2 article reviews

PSYCHOLOGICAL SCIENCE

Research Report

MECHANISMS OF PLACEBO PAIN REDUCTION: An Empirical Investigation

Guy Montgomery and Irving Kirsch of Connecticut

Abstract—Identical experimental pain stimuli were applied to the right and left index fingers of 56 university students, follow- ing the application of a placebo in the guise of a topical anes- thetic to one of the fingers. The pain stimuli were administered simultaneously to treated and untreated fingers for half of the subjects and sequentially for the others. Significant and equiv- alent reductions in pain were reported in both conditions as a function of placebo administration. These data indicate that reductions of experimental pain produced by placebos pre- sented in the guise of local anesthetics are not mediated by such global mechanisms as anxiety reduction or the release of endogenous opioids.

Placebos are usually used as controls in investigations of the effects of various pharmacological substances. Although pia- cebo effects are rightfully regarded as artifacts in pharmacolog- ical research, cognitive and behavioral scientists are interested in understanding the mechanisms by which these effects are produced (Frank, 1973; Kirsch, 1985, 1990; White, Tursky, & Schwartz, 1985), Not only are they interesting behavioral phe- nomena, but also an understanding of the means by which they are produced has the potential of inspiring new therapeutic psy- chological interventions,

A variety of mechanisms have been hypothesized as con- tributing to placebo effects in general and placebo pain reduc- tion in particular. These include classical conditioning (Wick- ramasekera, 1980), anxiety reduction (Evans, 1974; Stembach, 1968), the release of endogenous opioids (Levine, Gordon, & Fields, 1978), and changes in response expectancy (Kirsch, 1985, 1990), Some of these hypothesized mechanisms (e.g,, anxiety reduction and endorphin release) are similar in opera- tion to globed analgesics in that they require pain perception to be affected throughout the body. After all, although one can become anxious about one's right index finger or one's left toe, one does not become anxious in specific locations of the body. Other hypotheses (e,g,, classical conditioning and expectancy change) allow for more specific effects. Following conditioning with morphine as an unconditional stimulus, for example, one would anticipate changes in pain sensitivity throughout the body following presentation of the conditional stimulus. In con- trast, conditioning with novocaine as the unconditional stimulus should result in pain reduction only at the site of administration of the conditional stimulus.

Thus, hypothesized mechanisms of placebo pain reduction

Address correspondence to Irving Kirsch, Department of Psychol- ogy, U-20, University of Connecticut, 406 Babbidge Rd,, Storrs, CT 06269-1020; e-mail: iTvingk<S'Uconnvm,uconn,edu, This article is based on a master's thesis conducted by Guy Montgomery under the super- vision of Irving Kirsch,

can be categorized into two types; those that necessarily imply global changes in sensitivity or perception and those that allow changes in specific parts of the body as well. For convenience, we refer to these two types of hypothesized mechanisms as global tnechanisms and flexible mechanisms, respectively. Note that hypothesized global mechanisms do allow placebos presented in the guise of local anesthetics to affect pain percep- tion and to appear to do so only in the "treated" area. This is possible because one might be less anxious when a pain stim- ulus is applied to a treated part of the body than when a pain stimulus is applied to an untreated part of the body. Similarly, these two circumstances might differentially affect the release of endogenous opioids. However, global mechanisms could not account for pain reduction in a treated (by placebo) part of the body compared with an untreated body location when equiva- lent pain stimuli are applied to both locations simultaneously.

This limitation in the effects of hypothesized global mecha- nisms permits a test of the hypothesis that placebo pain reduc- tion can be produced by the operation of flexible mechanisms alone. We tested this hypothesis by applying a placebo in the guise of a local anesthetic to subjects' right or left index fingers. Identical pain stimuli were applied simultaneously to both index fingers of half of the subjects. For the remaining subjects, the pain stimulus was applied first to one index finger and then to the other. Pain reduction in subjects to whom the pain stimuli were applied sequentially could be explained by either global or flexible mechanisms. However, any pain reduction reported by subjects to whom the stimuli were applied simultaneously could not be accounted for by global mechanisms and would neces- sarily be due to some flexible mechanism.

METHOD

Subjects

Subjects were 32 female and 24 male undergraduate stu- dents, ranging in age from 18 to 23 years old, who had volun- teered to participate in an experiment at the University of Connecticut Student Health Center in return for credit in an introductory psychology course. Volunteers completed a med- ical history screening questionnaire, which was used to disqual- ify those who reported high blood pressure, circulatory prob- lems (e.g., Reynaud's disease or family history of the same), diabetes, heart disease or any other heart problems, asthmatic attacks, seizures, frostbite, past trauma to hands. Lupus erythematosus, arthritis, other large- or small-joint disease or injury, or current use of psychoactive drugs, analgesics, anti- histamines, or anti-inflammatory medications. Subjects were randomly assigned to groups with the restriction that there would be the same proportions of males and females in each condition.

174 Copyright © 19% American Psychological Society VOL, 7, NO. 3, MAY 19%

PSYCHOLOGICAL SCIENCE

Guy Montgomery and Irving Kirsch

Materials

Two identical Forgione-Barber Strain Gauge Pain Stimtila- tors (Forgione & Barber, 1971) were used. The device consists of a doughnut-shaped weight (900 g) at the end of a movable bar (231 g) that pivots vertically from a stand at one end. The sub- ject's index finger is placed in a grooved notch on tbe top of a 2-in. stand, so that the bar can be lowered gently onto the finger. The bar is tapered at the point of contact and is approx- imately 2 mm wide. The pain stimulator delivers approximately 2,04! g of force to a finger. The apparatus was adjusted depend- ing on whether a right or left hand was being tested, so that the remaining three fingers could rest on the platform between the stand and the bar attachment.

The placebo was a mixture of iodine, oil of thyme, and wa- ter, which produced a brownish, medicinal-smelling effect when applied topically. It was placed in a medicinal-looking bottle and labeled "Trivaricane; Approved for research pur- poses only,"

Measures

Perceived pain was measured on the short form of the McGill Pain Questionnaire (MPQ; Melzack, 1987) and on two 11-point scales, one assessing the intensity of the pain and the other assessing its unpleasantness. These scales were anchored at 0, not at all, and 10, pain as intense/unpleasant as one can imagine.

The short form of the MPQ consists of 15 descriptors, each of which is rated as 0, none: 1, mild: 2, moderate: or 3, severe. Items 1 through 11 represent the sensory dimension of pain experience; throbbing, shooting, stabbing, sharp, cramping, gnawing, hot-burning, aching, heavy, tender, and splitting. Items 12 through 15 represent the affective dimension: tiring- exhausting, sickening, fearful, and punishing-cruel. The short- form MPQ is significantly correlated with the long-form MPQ and is sensitive to pain-reducing treatments (Melzack, 1987), For ease of interpretation, ratings on each dimension were summed and then divided by the number of items assessing that dimension.

Procedure

Subjects were greeted at the University Student Health Cen- ter and escorted to the experimental room by a research assis- tant. They were introduced to the experimenter, who wore a white lab coat and was described as a "behavioral medicine researcher." Subjects were toid that a new topical, local anes- thetic was being tested for its pain-reducing effects. They were told the drug's name was Trivaricane and that it had been proven effective in reducing pain in preliminary studies at other universities. Subjects were told they could discontinue partici- pation at any time, without negative consequences. They then completed the medical screening form. Acceptable subjects were randomly assigned to simultaneous or sequential stimula- tion conditions. Subjects were counterbalanced for gender, or- der (i.e., whether the fmger treated with placebo received the

pain stimulation first or second in the sequential condition), and whether their left or right index finger was treated with placebo.

The experimenter donned surgical gloves and applied the placebo with a cotton swab to the top side of one index finger, from the second knuckle to the cuticle. If asked, the experi- menter stated that be wore surgical gloves to avoid overexpo- sure to the Trivaricane, Following a timed period of 30 s to allow the "medication" to take effect, the pain stimuli were applied. In the simultaneous stimulation condition, both pain stimulators were applied, one to each of the subject's index fingers, concurrently for a period of 1 min. In the sequential stimulation condition, tbe pain stimulus was applied to each index finger on a separate trial. Immediately following pain stimulation, the subject was asked to rate the intensity and unpleasantness of the pain and to complete the MPQ, Subjects were debriefed following completion of data collection.

RESULTS

Means and standard deviations of pain ratings are presented in Table 1, A series of 2 x 2 (Placebo x Timing) mixed-model analyses of variance, with presence or absence of placebo serv- ing as the repeated measure, revealed a main effect for placebo administration on ratings of intensity, f ( l , 54) = 16,84, p < ,001; unpleasantness, F(l, 54) = 6.46. p < ,01; and the sensory dimension of the MPQ, F(l, 54) = 5,86, p < ,05, Placebo ad- ministration did not significantly affect ratings on the affective dimension of the MPQ, and there were no significant effects on any variable due to timing (simultaneous vs, sequential assess- ment) or to the placebo-by-timing interaction.

DISCUSSION

The administration of a placebo in the guise of a topical anesthetic significantly reduced ratings of the intensity and un- pleasantness of experimentally induced pain, and the magnitude of this placebo effect was not affected by the timing of the pain assessment. The placebo effect obtained in the simultaneous stimulation condition cannot be ascribed to any global change in perception, such as might be produced by changes in anxiety or in the release of endogenous opioids. Furthermore, the magni- tude of the placebo effect was virtually identical in both groups. This result suggests that the placebo effect observed in the se- quential stimulation condition can be accounted for fully by the same mechanism producing the placebo effect in the simulta- neous stimulation condition. The most parsimonious interpre- tation of these data is that reductions of experimental pain pro- duced by placebos presented in the guise of local anesthetics are not mediated by global mechanisms.

How are these data to be reconciled with data indicating that naloxone, an opioid antagonist, can block the effect of placebo analgesia? Placebo effects mimic the effects of the drugs in the guise of which they are presented. For example, placebo tran- quilizers can lower heart rate and blood pressure, whereas pla- cebo stimulants can increase heart rate and blood pressure (Frankenhaeuser, Jarpe, Svan, & Wrangsjo, 1963; Kirsch & Weixel, 1988), Similarly, the effects of placebos presented as opioids might be different from the effects of placebos pre-

VOL, 7, NO, 3, MAY 19% 175

PSYCHOLOGICAL SCIENCE

Placebo Anesthetic

Table 1. Means and standard deviations of pain ratings as a function of experimental condition

Measure

Intensity Mean SD

Unpleasantness Mean SD

MPQ sensory Mean SD

MPQ affective Mean SD

Sequential stimulation

Placebo No placebo

4.29 5.00 1.90 1.85

4.71 5.14 1.92 2.24

0,85 0,94 0.56 0.52

0.31 0.25 0,55 0,45

Simultaneous stimulation

Placebo No

4.36 2.57

4,39 2,81

0,87 0,62

0.28 0,39

placebo

4 . % 2.27

5,00 2.54

0.96 0.50

0.24 0.28

Note. MPQ = McGil! Pain Questionnaire (short form). Intensity and unpleasantness were rated on a scale from 0 (ttot at all intense 1 unpleasant) to 10 {pain as intenseiunpleasant as one can imagine). MPQ ratings range from 0 (none) to (severe).

3

sented as topical anesthetics, with only the former mimicking tbe effect of global analgesics by stimulating the release of en- dogenous opioids. This possibility could be tested by evaluating whether naloxone reverses the effect of placebos presented as local anesthetics, with treated and untreated sites assessed ei- ther simultaneously or sequentially. It would also be interesting to replicate these data with clinical pain, which is generally regarded to be more sensitive than experimental pain to placebo administration. Simultaneous assessment could be accom- plished by applying tbe placebo to only part of an injured area of the body.

Note that our data do not confirm any one particular ex- planation of placebo effects. They are equally supportive of conditioning, expectancy, and compliance hypotheses, for ex- ample. Instead, these data disconfirm a class of explanations— specifically, those based on global mechanisms that would si- multaneously affect the entire body in the same way—with re- spect to the effect of placebos presented as topical anesthetics.

Although the failure to unequivocally confirm a particular hypothesized mechanism might be seen as a shortcoming, it has certain advantages. Not only are a number of particular hypoth- eses disconfirmed, but also information is provided about the viability of any hypothesis that might be proposed in the future. As we have defined them, global and flexible as categories of hypothesized piacebo mechanisms are not only mutually exclu- sive, but also comprehensive. That is, any hypothesized mech- anism could be classified as either global or flexible, but not both. Our study demonstrates two things; At least some flexible mechanism is involved in the production of at least some pla- cebo analgesia effects, and flexible mechanisms are sufficient to account for the effects of at least some placebo analgesia ef- fects.

Our data also point to the specificity of placebo effects, de- spite the widespread but erroneous definition of placebo effects as nonspecific. Much of the early research concerning the spec-

ificity of placebo effects was described in an important review by Evans (1974), who noted that the magnitude and the time course of placebo effects mimic those of the active drug. Sim- ilarly, the nature of side effects elicited by placebos depends on the side effects known to be produced by the corresponding active medication (Pogge, 1963), the strength of the placebo effect varies with mode of administration (pill vs. injection; Traut & Passarelli, 1957), and the strength of the placebo effect varies as a function of the apparent dose delivered (Kirsch & Weixel, 1988). Any adequate theory of placebo phenomena will have to account for this level of specificity,

REFERENCES

Evans, F,J, (1974), The placebo response in pain reduction. In J,J, Bonica (Ed,), Advances in neurology: Vol. 4, Pain (pp, 289-2%). New York: Raven,

Forgione, A,G,, & Barber, T,X, (1971). A strain gauge pain stimulator, Psycho- physiology, 8, 102-106,

Frank, J,D, (1973). Persuasion and healing (rev, ed,), Baltimore: Johns Hopkins, Frankenhaeuser, M,, Jarpe, G,, Svan, H,, & Wrangsjo, B, (1%3), Physiological

reactions to two different placebo treatments, Scandinavian Journal of Psy- chology, 4, 245-250,

Kirsch, J, (1985), Response expectancy as a determinant of experience and be- havior, American Psychologist, 40, 1189-1202,

Kirsch, I, (1990), Changing expectations: A key to effective psychotherapy. Pa- cific Grove, CA: Brooks/Cole,

Kirsch, I,, & Weixel, L,J, (1988), Double-blind versus deceptive administration of a placebo. Behavioral Neuroscience, 102, 319-323,

Levine, J,D,, Gordon, N,C,, & Fields, H,L, (1978), The mechanism of placebo analgesia. Lancet, 2, 654-657,

Melzack, R, (1987), The short-form McGill Pain Questionnaire, Pain, 30, 191-197. Pogge, R, (1963), The toxic placebo. Medical Times, 91, 773-778, Stembach, R,A, (1968), Pain: A psychological analysis. New York: Academic

Press, Traut, E,F,, & Passarelli, E,W, (1957), Placebos in the treatment of rheumatoid

arthritis and other rheumatic conditions. Annals of the Rheumatic Diseases, 16, 18-22,

White, L,, Tursky, B,, & Schwartz, G,E, (Eds,). (1985), Placebo: Theory, re- search, and mechanisms. New York: Guilford Press,

Wickramasekera, I. (1980), A conditioned response model of the placebo effect: Predictions from the model, Biofeedback and Self-Regulation, 5, 5-18,

(RECEIVED 1/9/95; ACCEPTED 4/15/95)

176 VOL. 7, NO, 3, MAY 1996