Samueli.stage.bridgelinedigital.net

THE JOURNAL OF ALTERNATIVE AND COMPLEMENTARY MEDICINE
Volume 10, Supplement 1, 2004, pp. S-103–S-112
Mary Ann Liebert, Inc.
Placebo Research: The Evidence Base for Harnessing HARALD WALACH, Ph.D.,1 and WAYNE B. JONAS, M.D.2 ABSTRACT
Placebo effects are often considered irrelevant at best and a nuisance at worst for determining what is valu- able in medicine. In this paper, we argue that research that involves placebo provides critical information forhow the mind, body, and culture heal. Following a newly proposed definition of placebo as a therapeutic mean-ing response, empirical evidence is reviewed that emphasizes the importance of these effects for developing ascience of healing. It is likely that the effects resulting from the individual meaning of an intervention are animportant factor of any therapeutic approach. It would be therapeutically desirable to maximize these factorsand have good evidence on which to base healing interventions. We show how this could be achieved.
INTRODUCTION
normally neglected. We argue in this review that this bodyof literature teaches how we could maximize healing by har- Until recently, many authors dealing with placebo effects nessing these factors in any therapeutic context.
were not interested in the placebo effect as such, nor in conceptual clarity. Rather, they combined a number ofpossible confounding factors such as spontaneous remission,measurement artifacts, and regression to the mean, all un- HISTORY OF THE NOTION OF PLACEBO
der the heading of “placebo.”1–4 This has fueled a debate AND THE MEANING RESPONSE
and obscured the real issue, namely, whether psychologicprocesses and social contexts that facilitate hope, expecta- The term “placebo” derives from the Latin psalm verse, tion, positive feelings, relief of anxiety and anticipation of “placebo Domino in regione vivorum” (Psalm 116:6 mod- improvement are able to truly affect physiologic processes, ern counting): I shall please the Lord in the land of the liv- and contribute to healing over and above pharmacologically ing.7,8 This psalm was part of the prayers offered at the deathbed in the Middle Ages. At later times, it was cus- It is necessary to distinguish between the true placebo ef- tomary to pay others to sing the rites. Hence the connota- fect and artifacts.5 Placebo-controlled clinical trials nor- tion of “placebo” emerged as a fraudulent replacement of mally cannot distinguish between a true therapeutic response the real. Placebos were popular in the era of medicine when and other confounding factors.6 Hence, it is not helpful to effective pharmacologic interventions were scarce and of- average improvement rates of placebo arms of clinical tri- ten fraught with side effects. The emergence of randomized als to find out about the magnitude of the placebo effect.
controlled trials (RCT) tagged the placebo as all those “un- However, evidence from the psychologic literature, and real” effects that were not the result of pharmacologic in- from experiments especially targeting the question of mech- terventions.9 Today, “placebos” normally mean inert sub- anisms of placebo effects, are beginning to clarify the mes- stances that are given to subjects, mostly in the context of sage that behind the facade of what we normally call placebo a scientific study, to control for psychologic and social or effects are the self-healing capacities of the person, a fact 1Department of Environmental Medicine and Hospital Epidemiology, Samueli Institute for Information Biology, University of 2Samueli Institute for Information Biology, Alexandria, VA.
WALACH AND JONAS
The many attempts at a definition of placebo effects all The study, while methodologically sound, can be criti- have one thing in common.10 They all define placebo ef- cized on three grounds: (1) by restricting the analysis only fects as something negative. As either the psychologic (side) to randomized studies, many intriguing results from earlier effect of an intervention, as unintended effects, as nonspe- studies that were not randomized were discarded. Many of cific effects or even as effects caused by a pharmacologi- those studies show clear and impressive results.15 Also, treat- cally inert substance. A more useful definition has been pre- ments that were not pharmacologic were dismissed, thereby sented recently by Moerman and Jonas,11 based on similar not including the effects of many healing interventions such and earlier work by Brody12: Placebo response is defined as as psychotherapy. A more narrative and qualitative approach the effect that is due to the meaning of a therapeutic inter- shows that the evidence is in favor of effects of placebo in- vention for a particular patient and context. This new defi- terventions compared to no treatment.10 (2) A recent review14 did not consider what type of placebo intervention had beenadministered. For instance, a study had been included that 1. It is a semiotic definition in that it acknowledges that hu- compared the administration of an analgesic in still uncon- mans are not deterministic machines reacting to me- scious patients after surgery to placebo and no treatment.16 chanical causes (e.g., pharmacologic agents). Rather, It is hardly surprising no placebo effect was observed. (3) they are responding to signs and the meaning those signs No attempt was made to differentiate between studies that generate in a highly complex, often self-determined and used measures to maximize the placebo effect and those that sometimes unpredictable fashion.13 The meaning is tried to minimize it or did not pay attention to it.17 If the 23 something that is not fully determined by the external studies that used placebo as a control procedure are analyzed stimuli themselves, but arises from the interaction be- separately and compared with 14 studies that tried to maxi- tween the external environment and the internal condi- mize meaning responses through suggestions, there is a clear tions of persons, their history, their social circumstances, and significant difference between those two sets of effect their individual predilections and their expectations.
sizes. Studies that did not attempt to maximize meaning re- 2. The meaning response definition allows the context fac- sponses showed an effect size of d ϭ 0.15 of placebo against tors of an intervention to enter the stage. This definition no treatment, and studies that tried to enhance meaning re- underlines the individual differences in response to oth- sponses had an effect size of d ϭ 0.95, which is both signif- erwise similar conditions, and brings into focus the im- icantly different and clinically important.
portance of individuality in therapy. This makes plausi- An indirect attempt at quantifying placebo effects against ble why one and the same situation, for instance surgery, no treatment controls was made by Kirsch and Sapirstein.18 may arouse hope in one patient and induce fright in an- In a meta-analysis of 19 antidepressant medication trials, other with completely different physiologic reactions and they looked at improvement rates with antidepressants and placebos. Additionally, they compared the improvementrates obtained in the placebo groups of the antidepressanttrials with those of the effects of psychotherapy trials that ARE PLACEBO EFFECTS REAL?
had included waiting list control groups. Thus, their com- COMPARISON BETWEEN PLACEBO
parison with no-treatment controls was indirect and weaker, GROUPS AND UNTREATED GROUPS IN
although the populations in the two sets of studies were com- CLINICAL TRIALS
parable. They found a medium effect of d ϭ 0.39 for thecomparison between pharmacologic intervention and place- Two-armed clinical trials with one placebo group do not bos, and an effect of d ϭ 1.6 for the comparison between tell us much about the placebo effect, because both groups psychotherapy and no treatment. By indirect comparison, control for many different factors affecting treatment apart the authors estimate the placebo effect against no treatment from the real intervention. Therefore, reviews of three- to be d ϭ 0.79, which is sizeable, but less than the one re- armed trials are preferred. A meta-analysis of 130 three- armed trials that included only randomized studies compared These results show that there is a beneficial effect of the placebo arms of those studies against the untreated con- placebo administration in the context of clinical trials over trols.14 It yielded equivocal results: those studies with bi- and above natural history, spontaneous remission, and re- nary outcomes showed a small effect of a relative risk ϭ 0.95 (95% confidence interval [CI]: 0.88 to 1.02). Studieswith continuous outcomes, mostly visual analogue scale(VAS) measures of pain, showed a small, but significant ef- CONTEXT AND EXPECTATION
fect of d ϭ Ϫ0.28 (95% CI: Ϫ0.38 to Ϫ0.19). These resultssuggest that in those trials, true placebo effects compared to The meaning model of the placebo effect suggests that natural history are therapeutically not very important, be- the context within which a treatment is offered changes its cause effect sizes are rather small.
meaning and hence its effects. This has been supported by PLACEBO RESEARCH
a systematic review19 that included studies that offered a compared interventions of acid blockers in ulcers. Earlier treatment and the corresponding placebo under different drugs had to be taken four times per day, while the newer conditions. This review concluded that neither interventions drugs had to be taken only twice a day. They compared the nor placebos are indifferent to meaning and context factors.
placebo response rates in 51 studies with a regimen of four Treatments can be more effective than their placebos or vice times per day dosage with 28 studies with a twice daily dose.
versa depending on those context factors. An example is the Although the difference in the response rate of the placebo study by Bergman et al.20 in which the same trial was con- groups is only 6% to 8%, the effect is significant with a ducted twice. Patients with cancer who regularly received number needed to treat of 14. This suggests that the more naproxen as a routine analgesic were either given their nor- frequent application of a drug raises different expectations mal bedside medication or a placebo. Some knew they were than an application twice daily to the point of needing to part of a trial, and some were informed about the trial and treat 14 more patients with a twice per day regimen to ob- knew that they would be randomized to either naproxen or tain the same effect as a four times per day regiment.
placebo. Informed patients who received the active medica- Perhaps one of the most direct proofs for the power of tion had the greatest pain relief, followed by their placebo expectation is a recent experimental study on irritable bowel counterparts. Patients who had received placebo in the for- syndrome.28 Thirteen (13) patients received an experimen- mal trial had greater pain relief than the patients who had tal rectal distention and heat stimulus as a pain stimulant to received the normal naproxen medication without knowing test different analgesic interventions. Apart from the natural history of the pain, each patient received in randomized or- Thus, expectation of potential effects is important. An- der all of the following interventions: rectal lidocaine, oral other example is a study of the effects of analgesics in post- lidocaine, rectal placebo (a lubricant) with the suggestion of partum pain in two separate trials, one with paracetamol improvement, and rectal placebo with the suggestion of po- against placebo, and one with paracetamol against naproxen.
tential aggravation (i.e., a nocebo intervention). Desire for Patients were informed in one study they would be receiv- pain control and expectation of pain reduction were also ing placebo with a chance of 50%, and in the other study measured. There was a highly significant effect for the one of two active medications. The difference in effective- placebo intervention, the oral and the rectal lidocaine com- ness of paracetamol was 20 mm on a 100-mm VAS between pared to natural history. While neither the rectal nor the oral the trials with all other variables including researchers, set- lidocaine were distinguishable from the placebo, the nocebo ting, time, and patient population being the same.21–23 intervention produced a nonsignificant increase in pain per- Another trial illustrates how expectancy is a factor me- ception. The effect size of placebo-pain reduction versus nat- diating a meaning response. In a study comparing the ex- ural history was d ϭ 2.0, and versus nocebo was d ϭ 2.27.
perimental drug to sumatriptane, an established drug and Desire and expectation of pain control could explain 77% placebo, the ethics committee decreed that the randomiza- of the variance in pain ratings in a regression model for tion ratio to placebo should be 16:1, because it is known placebo and 81% for lidocaine. Thus, expectation was the that triptanes are effective and hence as few patients as pos- most important factor in this study, even for the effective- sible should be exposed to the risk of placebo. This study ness of the pharmaceutically active agent. The authors un- was unable to show superiority of any of the two active drugs derline that it was probably important that the same physi- against placebo.24,25 The active medication showed an ef- cian who treated the patients in a normal context was fect in 42% of the cases, and the placebo an effect in 38%.
responsible for the experimental interventions, and thus Because patients in the placebo group knew the randomiza- these effects might depend on a good relationship between tion ratio, they had a strong hope and expectation of actu- ally receiving active treatment, and it was this expectation The same conclusions can be drawn from a recent clini- that contributed to the strong effect. A meta-analysis of all cal trial of massage, acupuncture, and self-education in 262 triptane studies26 shows that those studies that had a ran- patients with chronic low-back pain. This trial showed mas- domization ratio to placebo different from 1:1, and hence sage and self-education were superior to acupuncture after provided a greater likelihood for patients to expect real treat- a year.29 This trial also assessed general and specific ex- ment, yielded higher placebo response rates than studies pectation of patients and reanalyzed the data according to with symmetrical randomization. The authors conclude that expectations.30 Those patients who had the largest expecta- ethical requirements threaten scientific progress if they re- tion of change for a specific treatment had the largest ther- quire unsymmetrical randomization ratios, which in turn apeutic benefit. When all other factors were controlled sta- drive patients’ expectations, which again inflate placebo re- tistically, specific expectation alone showed a significant sponse rates due to these expectations.
odds ratio (OR) of 5.3, meaning that patients with high ex- Supportive evidence of the importance of such expecta- pectation had a fivefold chance of benefiting from the treat- tions resulting from the context comes from a meta-analy- ment, all other things being equal. The authors underline sis that studied the question whether more frequent ap- that this effect of expectation was larger than the treatment plication of placebos yields larger effects.27 The authors effect in comparable low-back pain trials.
WALACH AND JONAS
Taken together, these data show that expectation is prob- on top) and expectancy manipulations, and then view stim- ably the most important meaning factor of a treatment, and ulating visual material. Here, expectancy of receiving alco- can be as powerful as a specific pharmacologic intervention.
hol produces strong effects, independent of the substance It is plausible, then, to suppose that the meaning response actually ingested. Measures in these studies were objective is nourished to a large extent by the expectation a specific measures of sexual arousal, such as penile erection or vagi- nal blood flow, that address social desirability.2 A quantitative meta-analysis of 34 studies on alcohol and alcohol expectancy has found small but significant effects SUGGESTION AND EXPECTATION
of expectancy.40 Across all studies, the effect of alcohol issignificantly different from zero, g ϭ 0.18 (effect size mea- Expectations may be indirectly altered or manipulated di- sure Hedge’s g, similar to the normally used Cohen’s d).
rectly by suggestion. The power of suggestion historically The effect of expectation is smaller, but also significantly was brought to the attention of the scientific community by different from zero, g ϭ 0.08. What is more important is the the investigation of Mesmerism in Paris 1784.9 These stud- fact that expectancy effects can be quite sizeable and even ies showed that the claims made by the followers of Mes- larger than those of alcohol in situations, whereas the social mer, at least in formal studies, were largely the result of sug- setting, but not necessarily the pharmacology of alcohol, gestions, as the effects could only be seen when therapist suggests alcohol should have an effect such as settings of and subject had visible contact, allowing subtle communi- antisocial behavior (g ϭ 0.4), sexual arousal (g ϭ 0.3), or cation.31,32 The interest in suggestion waned with the rise craving (g ϭ 0.5). When interpreting these data, it is im- of behaviorism and later cognitive therapy. Interest in sug- portant to recognize that the amount of alcohol used in these gestion now is increasing,33 as the neurosciences are able to studies was small, normally equivalent to one drink (a can measure and better understand the intricate complexities of of beer, a glass of wine). In addition, experimental models subconscious and preattentive processes.34,35 are only proxies for clinical situations in which patients are Suggestion and the subconscious processes it triggers differently motivated by their desire to get healthy again.
may be the link between the meaning of an external situa- Recent studies have been published supporting the results tion and physiologic responses. In many ways placebo ef- fects are akin to therapeutic hypnosis,15,36 with the differ-ence being that in hypnosis a patient actively and willinglyagrees to the procedure. The clarifying psychological liter- MECHANISMS: ENDORPHINS,
ature is vast and has been reviewed elsewhere.15 However, CONDITIONING, AND
there are relevant aspects that highlight the mechanisms of CENTRAL PROCESSES
healing. In a convincing piece of evidence for the physio-logic effectiveness of suggestions, Butler and Steptoe37 gave a water aerosol as an inhalant to 12 subjects with asthmausing a balanced, crossover experimental study. While ini- When Levine et al.47–52 published findings on the rever- tially subjects were told the aerosol was a bronchodilator, sal of placebo analgesia by the administration of naloxone, in later sessions it was described as a powerful broncho- they thought they had solved the placebo puzzle. In these constrictor. The placebo, given with the suggestion of bron- studies, patients suffering from either postoperative or ex- chodilation, was able to reverse the suggested bronchocon- perimental pain, received either a placebo infusion without strictive effect, both compared to the control situation and their knowledge through a covered indwelling line, or nalox- one, an opiate antagonist. With their knowledge, they then Psychologic research has studied several pharmacologi- received another placebo injection that was claimed to be cally active substances, such as caffeine, alcohol, or cannabis, either a potent analgesic or a control substance. Levine and comparing their pharmacologic to psychological properties colleagues observed that the opiate antagonist reversed or in balanced placebo design.38 This design balances substance attenuated a placebo analgesia produced by the suggestion and expectation in a 2-by-2 factorial design, with one factor of administering an alleged painkiller. They concluded that being the substance versus placebo. The other factor ma- the substances responsible for this effect must be endoge- nipulates meaning by giving either correct or misleading in- nous opiates, which mediate centrally modulated pain and formation. This design allows for a separation of pharma- analgesia.53,54 Later studies,55,56 with added new controls or cological and psychological effects of substances,39 and the more sophisticated procedures,57,58 basically replicated the demonstration of the effects of suggestion. The first review38 initial findings of Levine and colleagues. A paper review- showed that strong effects of expectancy can be observed ing this evidence concludes that the effects are real and me- and vary by the setting. For instance, in studies on sexual arousal, subjects either receive alcohol or an appropriate The same conclusion was reached indirectly by Lichtig- placebo (normally tonic with a few drops of vodka sprinkled feld and Gillman,60 with the addition of the central role of PLACEBO RESEARCH
nitric oxide to the placebo response. These researchers con- those that are affected by naloxone.58 These opiate depen- ducted studies on the effects of nitric oxide in postwith- dent effects can be targeted toward specific body parts,80 a drawal depression of alcoholics. When titrated and admin- finding that supports the expectancy hypothesis. There are istered in low doses such that patients remain conscious, also objective effects (i.e., depression of lung function).
nitric oxide, normally a potent narcotic, relieves withdrawal Such objective opiate-dependent effects cannot be explained depression quickly and effectively.61–64 This psychotropic by expectancy, and likely are conditioned.81 There seem to analgesic nitric oxide (PAN) is superior to placebo (air)61 be two systems active, both of which are mediated by en- even though the response rate to this placebo can be as high dogenous opiates. One operates via expectancies, and the as 50% of cases. Because PAN is effective in alleviating other operates via conditioning. Additionally, there seem to 95% of the cases of postwithdrawal depression, the authors be conditioned effects which are not dependent on opiates.79 conclude that the endogeneous opiate system must be in- Response expectancies are also effective in clinically rele- volved. Because medical air and oxygen have the same ef- vant settings such as postoperative pain.78 The effect of ex- fect in many cases, this effect must be the result of activa- pectancy was determined by open (unblinded) or hidden tion of the endogenous opiate system. This is also suggested (blinded) administration of analgesics to patients, postopera- by Stefano et al.65 who imply that endogenous nitric oxide tively or experimentally, without administering placebo. Open is the hub around which both the immunologic and the af- administration involved an arousal of expectancies and pro- duced significantly stronger effects.77 Similar findings werefound with patients suffering from Parkinson’s disease,76who had had subthalamic electrodes implanted to stimulate dopaminergic neurons. In one condition, they were told about Is this system activated by expectancy or by condition- the actual stimulation levels and their reduction. In another ing, or both? That autonomous processes can be subject to condition, they expected the stimulation to be active but it was conditioning has been uncontested since Ader’s ground- in fact reduced. After 30 minutes, significantly different ef- breaking studies on conditioning immune responses in fects in motor tasks were seen, indicating that endogenic rats.66,67 These studies make plausible that autonomous processes were activated by the expectation. The most recent processes, such as the activation of the immune system or study75 has shown a clear effect of expectation in a pain model a neurotransmitter system, could be reinforced by operant while heart rate variability measures were also taken. Given conditioning or by a keying stimulus as in classic condi- as an alleged analgesic, placebo reduced pain and the low-fre- tioning. Thus, placebo effects could be conceived of as con- quency power spectrum component of heart rate variability ditioned reactions.68 On the other hand, social cognitive the- caused by sympathetic activation. This effect was antagonized orists suggest that the placebo effect is controlled by by naloxone, again supporting evidence that endogenous opi- response expectancies69: if a response is expected by an or- oids are responsible for the mediation of this expectancy ef- ganism, it is more likely to happen. There are studies sup- fect. It remains to be seen which class of opioids influencing porting a conditioning model,70–72 as well as studies sup- the parasympathetic system is responsible for these effects.
porting expectancy models. Current evidence suggests that There seem to be several processes that mediate mean- the conditioning effect is mediated by expectancies.73,74 ing responses, depending on the paradigm and the context.
Benedetti and colleagues58,75–84 have presented evidence It is reasonable to assume that neurotransmitters such as the both for the reality and clinical relevance of placebo effects endogenous opioids have a role. It is also likely other sys- and their mechanisms through different neurotransmitter tems like dopaminergic pathways85 mediate these effects. systems. There are some pertinent points relevant to heal- If central neurotransmitter systems are active in mobilizing ing effects in these complex experiments. First, natural his- expectation effects, it is also plausible that many other ef- tory control groups were run, showing that the effects in the fects in addition to pain alleviation or antidepressant re- placebo groups are clearly different from the effects in no- sponses85,86 are mediated by those systems. It is well-known, treatment control groups. Thus, the effects of placebo are for example, that practically every neurotransmitter,87 in- not merely artifacts. Second, these studies show both that cluding endogenous opioids88,89 and serotonin,90–92 have not only are endogenous opiate systems involved in placebo analgesia, but specific information and meaning given to apatient can modulate specific receptor families involved in producing analgesia and direct those effects to certain areasof the body. For instance, cholecystokinine (CCK) antago- There are two primary opiate networks in the brain, one nizes morphine effects that operate via opiate receptors.
in the brainstem, and one in the cortex, involving the ros- Proglumide is an antidote to CCK, and therefore acts syn- tral anterior cingulate cortex, (rACC) and the ventromedial ergistically with opiates. It was shown that proglumide not prefrontal cortex, respectively. In an experimental, counter- only potentiates placebo-mediated analgesia,84 but also op- balanced, within-subject design, nine subjects received heat erates via different receptor systems independently from or warmth, followed by opiate or placebo, while their re- WALACH AND JONAS
gional cerebral blood flow was monitored.93 It could be 1. Meaning and healing effects are real and can be quite demonstrated that in placebo responders, the same areas are active during placebo analgesia as with opiates, namely the 2. If conceived as an individual response to the meaning of nuclei in the rACC. Nuclei in the pons covaried with the ac- an intervention, many paradoxes inherent in traditional tivities of the rACC. Notably, the activation of the rACC usages of the concept of placebo disappear.
was only seen in placebo responders.
3. This latter usage of the concept can also contribute to a Another part of the puzzle has been illuminated by a broader understanding of healing responses, which seem positron emission tomography (PET) imaging study of the to be triggered by central processes, either through ex- effects of apomorphin and placebo on dopamine release in pectation, or through conditioning, or both, and can in- patients with Parkinson’s disease.94 The study utilized the volve multiple central mechanisms and neurotransmitter competition of radioactively marked raclopride (RAC) and endogenous dopamine. The authors observed a 17% and 19% 4. Placebo analgesia is mediated by endogenous opiate sys- diminuation of RAC by placebo administration in the nu- tems which are similar to those activated by exogenous cleus caudatus and the putamen, respectively. Both areas con- tain many dopamine producing neurons. This finding sug- 5. Other systems besides the endogenous opiate system are gests that patients expecting dopaminergic pharmacological involved, such as the dopaminergic system.
effects will produce dopamine. Dopamine is an importantneurotransmitter which activates the reward system,95 and is It is time to change our perspective on placebo and mean- important in learning.86,96,97 Thus, this finding illustrates ing effects in research and medicine. Rather than viewing how placebo may produce effects beyond pain relief and in- the placebo effect as an enemy that hampers clinical trials, fluence areas such as affect, learning and motivation.
it should be seen as a ubiquitous healing response mediated An imaging study of placebo effects in an antidepression by expectations and conditioning. Thus, it can be utilized to study using the selective serotonin reuptake inhibitor (SSRI) enhance or interfere with healing in many clinical settings.
fluoxetine showed a clear overlap of areas activated in By understanding the meaning response, we might under- placebo responders and drug treatment responders.98 While stand how optimal healing can be fostered.
activation of the thalamus was reduced in both groups and Complementary and alternative medical (CAM) therapies activation of prefrontal areas enhanced, fluoxetine showed may be elegant, efficient and comparatively harmless ways enhanced activity in the pons and reduced activity in the hip- to harness healing processes.100 We should view that possi- pocampus and striatum. These findings were not seen with bility as a virtue rather than a vice. But CAM is not the pro- placebo. Fluoxetine effects were generally more pronounced prietor of all meaning responses. These responses are overall, but activation of the right prefrontal cortex was more ubiquitous and occur in every healing context. Table 1 lists ways to enhance healing with any therapy derived from the This finding is qualified by another recent antidepressant research literature on placebo effects.
imaging study99 that used high-resolution quantitative elec- In addition, evidence points to the following suggestions trocardiogram (EEG) to locate areas of higher or lower elec- on how to harness these optimal healing processes: trical activity. In this study, it was found that placebo re-sponders, drug responders, and nonresponders had distinctive 1. Always work with and not against patients’ expectations.
activation patterns. While placebo responders showed in- If patients expect an intervention to be harmful, danger- creased activity in the prefrontal cortex, drug responders ous, fraught with side-effects, and not curative, they are showed decreased activity, and nonresponders showed no likely to experience just that. Thus, it should be manda- change. This seems to show that meaning produced specific tory for every physician and therapist to find out about areas of altered brain metabolism and activity.
those expectations and move them in a positive direction.
These findings are the first of their kind and replications If multiple treatment choices are available, use the one are called for to help confirm and clarify our understanding which is most conforming to patients’ expectations for of the mechanisms of meaning effects. They have already improvement. For instance, if a patient expects to get bet- dispersed doubt about the reality of meaning response ef- ter from a “natural product” rather than from a chemical one, it is likely that this preference has clinically impor-tant influences.
2. If patients’ expectations are unhealthy or harmful, work CONCLUDING REMARKS: HARNESSING
to change them first before jumping from intervention to MEANING EFFECTS
intervention. Although modern day patients are some-times surprisingly educated, they also sometimes cling to Although this review is not exhaustive, and has high- either outdated or faddish beliefs. It is likely that inter- lighted a selective list of findings, it has illustrated a num- ventions are unsuccessful or less effective if patients’ ex- ber of issues about healing and its mechanisms: pectations are not fulfilled. In addition, it is likely that PLACEBO RESEARCH
TABLE 1. WAYS TO ENHANCE HEALING RESPONSES medicine, made the effects produced by the general ambience of treatment the most important of therapeutic 1. Use more frequent dosing rather than less frequent 4. One of the greatest skills of a doctor, and a topic often 2. Apply therapies in “therapeutic” settings such as hospitals left out of the debate around evidence-based medicine, is individualization. It is in the subtle changes to therapy 3. Match the appearance, such as size and color, to the desired and how they are delivered by a skilled healer that the 4. Attend to the route of administration.a meaning response is harnessed to its fullest. It is expected 5. Deliver therapies in a warm and caring way.c that any therapist who individualizes his treatment will 6. Deliver therapies with confidence and in a credible way.d have better results, because he can harness the meaning 7. Determine what treatment your patient believes in or 5. Raising hope and alleviating anxiety in a credible way is 8. Be sure you as a therapist believe in the treatment and find one of the most therapeutic acts in general. It has been 9. Align all beliefs congruently: patient, doctor, family, shown empirically that a simple act, such as giving a clear diagnosis and prognosis, improves outcome.105 If patients 10. Deliver a benign but frequent conditioned stimulus along receive clear and positive communications conveyed with trust, credibility, and confidence, healing is more likely.
11. Use the newest and most prominent treatment available.h,i 12. Use a well known name brand identified with success.j This must be mastered in a world in which knowledge is 13. Cut or stick the skin or poke into an orifice whenever it is transitory, and changes quickly and frequently contra- dicts previous knowledge. Truth and integrity are inte- 14. Inform the patient what they can expect.20,21 gral components of a trusting relationship; however, the 15. Use a light, laser, or electronic device to deliver and track patient should not be made the primary target for trans- 16. Incorporate reassurance, relaxation, suggestion, and anxiety reduction methods into the delivery.28,33,65 6. A frequent assumption is that only specific causal effects 17. Listen and provide empathy and understanding.12,c count, like those produced by drugs or surgery. This re- view makes it plausible that other effects also count. Be- lieving that one has a potent therapeutic agent at one’s de Craen AJ, Tijssen JG, de Gans J, Kleijnen J. Placebo effect in the acute treatment of migraine: Subcutaneous placebos are bet- command may be the single most important ingredient ter than oral placebos. J Neurol 2000;247:183–188.
for producing a broad spectrum of meaning responses.
bde Craen AJ, Roos PJ, Leonard de Vries A, Kleijnen J. Effect For it is only when a physician believes in what he uses of colour of drugs: Systematic review of perceived effect of drugs that he can fully convey competence and positive ex- and of their effectiveness. Br Med J 1996;313:1624–1626.
pectations. Thus, applying interventions which patients Thomas KB. General practice consultations: Is there any point in being positive? Br Med J 1987;294:1200–1202.
demand without real conviction is not an evidence-based dUhlenhuth EH, Richels K, Fisher S, Park LC, Lipman RS, Mock J. Drug, doctor’s verbal attitude and clinical setting in the symp- 7. “Giving placebos” is not identical to using the meaning tomatic response to pharmacotherapy. Psychopharmacologia 1966; response therapeutically. One need not give sugar pills.
However, in some cases the use of nonactive or mini- Cassidy CM. Chinese medicine users in the United States. Part II: Preferred aspects of care. J Altern Complement Med 1998;4: mally active drugs might be a better option than contin- uous medication of toxic but effective therapies. Should fMoerman DE. Cultural variations in the placebo effect: Ulcers, the patient respond favorably, it would be a mistake to anxiety, and blood pressure. Med Anthropol Q 2000;14:1–22.
attribute the problems to “psychologic problems.” The Phillips DP, Ruth TE, Wagner LM. Psychology and survival.
meaning response teaches us that there is not a clear di- hLange RA, Hillis LD. Transmyocardial laser revascularization.
vide between the mental and physical.
8. Therapeutic rituals103 might be helpful in eliciting the iJohnson AG. Surgery as a placebo. Lancet 1994;344:1140– meaning response. A significant portion of the effects from modern devices used in both conventional and com- Margo CE. The placebo effect. Surv Ophthalmol 1999;44: plementary medicine may be caused by such effects. It may be useful to develop one’s own rituals with patients,like taking a drug after a morning bath, in a special room, going against expectations will reduce compliance, a fac- before or with prayer, or having it administered by a tor central to any therapeutic success.101,102 3. Talking can induce a response toward cure. Rapport between doctor and patient is an important vehicle for Taken together, we have shown that placebo effects, re- suggesting therapeutic effects and enhancing expecta- framed as meaning responses, can evoke powerful healing tions. Frank,103,104 in his seminal work on meaning in and should be cherished rather than chided. The meaning WALACH AND JONAS
response is ubiquitous, exists and can be used or abused in Interdisciplinary Explorations. Cambridge, MA: Harvard any therapeutic context. To ignore it is to risk having it pro- duce random and possibly harmful effects. To understand it 16. Sinclair R, Cassuto J, Högström S, et al. Topical anaesthesia and use it intelligently is to increase therapeutic benefit.
with lidocaine aerosol in the control of postoperative pain.
Placebo research can and should be directed toward pro- viding the evidence base for developing optimal healing 17. Vase L, Riley JL, Price DD. A comparison of placebo effects in clinical analgesic trials versus studies of placebo analge- 18. Kirsch I, Sapirstein G. Listening to prozac but hearing placebo: A meta-analysis of antidepressant medication. Pre- ACKNOWLEDGMENT
vention & Treatment 1998;1:2a. Online document at: jour-nals apa org/prevention 19. Kleijnen J, de Craen AJM, Van Everdingen J, Krol L. Placebo Our work is supported by the Samueli Institute, Newport effect in double-blind clinical trials: A review of interactions with medications. Lancet 1994;344:1347–1349.
20. Bergmann J-F, Chassany O, Gandiol J, et al. A randomised clinical trial of the effect of informed consent on the anal-gesic activity of placebo and naproxen in cancer patients. Clin REFERENCES
21. Skovlund E. Should we tell trials patients that they might re- 1. Beecher HK. The powerful placebo. JAMA 1955;159:1602– ceive placebo? Lancet 1991;337:1041.
22. Skovlund E, Fyllingen G, Landre H, Nesheim B-I. Compar- 2. Kienle GS, Kiene H. The powerful placebo effect: Fact or ison of postpartum pain treatments using a sequential trial de- fiction? J Clin Epidemiol 1997;50:1311–1318.
sign I: Paracetamol versus placebo. Eur J Clin Pharmacol 3. Kienle G, Kiene H. Placeboeffekt und Placebokonzept-eine kritische methodologische und konzeptionelle Analyse von 23. Skovlund E, Fyllingen G, Landre H, Nesheim B-I. Compar- Angaben zum Ausmass des Placeboeffekts. Forschende ison of postpartum pain treatments using a sequential trial de- Komplementärmedizin 1996;3:121–138.
sign II: naproxen versus paracetamol. Eur J Clin Pharmacol 4. Kaptchuk TJ. Powerful placebo: The dark side of the ran- domised controlled trial. Lancet 1998;351:1722–1725.
24. Diener HC. Issues in migraine trial design: A case study. The 5. Ernst E, Resch KL. Concept of true and perceived placebo 311C Symposium Presentations given in the 311C90 Sym- posion of the 3rd European Headache Conference (Abstract).
6. Hróbjartsson A. What are the main methodological problems June 5–8, 1996, S Margherita die Pula 1996;10–11.
in the estimation of placebo effects? J Clin Epidemiol 2002; 25. Diener HC, Klein KB, Multinational Oral 311C90 and Suma- triptan Comparative Study Group. The first comparison of the 7. Shapiro AK, Morris LA. The placebo effect in medicine and efficacy and safety of 311C90 and sumatriptan in the treat- psychological therapies. In: Garfield SL, Bergin A, eds.
ment of migraine (Abstract). 3rd European Headache Con- Handbook of Psychotherapy and Behavior Change: An Em- ference. S Margherita di Pula, Italy, June 5–8, 1996.
pirical Analysis, 2nd ed. New York: Wiley, 1978:369–410.
26. Diener H-C, Dowson AJ, Ferrari M, Nappi G, Tfelt-Hansen 8. Illhardt FJ. Plazebo und Ethik. Zeitschrift für Allgemein- P. Unbalanced randomization influences placebo response: Scientific versus ethical issues around the use of placebo in 9. Kaptchuk TJ. Intentional ignorance: A history of blind as- migraine trials. Cephalalgia 1999;19:699–700.
sessment and placebo controls in medicine. Bull History Med 27. de Craen AJM, Moerman DE, Heisterkamp SH, Tytgat GNJ, Tijssen J, Kleijnen J. Placebo effect in the treatment of duo- 10. Walach H, Sadaghiani C. Plazebo und Plazebo-Effekte: Eine denal ulcer. Br J Clin Pharmacol 1999;48:853–860.
Bestandsaufnahme. Psychotherapie, Psychosomatik, medi- 28. Vase L, Robinson ME, Verne GN, Price DD. The contribu- zinische Psychologie 2002;52:332–342.
tions of suggestion, desire, and expectation to placebo effects 11. Moerman DE, Jonas WB. Deconstructing the placebo effect in irritable bowel syndrome patients: An empirical investi- and finding the meaning response. Ann Intern Med 2002;136: 29. Cherkin DC, Eisenberg D, Sherman KJ, et al. Randomized 12. Brody H. The placebo response. Recent research and impli- trial comparing traditional Chinese medical acupuncture, cations for family medicine. J Family Pract 2000;49:649–654.
therapeutic massage, and self-care education for chronic low 13. Uexküll TV. Biosemiotic research and not further molecular back pain. Arch Intern Med 2001;161:1081–1088.
analysis is necessary to describe pathways between cells, per- 30. Kalauokalani D, Cherkin DC, Sherman KJ, Koepsell TD, sonalities, and social systems. Adv J Mind–Body Health Deyo RA. Lessons from a trial of acupuncture and massage for low back pain. Spine 2001;26:1418–1424.
14. Hróbjartsson A, Gotzsche PC. Is the placebo powerless? An 31. Miller J. Going unconscious. In: Silver RB, ed. Hidden His- analysis of clinical trials comparing placebo with no treat- tories of Science, London: Granta Books, 1997:1–35.
ment. N Engl J Med 2001;344:1594–1602.
32. Florey E. Ars Magnetica. Franz Anton Mesmer 1734–1815: 15. Kirsch I. Specifying nonspecifics: Psychological mechanisms Magier vom Bodensee. Konstanz: Universitätsverlag Kon- of placebo effects. In: Harrington A, ed. The Placebo Effect: PLACEBO RESEARCH
33. Gheorghiu VA, Netter P, Eysenck HJ, Rosenthal R, eds. Sug- 52. Levine JD, Gordon NC, Jones RT, Fields HL. The narcotic gestion and Suggestibility: Theory and Research. Heidelberg: antagonist naloxone enhances clinical pain. Nature 1978;272: 34. Berner W. Zu den empirischen Ergebnissen der Neurowis- 53. Kosterlitz HW, Hughes J. Peptides with morphine-like ac- senschaften und ihren Auswirkungen auf die Theorie des tions in the brain. Br J Psychiatry 1977;130:298–304.
Mentalen und die Theorie der Technik der Psychoanalyse. In: 54. Sher L. The placebo effect on mood and behavior: the role Giampieri-Deutsch P, ed. Psychoanalyse im Dialog der Wis- of the endogenous opioid system. Med Hypotheses 1997;48: senschaften. Band 1: Europäische Perspektiven, Stuttgart: 55. Gracely RH, Dubner R, Deeter WR, Wolskee PJ. Clinicians 35. Roth G. Das Gehirn und seine Wirklichkeit. Kognitive Neu- expectation influence placebo analgesia [letter]. Lancet 1985; robiologie und ihre philosophischen Konsequenzen. Frank- 56. Gracely RH, Dubner R, Wolskee PJ, Deeter WR. Placebo 36. Kirsch I, Baker SL. Clinical implications of expectancy re- and naloxone can alter post surgical pain by separate mech- search: Activating placebo effects without deception. Con- 57. Posner J, Burke CA. The effects of naloxone on opiate and 37. Butler C, Steptoe A. Placebo responses: an experimental placebo analgesia in healthy volunteers. Psychopharmacol- study of psychophysiological processes in asthmatic volun- teers. Br J Clin Psychol 1986;25:173–183.
58. Benedetti F. The opposite effects of the opiate antagonist 38. Marlatt GA, Rohsenow DJ. Cognitive processes in alcohol naloxone and the cholecystokinin antagonist proglumide on use: Expectancy and the balanced placebo design. In: Mello placebo analgesia. Pain 1996;64:535–543.
N, ed. Advances in Substance Abuse. Greenwich: JAI Press, 59. ter Riet G, de Craen AJM, de Boer A, Kessels AGH. Is placebo analgesia mediated by endogenous opioids? A sys- 39. Lienert GA. Die Bedeutung der Suggestion in pharmakopsy- tematic review. Pain 1998;76:273–275.
chologischen Unterschungen. Zeitschrift für experimentelle 60. Lichtigfeld FJ, Gillman MA. Possible role of the endogenous und angewandte Psychologie 1955;3:418–438.
opioid system in the placebo response in depression. Int J 40. Hull JG, Bond CF. Social and behavioral consequences of al- Neuropsychopharmacol 2002;5:107–108.
cohol consumption and expectancy: A meta-analysis. Psy- 61. Gillman MA, Lichtigfeld FJ. Randomized double-blind trial of psychotropic analgesic nitrous oxide compared with di- 41. Fillmore MT, Mulvihill LE, Vogel-Sprott M. The expected azepam for alcohol withdrawal state. J Subs Abuse Treat drug and its expected effect interact to determine placebo re- sponses to alcohol and caffeine. Psychopharmacology 1994; 62. Gillman MA, Lichtigfeld FJ. Placebo and analgesic nitrous oxide for treatment of the alcohol withdrawal state. Br J Psy- 42. Kirsch I, Rosadino MJ. Do double-blind studies with in- formed consent yield externally valid results? An empirical 63. Lichtigfeld FJ, Gillman MA. Analgesic nitrous oxide for al- test. Psychopharmacology 1993;110:437–442.
cohol withdrawal is better than placebo. Int J Neurosci 43. Fillmore M, Vogel-Sprott M. Expected effect of caffeine on motor performance predicts the type of response to placebo.
64. Gillman MA, Lichtigfeld FJ. Minimal sedation required with Psychopharmacology 1992;106:209–214.
nitrous oxide-oxygen treatment of the alcohol withdrawal 44. Mikalsen A, Bertelsen B, Flaten MA. Effects of caffeine, caf- state. Br J Psychiatry 1986;148:604–606.
feine-associated stimuli, and caffeine-related information on 65. Stefano GB, Fricchione GL, Slingsby BT, Benson H. The physiological and psychological arousal. Psychopharmacol- placebo effect and relaxation response: Neural processes and their coupling to constitutive nitric oxide. Brain Res Rev 45. Flaten MA, Blumenthal TD. Caffeine-associated stimuli elicit conditioned response: An experimental model of the placebo 66. Ader R. Behavioral conditioning and immunity. In: Fabris N, effect. Psychopharmacology 1999;145:105–112.
Garaci E, Hadden J, Mitchison NA, eds. Immunoregulation.
46. Andrews SE, Blumenthal TD, Flaten MA. Effects of caffeine New York: Plenum Press, 1983:283–313.
and caffein-associated stimuli on the human startle eyeblink 67. Ader R, Cohen N. Behaviorally conditioned immunosup- reflex. Pharmacol Biochem Behav 1998;59:39–44.
pression. Psychosomat Med 1975;37:333–340.
47. Levine JD, Gordon NC. Influence of the method of drug ad- 68. Wickramasekera I. A conditioned response model of ministration on analgesic response. Nature 1984;312:755– the placebo effect: Predictions from the model. In: White L, Tursky B, Schwartz GE, eds. Placebo—Theory, Re- 48. Levine JD, Gordon NC, Smith R, Fields HL. Analgesic re- search, Mechanisms. New York: Guilford Press, 1985:255– sponses to morphine and placebo in individuals with post- operative pain. Pain 1981;10:379–389.
69. Kirsch I. Response expectancy as a determinant of experi- 49. Levine JD, Gordon NC, Bornstein JC, Fields HL. Role of pain ence and behavior. Am Psychol 1985;40:1189–1202.
in placebo analgesia. Proc Nat Acad Sci 1979;76:3528–3531.
70. Voudouris NJ, Peck CL, Coleman G. The role of condition- 50. Levine JD, Gordon NC, Fields HL. Naloxone dose depen- ing and verbal expectancy in the placebo response. Pain 1990; dently produces analgesia and hyperalgia in postoperative 71. Voudouris NJ, Peck CL, Coleman G. Conditioned response 51. Levine JD, Gordon NC, Fields HL. The mechanism of models of placebo phenomena: Further support. Pain 1989; placebo analgesia. Lancet 1978;ii:654–657.
WALACH AND JONAS
72. Voudouris NJ, Peck CL, Coleman G. Conditioned placebo 91. Masek K, Petrovicky P, Sevcik J, Zidek Z, Frankova D. Past, responses. J Person Soc Psychol 1985;48:47–53.
present and future of psychoneuroimmunology. Toxicology 73. Montgomery GH, Kirsch I. Classical conditioning and the placebo effect. Pain 1997;72:107–113.
92. Mössner R, Lesch K-P. Role of serotonin in the immune sys- 74. Montgomery GH, Kirsch I. Mechanisms of placebo pain re- tem and in neuroimmune interactions. Brain Behav Immun duction: An empirical investigation. Psychol Sci 1996;7:174– 93. Petrovic P, Kalso E, Petersson KM, Ingvar M. Placebo and 75. Pollo A, Vighetti S, Rainero I, Benedetti F. Placebo analge- opioid analgesia—Imaging a shared neuronal network. Sci- sia and the heart. Pain 2003;102:125–133.
76. Pollo A, Torre E, Lopiano L, et al. Expectation modulates 94. de la Fuente-Fernández R, Ruth TJ, Sossi V, Schulzer M, the response to subthalamic nucleus stimulation in Parkin- Calne DB, Stoessl AJ. Expectation and dopamine release: sonian patients. Neuroreport 2002;13:1383–1386.
Mechanism of the placebo effect in Parkinson’s disease. Sci- 77. Amanzio M, Pollo A, Maggi G, Benedetti F. Response vari- ability to analgesics: A role for non-specific activation of en- 95. Pitchot W, Reggers J, Pinto E, et al. Reduced dopaminergic dogenous opioids. Pain 2001;90:205–215.
activity in depressed suicides. Psychoneuroendocrinology 78. Pollo A, Amanzio M, Arlsanian A, Casadio C, Maggi G, Benedetti F. Response expectancies in placebo analgesia and 96. Schultz W. The primate basal ganglia and the voluntary con- their clinical relevance. Pain 2001;93:77–84.
trol of behaviour. J Consciousness Stud 1999;6:31–45.
79. Amanzio M, Benedetti F. Neuropharmacological dissection 97. Waelti P, Dickinsen A, Schultz W. Dopamine responses com- of placebo analgesia expectation-activated opioid systems ply with basic assumptions of formal learning theory. Nature versus conditioning-activated specific subsystems. J Neurosci 98. Mayberg HS, Silva JA, Brannan SK, et al. The functional 80. Benedetti F, Arduino C, Amanzio M. Somatotopic activation neuroanatomy of the placebo effect. Am J Psychiatry 2002; of opioid systems by target-directed expectations of analge- sia. J Neurosci 1999;19:3639–3648.
99. Leuchter AF, Cook IA, Witte EA, Morgan M, Abrams M.
81. Benedetti F, Amanzio M, Baldi S, Casadio C, Maggi G. In- Changes in brain function of depressed subjects during treat- ducing placebo respiratory depressant responses in humans ment with placebo. Am J Psychiatry 2002;159:122–129.
via opioid receptors. Eur J Neurosci 1999;11:625–631.
100. Walach H. Das Wirksamkeitsparadox in der Komplemen- 82. Benedetti F, Amanzio M, Baldi S, et al. The specific effects tärmedizin. Forschende Komplementärmedizin und Klassis- of prior opioid exposure on placebo analgesia and placebo che Naturheilkunde 2001;8:193–195.
respiratory depression. Pain 1998;75:313–319.
101. Coronary Drug Project Research Group. Influence of adher- 83. Benedetti F, Amanzio M. The neurobiology of placebo anal- ence to treatment and response of cholesterol on mortality in gesia: From endogenous opioids to cholecystekinin. Prog the Coronary Drug Project. N Engl J Med 1980;303:1038– 84. Benedetti F, Amanzio M, Maggi G. Potentiation of placebo 102. Horwitz RI, Horwitz SM. Adherence to treatment and health analgesia by proglumide. Lancet 1995;346:1231.
outcomes. Arch Intern Med 1993;153:1863–1868.
85. de la Fuente-Fernández R, Schulzer M, Stoessl AJ. The 103. Frank JD. Non-specific aspects of treatment: The view of a placebo effect in neurological disorders. Lancet Neurol 2002; psychotherapist. In: Shepherd M, Sartorius N, eds. Non-Spe- cific Aspects of Treatment. Bern: Huber, 1989:95–114.
86. Rossi EL. A conceptual review of the psychosocial genomics 104. Frank JD. Persuasion and Healing: A Comparative Study of of expectancy and surprise: Neuroscience perspectives about Psychotherapy. Baltimore: Johns Hopkins University Press, the deep psychobiology of therapeutic hypnosis. Am J Clin 105. Thomas KB. General practice consultations: Is there any 87. Blalock JE. The syntax of immune-neuroendocrine commu- point in being positive? Br Med J 1987;294:1200–1202.
nication. Immunol Today 1994;15:504–511.
88. Cabot PJ, Carter L, Schäfer M, Stein C. Methionine- enkephalin dynorphin A-release from immune cells and con- trol of inflammatory pain. Pain 2001;93:207–212.
89. Maestroni GJM, Conti A. Role of the pineal neurohormone Department of Environmental Medicine and melatonin in the psycho-neuroendocrine-immune network.
In: Ader R, Felten DL, Cohen N, eds. Psychoneuroimmu-nology, 2nd ed. San Diego, CA: Academic Press, 1991:495– Samueli Institute for Information Biology 90. Grimaldi B, Bonnin A, Fillion MP, et al. Interaction of the serotonergic and the immune systems: 5HT-Moduline. In: Müller N, ed. Psychiatry, Psychoimmunology, and Viruses.
Wien/New York: Springer, 1999:45–54.
E-mail: [email protected]

Source: http://samueli.stage.bridgelinedigital.net/File%20Library/Knowledge%20Center/Publications/Walach-and-Jonas.pdf