possible explanation for this finding may be the carryover effect of the first session on the second session in the group performing the control condition first.23 Patients in that subgroup did not expect or experience analgesia in that first session. The experience of ‘‘no changes’’ in pain in the context of the first session might have triggered some anti-analgesic mechanisms that partially blocked the placebo effect in the second session. Recent models have emphasized the relative role of expectation and conditioning in placebo analgesia. Some studies q 2006 Lippincott Williams & Wilkins
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Clin J Pain Volume 22, Number 2, February 2006
have shown that anti-analgesic expectations can counteract conditioned analgesia.17,32 In contrast, the results of the present study suggest that conditioned anti-analgesia, produced by the control session performed first, may have blocked the proanalgesic effect of relief expectancy.
Limitations of the Study
Placebo Effects on Clinical and Experimental Pain
4. 5. 6.
Several issues raised by this discussion will need to be confirmed in future studies. First, we mentioned the possibility that the desire for relief may contribute to the difference between clinical and experimental pain in the placebo effects. However, subjective ratings of this factor were not included in this study. Furthermore, we did not control for the heterogeneity of patients, and we did not have the statistical power to examine the potential effects of individual differences in treatment history. However, the chronic pain state of these patients attests to a relative failure of treatment, so one would expect that they might have become less responsive to placebo treatment rather than more responsive.
11.
CONCLUSIONS
12. 13.
Our results confirmed that placebo effects are stronger for clinical than experimental pain. These results also show two cases of dissociation between high levels of expectations and placebo effects. First, the mediating role of expectancy was found in the effective placebo condition on low back pain, whereas high expectations did not lead to strong placebo effects on experimental pain. Second, high expectations of relief in low back pain produced no placebo effect on concurrent ratings and a reduced placebo effect on perceived relief when the placebo session was performed after the control session. We emphasize the need for a more comprehensive examination of motivational factors that may facilitate or obstruct expectation-induced placebo effects. Future studies should also systematically include measurements of pain sensation and pain affect, as these dimensions of pain may be differentially affected by placebo procedures and may be selectively related to different aspects of placebo mediators. Finally, we raise the unexpected possibility that the anti-analgesic effect of a single experience with an ineffective treatment may critically influence future responses to a treatment, despite pro-analgesic expectations associated with that treatment. This underscores the importance of previous experience on pain relief and attests to the remarkable flexibility of pro- and anti-analgesic processes affecting the magnitude of placebo effects. ACKNOWLEDGMENTS The authors thank Paule Julien, Serge Daigle, Line Fecteau, Guylaine Leblond, and Anie Paiement-Lamothe for their valuable support and Philippe Goffaux for his valuable revision of the manuscript. REFERENCES 1. Pollo A, Amanzio M, Arslanian A, et al. Response expectancies in placebo analgesia and their clinical relevance. Pain. 2001;93:77–84. 2. Price DD. Assessing placebo effects without placebo groups: an untapped possibility? Pain. 2001;90:201–203. 3. Price DD, Soerensen LV. Endogenous opioid and non-opioid pathways as mediators of placebo analgesia. In: Guess HA, Kleinman A, Kusek DJW,
q 2006 Lippincott Williams & Wilkins
7. 8. 9. 10.
14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32.
et al, eds. The Science of Placebo: Toward an Interdisciplinary Research Agenda. London: BMJ Books, 2002:183–206. Price DD. Factors that determine the magnitude and presence of placebo analgesia. In: Devor M, Rowbotham MC, Wiesenfeld-Hallin Z, eds. Proceedings of the IXth World Congress on Pain. Seattle, 2000:1085–1095. Amanzio M, Pollo A, Maggi G, et al. Response variability to analgesics: a role for non-specific activation of endogenous opioids. Pain. 2001;90:205–215. Price DD, Milling LS, Kirsch I, et al. An analysis of factors that contribute to the magnitude of placebo analgesia in an experimental paradigm. Pain. 1999;83:147–156. Benedetti F, Arduino C, Amanzio M. Somatotopic activation of opioid systems by target-directed expectations of analgesia. J Neurosci. 1999;19: 3639–3648. Amanzio M, Benedetti F. Neuropharmacological dissection of placebo analgesia: expectation-activated opioid systems versus conditioningactivated specific subsystems. J Neurosci. 1999;19:484–494. Roelofs J, Ter Riet G, Peters ML, et al. Expectations of analgesia do not affect spinal nociceptive R-III reflex activity: an experimental study into the mechanism of placebo-induced analgesia. Pain. 2000;89:75–80. De Pascalis V, Chiaradia C, Carotenuto E. The contribution of suggestibility and expectation to placebo analgesia phenomenon in an experimental setting. Pain. 2002;96:393–402. Petrovic P, Kalso E, Petersson KM, et al. Placebo and opioid analgesia: imaging a shared neuronal network. Science. 2002;295:1737–1740. Beecher HK. The powerful placebo. JAMA. 1955;159:1602–1606. Jospe M. The Placebo Effect in Healing. Lexington, MA: Lexington Books, 1978. Vase L, Riley JL III, Price DD. A comparison of placebo effects in clinical analgesic trials versus studies of placebo analgesia. Pain. 2002;99:443–452. Harkness E, Ernst E. The enigmatic placebo effect: a systemic review to define its determinants. Perfusion. 2000;13:164–170. Kirsch I. Response expectancy as a determinant of experience and behavior. Am Psychol. 1985;40:1189–1202. Montgomery GH, Kirsch I. Classical conditioning and the placebo effect. Pain. 1997;72:107–113. Charron J, Rainville P, Marchand S. Placebo Analgesia: Clinical Pain Is More Sensitive than Experimental Pain. San Diego, August Edition, 2002. Rainville P, Feine JS, Bushnell MC, et al. A psychophysical comparison of sensory and affective responses to 4 modalities of experimental pain. Somatosens Mot Res. 1992;9:265–277. Marchand S, Charest J, Li J, et al. Is TENS purely a placebo effect? A controlled study on low back pain. Pain. 1993;54:99–106. Price DD, McGrath P, Rafii A, et al. The validation of visual analogue scales as ratio scale measures for chronic and experimental pain. Pain. 1983;17:45–56. Marchand S, Arsenault P. Spatial summation for pain perception: interaction of inhibitory and excitatory mechanisms. Pain. 2002;95:201–206. Suchman AL, Ader R. Classic conditioning and placebo effects in crossover studies. Clin Pharmacol Ther. 1992;52:372–377. Kirsch I. Specifying nonspecific psychological mechanisms of placebo effects. In: Harrington A, ed. The Placebo Effect: An Interdisciplinary Exploration. Cambridge, MA: Harvard University Press, 1997:166–186. Vase L, Robinson ME, Verne GN, et al. The contributions of suggestion, desire, and expectation to placebo effects in irritable bowel syndrome patients. An empirical investigation. Pain. 2003;105:17–25. Price DD, Barrell JJ, Gracely RH. A psychophysical analysis of experimental factors that selectively influence the affective dimension of pain. Pain. 1980;8:137–149. Jensen MP, Karoly P. Motivation and expectancy factors in symptom perception: a laboratory study of the placebo effect. Psychosom Med. 1991;53:144–152. Price DD. Placebo analgesia. In: Price DD, ed. Psychological Mechanisms of Pain and Analgesia. New York: Raven Press, 1999:155–181. Feine JS, Lavigne G, Dao TTT, et al. Memories of chronic pain and perceptions of relief. Pain. 1998;77:137–141. Gedney JJ, Logan H, Baron RS. Predictors of short-term and long-term memory of sensory and affective dimensions of pain. J Pain. 2003;4:47–55. Rainville P, Doucet JC, Fortin MC, et al. Rapid deterioration of pain sensory-discriminative information in short-term memory. Pain. 2004;110: 605–615. Marchand S, Kupers RC, Bushnell MC, et al. Analgesic effects of normal and placebo thalamic stimulation. Pain. 2003;105:481–488.
211
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
