Journal of Psychosomatic Research 78 (2015) 484–488
Contents lists available at ScienceDirect
Journal of Psychosomatic Research
Galvanic Vestibular Stimulation: A new model of placebo-induced nausea V.F. Quinn ⁎, H.G. MacDougall, B. Colagiuri School of Psychology, The University of Sydney, Australia
a r t i c l e
i n f o
Article history: Received 7 July 2014 Received in revised form 17 December 2014 Accepted 20 December 2014 Keywords: Placebo Nausea Conditioning
a b s t r a c t Objective: Traditional rotation-based models of placebo nausea are limited because they do not have vehicle settings and are tied to their context. The present study introduces a new model for examining placebo-induced nausea in the laboratory that overcomes these limitations, namely, Galvanic Vestibular Stimulation (GVS). GVS stimulates the vestibular system to cause nausea through sensory mismatch with visual cues and importantly has a non-nauseating placebo setting. Using this, we tested whether conditioning could elicit placebo nausea when participants were later exposed to placebo stimulation as well as whether this placebo nausea was generalised across contexts — something that is extremely difficult to test with rotation-based models of placebo nausea. Methods: Thirty healthy undergraduate students were randomised to receive either placebo GVS (controls) or active GVS during training (Context-Consistent and Context-Change). On test, all groups received placebo GVS. The controls and Context-Consistent groups were tested in the same context as training, whereas the ContextChange group was tested in a new context. Results: Participants conditioned with nausea during training had significantly higher nausea symptom ratings after placebo stimulation on test than those given no conditioning. This placebo-induced nausea also generalised to a novel test context with no differences observed between the Context-Change and Context-Consistent groups. Conclusion: GVS provides a new model of placebo-induced nausea that overcomes limitations to traditional rotation-based paradigms. Future studies should use this device to explore the effect of instructions and conditioning on the development of placebo nausea and to assess the efficacy of conditioning-based interventions for clinical use. © 2015 Elsevier Inc. All rights reserved.
Introduction Nausea is problematic in many clinical settings, including postoperative care and chemotherapy. Our understanding of the multiple pathways through which nausea develops has led to a growing appreciation for the role of psychological processes [1]. The placebo effect may be one contributor, with multiple clinical demonstrations providing evidence that a patient's expectation can predict the severity of nausea experienced (e.g., chemotherapy, [2], postoperative care, [3]). A number of laboratory studies have attempted to model this in order to better understand how both positive and negative expectations contribute to nausea. To date, these studies have produced mixed findings (see [4] for a detailed review). For example, two found that verbal instructions modulate nausea in the direction of the suggestion ([5,6]: Experiment 2), while two others found that the suggestion of increased nausea actually reduced nausea [7,8]. Similarly, one study found that a conditioning ⁎ Corresponding author at: School of Psychology, A18, The University of Sydney, NSW 2006, Australia. Tel.: +61 2 9036 7265. E-mail address: [email protected] (V.F. Quinn).
http://dx.doi.org/10.1016/j.jpsychores.2014.12.011 0022-3999/© 2015 Elsevier Inc. All rights reserved.
manipulation affected anticipatory, but not reactive nausea [9], while another found that both were affected, but only in women ([6]: Experiment 1). One limitation to current laboratory models of placebo-induced nausea is that they rely on rotation to induce nausea, e.g., rotation chairs or optokinetic drums. Such devices have a number of potential problems. First, highly expectant participants can introduce compensatory behaviours, such as, eye-closing (or opening depending on the type of device) or reducing mandatory head movements to counteract the nausea, which may interfere with the target manipulation and could explain null or reverse manipulation effects [7,8]. Second, in contrast to placebo treatments such as pills or injections, rotation devices do not have placebo settings, meaning that any attempt to stop or surreptitiously reduce rotation is limited by what participants can discern. Third, rotation devices render contextual manipulations difficult, as they are nonportable and often dominate the context in which they are placed. This is in contrast to traditional animal models of conditioned nausea in which a nauseating agent can be delivered in one context and the vehicle delivered in another (e.g., [10]), a process not replicable with rotation. Without being able to understand the way in which placebo
V.F. Quinn et al. / Journal of Psychosomatic Research 78 (2015) 484–488
processes may alter the nauseous response in the laboratory, our capacity to develop methods to intervene clinically is reduced. It is therefore important to develop laboratory paradigms in which the experience of placebo nausea can be examined systematically. The present study, therefore, aimed to develop and test a new model for examining placebo-induced nausea, namely Galvanic Vestibular Stimulation (GVS). GVS involves passing small electrical currents to the vestibular end organs via surface electrodes over the mastoid processes that alter the firing rate of vestibular afferents, often used to explore functions of the vestibular system [11]. It has been reported that one consequence of extended bipolar GVS for some individuals can be feelings of motion sickness, potentially as the individual experiences a sensory mismatch between visual motion cues and the perception of movement that the GVS causes, likened to a rocking boat [12,13]. These illusions of movement during bipolar GVS are frequency specific and detected around the yaw and roll axes [14]. Devices used to generate GVS currents are usually small and portable, and are attached to the individual via cutaneous electrodes that are arguably a more ecologically valid imitation of the clinical setting in which therapeutic agents that cause nausea are often infused, such as chemotherapy. The primary reason that the GVS may offer an important alternative to rotation chairs or optokinetic drums is that binaural GVS enables the researcher to either send bipolar (asymmetrical) or monopolar (symmetrical) electrical signals that stimulate the entire labyrinth of both the left and right vestibular systems. In bipolar stimulation the vestibular signal is modulated as the firing rates are increased on the cathodal side and decreased on the anodal side [15]. Using a pseudorandom sinusoidal current with peak amplitude of ±5 mA (see Fig. 1) the two terminals of the GVS unit can be positive or negative. This means that rather than having one ear receiving the anode current and one the cathode, throughout stimulation the vestibular end organs that receive the anode or the cathode continuously switch. This effectively doubles the range of stimuli without exceeding the 5 mA limit. This vestibular input, which leads to perceptions of sideways movements and rolling [14], is in conflict with the constant visual input in the individual's environment, leading to motion sickness [13]. In contrast, during monopolar stimulation the spontaneous firing rate of vestibular afferents is still increased and decreased, but this occurs symmetrically. Since the central nervous system largely interprets movement through the difference in firing rate between afferents in the vestibular end organs [11], symmetrical stimulation should not lead to the same experience of motion, as there is comparatively little effect on the perception of motion and reflexive eye movements, and hence nausea [16,17]. However during both bipolar and monopolar stimulation, there is skin prickling and a metallic taste experienced, making the two virtually indistinguishable. This device is therefore unique in that in can be used both to induce acute nausea safely in healthy participants (through
485
bipolar stimulation) and can also be used to produce a non-nauseating placebo-like stimulation (through monopolar stimulation). Using this new model, we tested whether individuals led to expect nausea through conditioning would experience more nausea when later exposed to vehicle (ie. placebo) stimulation, compared with those receiving no conditioning. To extend previous findings suggesting such effects and demonstrate the versatility of GVS, we also tested the extent to which any such placebo-induced nausea would generalise to a new context, by including a Context-Change group who also received conditioning but who were placed in a different context on the test day to that which they had been trained in. To our knowledge, this is the first study to use GVS to explore placebo-induced nausea and its generalizability across contexts. Method Participants Participants were 30 undergraduates from the University of Sydney, who were reimbursed AUD$50 for their time (approx. $20/hour). They had an average age of 23.3 (SD = 4.70) and 16 were female. Design Table 1 illustrates the three-group between-subject design. The type of stimulation was varied during training such that controls received binaural monopolar (non-nauseating, i.e., vehicle) stimulation and the two experimental groups received binaural bipolar (nauseating, i.e., active) stimulation. On test, all groups received vehicle GVS. The controls and Context-Consistent groups were tested in the same context as in training, whereas the Context-Change group was tested in a novel context. The dependent variable was nausea symptom ratings. Apparatus GVS was delivered by a small, battery powered constant-current generator that was pre-programmed to deliver the signal depicted in Fig. 1, identical to the stimulus used by MacDougall, Moore, Curthoys and Black [18] who used GVS to investigate postural instability. The electrodes used were three 10 cm2 grounded plate electrodes attached to adhesive pads, which were additionally treated with a thick layer of electroconductive gel to ensure good skin contact. One of these was placed across the mastoid bone behind each ear and then secured with foam pads and an elastic strap. The third pad was placed on the participant's back centered at about the height of the c7 vertebra. Although this third electrode placement is only necessary during monopolar stimulation, this was performed on all participants so that those in the experimental groups could not notice a change from training to test. During bipolar stimulation, the third electrode transmitted the same signal as the right mastoid process received. The terminal setup to deliver bipolar and monopolar stimulation using the same device, electrode configuration, and signal is shown in Fig. 2. The experimental contexts were a brightly coloured and well lit medical consultation room and a small, dark testing cubicle with all walls and ceiling painted black located in a different building in thecampus. Procedure
Fig. 1. Diagram depicting a 10 second example of the galvanic stimulus used: a pseudorandom current that was a modified sum-of-sines signal with clippings and random weights assigned, with maximum peak amplitude of 5 mA. During monopolar stimulation both labyrinths receive this wave, whereas during bipolar stimulation one labyrinth receives this wave, and the other receives the additive inverse.
The study was conducted on three non-consecutive days, scheduled to occur within a fortnight and at the same time of the day for each participant. Participants were given a cover story that the study was assessing the effects of experiencing simulated motion on their spatial awareness, which would be assessed through several spatial and cognitive tasks, but for ethical reasons were informed that the device may temporarily cause low levels of nausea. They were told that symptom ratings were being taken as the researchers needed a detailed record
486
V.F. Quinn et al. / Journal of Psychosomatic Research 78 (2015) 484–488
Table 1 Experimental design. Group
Controls (n = 10) Context-Consistent (n = 10) Context-Change (n = 10)
Training
Test
Day 1
Day 2
Day 3
Context A Vehicle Context A Active Context B Active
Context A Vehicle Context A Active Context B Active
Context A Vehicle Context A Vehicle Context A Vehicle
Note. The assignment of contexts as A or B was counterbalanced across participants.
of any side effects of stimulation experienced to monitor their tolerance and that it was important for their own safety that ratings were as accurate as possible. Nausea ratings were based on Klosterhalfen et al.'s scale [19] and involved ratings of nausea-related symptoms (urge to vomit, stomach awareness, nausea, headache, fatigue, and dizziness) on a visual analogue scale from 0 ‘not at all’ to 10 ‘severe’, with two bogus side effects added to divert attention away from the focus on nausea. After providing written consent and filling out a demographics questionnaire, participants were asked to make baseline nausea ratings. The skin over the mastoid processes was then cleaned and checked for any minor abrasions, and the electrodes were attached. The participants then received 25 min of stimulation, with the experimental groups receiving active GVS and the control group instead receiving vehicle stimulation. During this time they were asked to conduct a series of spatial tasks designed both to enhance the experience of motion mismatch, as well as uphold the cover story. For each of the training sessions this involved four questions taken from a Raven's Progressive Matrices inventory, a task in which participants had to copy a complex shape from memory, a computerised digit substitution task, as well as a balancing and ball-catch task. After this, the device was switched off and participants were asked to make post-stimulation ratings. This procedure was replicated on the second training day. On the test day, the procedure was identical to training except that all participants received vehicle stimulation and the Context-Change group was tested in the alternate room to the one in which their training was given. The only other difference was that the balance and ball-catch tasks were omitted, as pilot testing had revealed that these allowed participants in the bipolar stimulation conditions to determine there had been a change in stimulation. Finally, participants completed a manipulation check that asked them what they thought the aim of the study was and were then fully debriefed. The project received approval from the University of Sydney Human Research Ethics Committee. Statistical analysis As not all individuals receiving the same GVS develop the same levels of nausea [12], it was of interest to determine the rate of response to GVS with nausea across the active and vehicle stimulation, simply to confirm that active GVS induces nausea and vehicle does not. A nonresponder was classified as someone who reported on average less
Fig. 2. Diagram showing the electrode and terminal set up for the bipolar, or active, stimulation (left) and monopolar, or vehicle, stimulation (right).
than or equal to 6 points more on the symptom rating scale after stimulation than before. In order to account for pre-existing symptoms before stimulation symptom experience, we calculated difference scores between poststimulation and pre-stimulation symptom ratings as the primary dependent variable. For this reason we carried out between-subjects oneway ANOVAs on the average baseline ratings during both training and on test to ensure that no group differences existed before stimulation. A one-way between-subjects ANOVA was then used to determine the efficacy of the GVS to elicit nausea during training. This was achieved by computing a training average, the mean of the difference ratings across the two training days, to use as the dependent variable. For the test scores, a one way between-subjects ANOVA was carried out on the test day difference ratings. All between-subject ANOVAs were followed up with planned orthogonal group contrasts that first compared the control group to the two experimental groups combined – to test for placebo-induced nausea overall, and then compared the two experimental groups to each to other – to test the effect of context on the strength of placebo-induced nausea. Analysis was conducted using SPSS (V20) and results were considered statistically significant when p b .05. Results GVS and nausea In the two experimental groups, there were two participants in the ContextConsistent group, and one in the Context-Change group who were non-responders during training. This means that active GVS successfully induced substantial nausea in the vast majority (85%) of the experimental participants during training. The demographic characteristics of these three non-responders to active GVS did not differ appreciably from the other participants. Importantly, in the control group, all ten participants were classified as non-responders, indicating that without any conditioning, vehicle stimulation does not produce nausea. Thus, active GVS effectively induced nausea and vehicle GVS did not (unconditionally) induce nauea. This was also evident in the nausea symptom ratings during training. An analysis of baseline ratings during training found that there were no differences between groups F(2, 27) = .38, p = .69, and so difference scores were computed and used in the primary analysis. As shown in Fig. 3, averaged across the two training days, the two experimental groups rated nausea following stimulation significantly higher by an average of 16.2 (SD = 16.1) points than controls relative to baseline, F(1, 27) = 27.4, p b .001, with no difference between the two experimental groups, F(1, 27) = 1.14, p = .30. Placebo nausea Analysis of baseline ratings on the test day demonstrated that there were again no group differences before stimulation, F(2, 27) = .07, p = .93, and so difference scores were calculated. As shown in Fig. 4, on the test day when all groups received vehicle stimulation, the two experimental groups demonstrated statistically significant placeboinduced placebo, rating their nausea an average of 6.7 (SD = 11.1) points higher than the control group, F(1, 27) = 6.97, p = .01. The magnitude of this placebo-induced nausea did not differ significantly between the two experimental groups, F(1, 27) = 0.07, p = .79. In the manipulation check, no participant guessed the true aims of the experiment or mentioned nausea as a dependent variable.
Discussion Using a new model to induce nausea, namely Galvanic Vestibular Stimulation, we found clear evidence of placebo-induced nausea. Participants conditioned to expect nausea during training had significantly higher nausea symptom ratings to vehicle stimulation on test than those given no conditioning. This demonstration provides evidence that through conditioning, in combination with initial instructions, GVS can be used to elicit placebo nausea. This effect is consistent with previous findings that suggest placebo-induced nausea can be experimentally established via instruction and conditioning [5,6]. Extending previous research, we found that the placebo-induced nausea generalised perfectly well to a new test context with no difference in placebo-induced nausea on test between the Context-Change and Context-Consistent groups. Our understanding of human conditioned nausea is largely informed in the laboratory by animal models
V.F. Quinn et al. / Journal of Psychosomatic Research 78 (2015) 484–488
487
Fig. 3. Mean (+SE) nausea ratings both before and after stimulation across the three groups, for the average of the two training days.
of context-nausea learning (e.g., [20,21]). However there are obvious differences between animal learning paradigms, where motivation to use information from the context is high, compared to in human learning where there is explicit knowledge that the nausea-inducing device, i.e., the unconditioned stimulus, is responsible for the nausea experienced. The strong generalisation of the placebo nausea to the novel context observed here suggests that the device overshadowed conditioning to the context. This has important clinical implications as similar processes may be at play in the medical context, if a drug injection is more salient than the context in which it is injected. Importantly, this perfect generalisation of placebo-induced nausea was only possible to detect because the GVS can be applied across multiple contexts, unlike rotation chairs and optokinetic drums. The overall pattern of results demonstrates that the GVS offers researchers an important method of examining the development of placebo-induced nausea using a portable and highly controlled stimulus, which may also offer a more ecologically valid simulation of the treatment context. As well as sending vehicle stimulation, the GVS device can be used to simulate a surreptitious reduction procedure commonly utilised in pain research (e.g., [22,23]) through a dial that varies the amplitude of the signal from ±0 mA to ± 5 mA. The combination of these features means that GVS is uniquely placed to answer several questions regarding the role of instruction and conditioning in placebo nausea and may be used in future studies to shed light on conflicting findings present in the literature [4]. Not only is this current
Fig. 4. Mean (+SE) nausea ratings both before and after stimulation across the three groups on test, when all participants received vehicle (ie. placebo) stimulation.
experimental set up ideal for use in placebo nausea research, but also for the understanding of other human conditioning processes which are mediated via nausea, such as conditioned taste aversion. Although we observed perfect generalisation to the test context in the Context-Change group, this does not mean that humans cannot exhibit contextually conditioned nausea. A more sensitive discrimination that assesses the context specificity of conditioned nausea in humans could involve using two different experimenters in two different rooms, one in which bipolar stimulation is always delivered and one in which monopolar stimulation is always delivered. In such a design, the device itself would not predict nausea, but the context would. This would reduce any overshadowing by the device itself and therefore provide a test of whether any nausea can be conditioned to the context at all. It is also possible that overlapping features of the two contexts may have facilitated generalisation rather than purely the device overshadowing the context, as both contexts had the same experimenter and were located in the same university with basic structural features in common. Experimenting with more salient context changes could also be one avenue through which future studies can shed light on the context specificity of human nausea learning. It should also be noted that monopolar GVS has been found to lead to very small postural sways, potentially due to a small net pitch effect [24]. However, since our participants were seated, it is unlikely that any small net pitch effect would induce even minor postural sways and certainly none sufficient to produce nausea. This is supported by none of the participants in the control group being classified as nausea responders. However, one at least theoretical confound, would be that the experimental groups may have experienced sensitisation to GVS across the two bipolar GVS training days, which could lead to differential responsiveness to monopolar GVS on test, compared with controls who never experienced bipolar GVS. This appears unlikely, as during training participants reported weaker nausea in response to bipolar GVS on the second training day than the first (data not shown), suggesting that, if anything, there was habituation to GVS. One other potential limitation of the present study was the reliance on self-report assessments of nausea, as nauseating stimulation during training could have caused a response shift in participants in the two experimental groups. Although demand characteristics are also possible, the failure of any participants to discern the true aims of the experiment may mean that this cover story offers a suitable method of examining placebo nausea in the absence of perceived demands. The problems inherent with using self-report assessments could be resolved in future designs using the GVS by including a behavioural assessment such as stimulation tolerance, or physiological assessments such as electrogastrogram. However, these types of outcomes are not without their own limitations [4,25].
488
V.F. Quinn et al. / Journal of Psychosomatic Research 78 (2015) 484–488
In summary, we found compelling evidence that GVS can be used to examine placebo related nausea that resolves methodological limitations of rotation chairs and optokinetic drums. Further, the current study demonstrates perfect generalizability of placebo-induced nausea across contexts, suggesting the nauseating agent or device may be the most important cue in producing placebo-induced nausea. This may be particularly important when trying to reduce nausea in the clinic. Acknowledgments We would like to thank Luana Colloca for her insights into the manuscript. This project was supported by an Australian Post Graduate Award. References [1] Sanger GJ, Andrews PLR. Treatment of nausea and vomiting: gaps in our knowledge. Auton Neurosci 2006;129:3–16. http://dx.doi.org/10.1016/j.autneu.2006.07.009. [2] Colagiuri B, Zachariae R. Patient expectancy and post-chemotherapy nausea: a metaanalysis. Ann Behav Med 2010;40:3–14. http://dx.doi.org/10.1007/s12160-0109186-4. [3] Montgomery GH, Schnur JB, Erblich J, Diefenbach MA, Bovbjerg DH. Pre-surgery psychological factors predict pain, nausea and fatigue one week following breast cancer surgery. J Pain Symptom Manage 2010;39:1043–52. http://dx.doi.org/10.1016/j. jpainsymman.2009.11.318. [4] Quinn VF, Colagiuri B. Placebo Interventions for Nausea: a Systematic Review. Ann Behav Med 2014;1–14. http://dx.doi.org/10.1007/s12160-014-9670-3. [5] Horing B, Weimer K, Schrade D, Muth ER, Scisco JL, Enck P, et al. Reduction of motion sickness with an enhanced placebo instruction: an experimental study with healthy participants. Psychosom Med 2013;75:497–504. http://dx.doi.org/10.1097/PSY. 0b013e3182915ee7. [6] Klosterhalfen S, Kellermann S, Braun S, Kowalski A, Schrauth M, Zipfel S, et al. Gender and the nocebo response following conditioning and expectancy. J Psychosom Res 2009;66:323–8. http://dx.doi.org/10.1016/j.jpsychores.2008.09.019. [7] Levine ME, Stern RM, Koch KL. The effects of manipulating expectations through placebo and nocebo administration on gastic tachyarrhythmia and motioninduced nausea. Psychosom Med 2006;68:478–86. http://dx.doi.org/10.1097/01. psy.0000221377.52036.50. [8] Williamson MJ, Thomas MJ, Stern RM. The contribution of expectations to motion sickness symptoms and gastric activity. J Psychosom Res 2004;56:721–6. http://dx.doi.org/10.1016/S0022-3999(03)00130-2. [9] Klosterhalfen S, Kellermann S, Stockhorst U, Wolf J, Kirschbaum C, Hall G, et al. Latent inhibition of rotation chair-induced nausea in healthy male and female
[10]
[11] [12]
[13]
[14]
[15] [16]
[17]
[18]
[19]
[20] [21]
[22] [23] [24] [25]
volunteers. Psychosom Med 2005;67:335–40. http://dx.doi.org/10.1097/01.psy. 0000156930.00201.e0. Boakes RA, Westbrook RF, Elliot M, Swinbourne AL. Context dependency of conditioned aversions to water and sweet tastes. J Exp Psychol Anim Behav Process 1997;23:56–67. http://dx.doi.org/10.1037/0097-7403.23.1.56. Fitzpatrick RC, Day BL. Probing the human vestibular system with galvanic stimulation. J Appl Physiol 2004;96:2301–16. Grewal T, James C, Macefield VG. Frequency-dependent modulation of muscle sympathetic nerve activity by sinusoidal galvanic vestibular stimulation in human subjects. Exp Brain Res 2009;197:379–86. http://dx.doi.org/10.1007/s00221-0091926-y. Hammam E, Dawood T, Macefield VG. Low-frequency galvanic vestibular stimulation evokes two peaks of modulation in skin sympathetic nerve activity. Exp Brain Res 2012;219:441–6. http://dx.doi.org/10.1007/s00221-012-3090-z. Stephan T, Deutschlander A, Nolte A, Schneider E, Wiesmann M, Brandt T, et al. Functional MRI of galvanic vestibular stimulation with alternating currents at different frequencies. Neuroimage 2005;26:721–32. http://dx.doi.org/10.1016/j. neuroimage.2005.02.049. Wardman DL, Day BL, Fitzpatrick RC. Position and velocity responses to galvanic vestibular stimulation in human subjects during standing. J Physiol 2003;547:293–9. Karlberg M, McGarvie L, Magnsson M, Aw ST, Halmagyi GM. The effects of galvanic vestibular stimulation on the human vestibulo-ocular reflex. Audit Vesti Syst 2000; 11:3897–901. Watson SRD, Brizuela AE, Curthoys IS, Colebatch JG, MacDougall HG, Halmagyi GM. Maintained ocular tension produced by bilateral and unilateral galvanic (DC) vestibular stimulation in humans. Exp Brain Res 1998;122:453–8. MacDougall HG, Moore ST, Curthoys IS, Black FO. Modeling postural instability with galvanic vestibular stimulation. Exp Brain Res 2006;172:208–20. http://dx.doi.org/ 10.1007/s00221-005-0329-y. Klosterhalfen S, Ruttgers A, Krumrey E, Otto B, Stockhorst U, Riepl RL, et al. Pavlovian conditioning of taste aversion using a motion sickness paradigm. Psychosom Med 2000;62:671–7 [doi: 0033-3174/00/6205-0671]. Rodriguez M, Lopez M, Symonds M, Hall G. Lithium-induced context aversion in rats as a model of anticipatory nausea in humans. Physiol Behav 2000;71:571–9. Roscoe JA, Morrow GR, Aapro MS, Molassiotis A, Olver I. Anticipatory nausea and vomiting. Support Care Cancer 2011;19:1533–8. http://dx.doi.org/10.1007/s00520010-0980-0. Colloca L, Benedetti F. How prior experience shapes placebo analgesia. Pain 2006; 124:126–33. http://dx.doi.org/10.1016/j.pain.2006.04.005. Au Yeung S, Colagiuri B, Lovibond P, Colloca L. Partial reinforcement, extinction, and placebo analgesia. Pain 2014;155:1110–7. Magnusson M, Johansson R, Wiklund J. Galvanically induced body sway in the anterior–posterior plane. Acta Otolaryngol 1990;110:11–7. Colagiuri B, Lovibond P. Psychological processes that can bias responses to placebo treatment for pain. In: Colloca L, Flaten M, Meissner K, editors. Placebo and pain: from bench to bedside. London: Academic Press; 2013. p. 175–82.
Contents lists available at ScienceDirect
Journal of Psychosomatic Research
Galvanic Vestibular Stimulation: A new model of placebo-induced nausea V.F. Quinn ⁎, H.G. MacDougall, B. Colagiuri School of Psychology, The University of Sydney, Australia
a r t i c l e
i n f o
Article history: Received 7 July 2014 Received in revised form 17 December 2014 Accepted 20 December 2014 Keywords: Placebo Nausea Conditioning
a b s t r a c t Objective: Traditional rotation-based models of placebo nausea are limited because they do not have vehicle settings and are tied to their context. The present study introduces a new model for examining placebo-induced nausea in the laboratory that overcomes these limitations, namely, Galvanic Vestibular Stimulation (GVS). GVS stimulates the vestibular system to cause nausea through sensory mismatch with visual cues and importantly has a non-nauseating placebo setting. Using this, we tested whether conditioning could elicit placebo nausea when participants were later exposed to placebo stimulation as well as whether this placebo nausea was generalised across contexts — something that is extremely difficult to test with rotation-based models of placebo nausea. Methods: Thirty healthy undergraduate students were randomised to receive either placebo GVS (controls) or active GVS during training (Context-Consistent and Context-Change). On test, all groups received placebo GVS. The controls and Context-Consistent groups were tested in the same context as training, whereas the ContextChange group was tested in a new context. Results: Participants conditioned with nausea during training had significantly higher nausea symptom ratings after placebo stimulation on test than those given no conditioning. This placebo-induced nausea also generalised to a novel test context with no differences observed between the Context-Change and Context-Consistent groups. Conclusion: GVS provides a new model of placebo-induced nausea that overcomes limitations to traditional rotation-based paradigms. Future studies should use this device to explore the effect of instructions and conditioning on the development of placebo nausea and to assess the efficacy of conditioning-based interventions for clinical use. © 2015 Elsevier Inc. All rights reserved.
Introduction Nausea is problematic in many clinical settings, including postoperative care and chemotherapy. Our understanding of the multiple pathways through which nausea develops has led to a growing appreciation for the role of psychological processes [1]. The placebo effect may be one contributor, with multiple clinical demonstrations providing evidence that a patient's expectation can predict the severity of nausea experienced (e.g., chemotherapy, [2], postoperative care, [3]). A number of laboratory studies have attempted to model this in order to better understand how both positive and negative expectations contribute to nausea. To date, these studies have produced mixed findings (see [4] for a detailed review). For example, two found that verbal instructions modulate nausea in the direction of the suggestion ([5,6]: Experiment 2), while two others found that the suggestion of increased nausea actually reduced nausea [7,8]. Similarly, one study found that a conditioning ⁎ Corresponding author at: School of Psychology, A18, The University of Sydney, NSW 2006, Australia. Tel.: +61 2 9036 7265. E-mail address: [email protected] (V.F. Quinn).
http://dx.doi.org/10.1016/j.jpsychores.2014.12.011 0022-3999/© 2015 Elsevier Inc. All rights reserved.
manipulation affected anticipatory, but not reactive nausea [9], while another found that both were affected, but only in women ([6]: Experiment 1). One limitation to current laboratory models of placebo-induced nausea is that they rely on rotation to induce nausea, e.g., rotation chairs or optokinetic drums. Such devices have a number of potential problems. First, highly expectant participants can introduce compensatory behaviours, such as, eye-closing (or opening depending on the type of device) or reducing mandatory head movements to counteract the nausea, which may interfere with the target manipulation and could explain null or reverse manipulation effects [7,8]. Second, in contrast to placebo treatments such as pills or injections, rotation devices do not have placebo settings, meaning that any attempt to stop or surreptitiously reduce rotation is limited by what participants can discern. Third, rotation devices render contextual manipulations difficult, as they are nonportable and often dominate the context in which they are placed. This is in contrast to traditional animal models of conditioned nausea in which a nauseating agent can be delivered in one context and the vehicle delivered in another (e.g., [10]), a process not replicable with rotation. Without being able to understand the way in which placebo
V.F. Quinn et al. / Journal of Psychosomatic Research 78 (2015) 484–488
processes may alter the nauseous response in the laboratory, our capacity to develop methods to intervene clinically is reduced. It is therefore important to develop laboratory paradigms in which the experience of placebo nausea can be examined systematically. The present study, therefore, aimed to develop and test a new model for examining placebo-induced nausea, namely Galvanic Vestibular Stimulation (GVS). GVS involves passing small electrical currents to the vestibular end organs via surface electrodes over the mastoid processes that alter the firing rate of vestibular afferents, often used to explore functions of the vestibular system [11]. It has been reported that one consequence of extended bipolar GVS for some individuals can be feelings of motion sickness, potentially as the individual experiences a sensory mismatch between visual motion cues and the perception of movement that the GVS causes, likened to a rocking boat [12,13]. These illusions of movement during bipolar GVS are frequency specific and detected around the yaw and roll axes [14]. Devices used to generate GVS currents are usually small and portable, and are attached to the individual via cutaneous electrodes that are arguably a more ecologically valid imitation of the clinical setting in which therapeutic agents that cause nausea are often infused, such as chemotherapy. The primary reason that the GVS may offer an important alternative to rotation chairs or optokinetic drums is that binaural GVS enables the researcher to either send bipolar (asymmetrical) or monopolar (symmetrical) electrical signals that stimulate the entire labyrinth of both the left and right vestibular systems. In bipolar stimulation the vestibular signal is modulated as the firing rates are increased on the cathodal side and decreased on the anodal side [15]. Using a pseudorandom sinusoidal current with peak amplitude of ±5 mA (see Fig. 1) the two terminals of the GVS unit can be positive or negative. This means that rather than having one ear receiving the anode current and one the cathode, throughout stimulation the vestibular end organs that receive the anode or the cathode continuously switch. This effectively doubles the range of stimuli without exceeding the 5 mA limit. This vestibular input, which leads to perceptions of sideways movements and rolling [14], is in conflict with the constant visual input in the individual's environment, leading to motion sickness [13]. In contrast, during monopolar stimulation the spontaneous firing rate of vestibular afferents is still increased and decreased, but this occurs symmetrically. Since the central nervous system largely interprets movement through the difference in firing rate between afferents in the vestibular end organs [11], symmetrical stimulation should not lead to the same experience of motion, as there is comparatively little effect on the perception of motion and reflexive eye movements, and hence nausea [16,17]. However during both bipolar and monopolar stimulation, there is skin prickling and a metallic taste experienced, making the two virtually indistinguishable. This device is therefore unique in that in can be used both to induce acute nausea safely in healthy participants (through
485
bipolar stimulation) and can also be used to produce a non-nauseating placebo-like stimulation (through monopolar stimulation). Using this new model, we tested whether individuals led to expect nausea through conditioning would experience more nausea when later exposed to vehicle (ie. placebo) stimulation, compared with those receiving no conditioning. To extend previous findings suggesting such effects and demonstrate the versatility of GVS, we also tested the extent to which any such placebo-induced nausea would generalise to a new context, by including a Context-Change group who also received conditioning but who were placed in a different context on the test day to that which they had been trained in. To our knowledge, this is the first study to use GVS to explore placebo-induced nausea and its generalizability across contexts. Method Participants Participants were 30 undergraduates from the University of Sydney, who were reimbursed AUD$50 for their time (approx. $20/hour). They had an average age of 23.3 (SD = 4.70) and 16 were female. Design Table 1 illustrates the three-group between-subject design. The type of stimulation was varied during training such that controls received binaural monopolar (non-nauseating, i.e., vehicle) stimulation and the two experimental groups received binaural bipolar (nauseating, i.e., active) stimulation. On test, all groups received vehicle GVS. The controls and Context-Consistent groups were tested in the same context as in training, whereas the Context-Change group was tested in a novel context. The dependent variable was nausea symptom ratings. Apparatus GVS was delivered by a small, battery powered constant-current generator that was pre-programmed to deliver the signal depicted in Fig. 1, identical to the stimulus used by MacDougall, Moore, Curthoys and Black [18] who used GVS to investigate postural instability. The electrodes used were three 10 cm2 grounded plate electrodes attached to adhesive pads, which were additionally treated with a thick layer of electroconductive gel to ensure good skin contact. One of these was placed across the mastoid bone behind each ear and then secured with foam pads and an elastic strap. The third pad was placed on the participant's back centered at about the height of the c7 vertebra. Although this third electrode placement is only necessary during monopolar stimulation, this was performed on all participants so that those in the experimental groups could not notice a change from training to test. During bipolar stimulation, the third electrode transmitted the same signal as the right mastoid process received. The terminal setup to deliver bipolar and monopolar stimulation using the same device, electrode configuration, and signal is shown in Fig. 2. The experimental contexts were a brightly coloured and well lit medical consultation room and a small, dark testing cubicle with all walls and ceiling painted black located in a different building in thecampus. Procedure
Fig. 1. Diagram depicting a 10 second example of the galvanic stimulus used: a pseudorandom current that was a modified sum-of-sines signal with clippings and random weights assigned, with maximum peak amplitude of 5 mA. During monopolar stimulation both labyrinths receive this wave, whereas during bipolar stimulation one labyrinth receives this wave, and the other receives the additive inverse.
The study was conducted on three non-consecutive days, scheduled to occur within a fortnight and at the same time of the day for each participant. Participants were given a cover story that the study was assessing the effects of experiencing simulated motion on their spatial awareness, which would be assessed through several spatial and cognitive tasks, but for ethical reasons were informed that the device may temporarily cause low levels of nausea. They were told that symptom ratings were being taken as the researchers needed a detailed record
486
V.F. Quinn et al. / Journal of Psychosomatic Research 78 (2015) 484–488
Table 1 Experimental design. Group
Controls (n = 10) Context-Consistent (n = 10) Context-Change (n = 10)
Training
Test
Day 1
Day 2
Day 3
Context A Vehicle Context A Active Context B Active
Context A Vehicle Context A Active Context B Active
Context A Vehicle Context A Vehicle Context A Vehicle
Note. The assignment of contexts as A or B was counterbalanced across participants.
of any side effects of stimulation experienced to monitor their tolerance and that it was important for their own safety that ratings were as accurate as possible. Nausea ratings were based on Klosterhalfen et al.'s scale [19] and involved ratings of nausea-related symptoms (urge to vomit, stomach awareness, nausea, headache, fatigue, and dizziness) on a visual analogue scale from 0 ‘not at all’ to 10 ‘severe’, with two bogus side effects added to divert attention away from the focus on nausea. After providing written consent and filling out a demographics questionnaire, participants were asked to make baseline nausea ratings. The skin over the mastoid processes was then cleaned and checked for any minor abrasions, and the electrodes were attached. The participants then received 25 min of stimulation, with the experimental groups receiving active GVS and the control group instead receiving vehicle stimulation. During this time they were asked to conduct a series of spatial tasks designed both to enhance the experience of motion mismatch, as well as uphold the cover story. For each of the training sessions this involved four questions taken from a Raven's Progressive Matrices inventory, a task in which participants had to copy a complex shape from memory, a computerised digit substitution task, as well as a balancing and ball-catch task. After this, the device was switched off and participants were asked to make post-stimulation ratings. This procedure was replicated on the second training day. On the test day, the procedure was identical to training except that all participants received vehicle stimulation and the Context-Change group was tested in the alternate room to the one in which their training was given. The only other difference was that the balance and ball-catch tasks were omitted, as pilot testing had revealed that these allowed participants in the bipolar stimulation conditions to determine there had been a change in stimulation. Finally, participants completed a manipulation check that asked them what they thought the aim of the study was and were then fully debriefed. The project received approval from the University of Sydney Human Research Ethics Committee. Statistical analysis As not all individuals receiving the same GVS develop the same levels of nausea [12], it was of interest to determine the rate of response to GVS with nausea across the active and vehicle stimulation, simply to confirm that active GVS induces nausea and vehicle does not. A nonresponder was classified as someone who reported on average less
Fig. 2. Diagram showing the electrode and terminal set up for the bipolar, or active, stimulation (left) and monopolar, or vehicle, stimulation (right).
than or equal to 6 points more on the symptom rating scale after stimulation than before. In order to account for pre-existing symptoms before stimulation symptom experience, we calculated difference scores between poststimulation and pre-stimulation symptom ratings as the primary dependent variable. For this reason we carried out between-subjects oneway ANOVAs on the average baseline ratings during both training and on test to ensure that no group differences existed before stimulation. A one-way between-subjects ANOVA was then used to determine the efficacy of the GVS to elicit nausea during training. This was achieved by computing a training average, the mean of the difference ratings across the two training days, to use as the dependent variable. For the test scores, a one way between-subjects ANOVA was carried out on the test day difference ratings. All between-subject ANOVAs were followed up with planned orthogonal group contrasts that first compared the control group to the two experimental groups combined – to test for placebo-induced nausea overall, and then compared the two experimental groups to each to other – to test the effect of context on the strength of placebo-induced nausea. Analysis was conducted using SPSS (V20) and results were considered statistically significant when p b .05. Results GVS and nausea In the two experimental groups, there were two participants in the ContextConsistent group, and one in the Context-Change group who were non-responders during training. This means that active GVS successfully induced substantial nausea in the vast majority (85%) of the experimental participants during training. The demographic characteristics of these three non-responders to active GVS did not differ appreciably from the other participants. Importantly, in the control group, all ten participants were classified as non-responders, indicating that without any conditioning, vehicle stimulation does not produce nausea. Thus, active GVS effectively induced nausea and vehicle GVS did not (unconditionally) induce nauea. This was also evident in the nausea symptom ratings during training. An analysis of baseline ratings during training found that there were no differences between groups F(2, 27) = .38, p = .69, and so difference scores were computed and used in the primary analysis. As shown in Fig. 3, averaged across the two training days, the two experimental groups rated nausea following stimulation significantly higher by an average of 16.2 (SD = 16.1) points than controls relative to baseline, F(1, 27) = 27.4, p b .001, with no difference between the two experimental groups, F(1, 27) = 1.14, p = .30. Placebo nausea Analysis of baseline ratings on the test day demonstrated that there were again no group differences before stimulation, F(2, 27) = .07, p = .93, and so difference scores were calculated. As shown in Fig. 4, on the test day when all groups received vehicle stimulation, the two experimental groups demonstrated statistically significant placeboinduced placebo, rating their nausea an average of 6.7 (SD = 11.1) points higher than the control group, F(1, 27) = 6.97, p = .01. The magnitude of this placebo-induced nausea did not differ significantly between the two experimental groups, F(1, 27) = 0.07, p = .79. In the manipulation check, no participant guessed the true aims of the experiment or mentioned nausea as a dependent variable.
Discussion Using a new model to induce nausea, namely Galvanic Vestibular Stimulation, we found clear evidence of placebo-induced nausea. Participants conditioned to expect nausea during training had significantly higher nausea symptom ratings to vehicle stimulation on test than those given no conditioning. This demonstration provides evidence that through conditioning, in combination with initial instructions, GVS can be used to elicit placebo nausea. This effect is consistent with previous findings that suggest placebo-induced nausea can be experimentally established via instruction and conditioning [5,6]. Extending previous research, we found that the placebo-induced nausea generalised perfectly well to a new test context with no difference in placebo-induced nausea on test between the Context-Change and Context-Consistent groups. Our understanding of human conditioned nausea is largely informed in the laboratory by animal models
V.F. Quinn et al. / Journal of Psychosomatic Research 78 (2015) 484–488
487
Fig. 3. Mean (+SE) nausea ratings both before and after stimulation across the three groups, for the average of the two training days.
of context-nausea learning (e.g., [20,21]). However there are obvious differences between animal learning paradigms, where motivation to use information from the context is high, compared to in human learning where there is explicit knowledge that the nausea-inducing device, i.e., the unconditioned stimulus, is responsible for the nausea experienced. The strong generalisation of the placebo nausea to the novel context observed here suggests that the device overshadowed conditioning to the context. This has important clinical implications as similar processes may be at play in the medical context, if a drug injection is more salient than the context in which it is injected. Importantly, this perfect generalisation of placebo-induced nausea was only possible to detect because the GVS can be applied across multiple contexts, unlike rotation chairs and optokinetic drums. The overall pattern of results demonstrates that the GVS offers researchers an important method of examining the development of placebo-induced nausea using a portable and highly controlled stimulus, which may also offer a more ecologically valid simulation of the treatment context. As well as sending vehicle stimulation, the GVS device can be used to simulate a surreptitious reduction procedure commonly utilised in pain research (e.g., [22,23]) through a dial that varies the amplitude of the signal from ±0 mA to ± 5 mA. The combination of these features means that GVS is uniquely placed to answer several questions regarding the role of instruction and conditioning in placebo nausea and may be used in future studies to shed light on conflicting findings present in the literature [4]. Not only is this current
Fig. 4. Mean (+SE) nausea ratings both before and after stimulation across the three groups on test, when all participants received vehicle (ie. placebo) stimulation.
experimental set up ideal for use in placebo nausea research, but also for the understanding of other human conditioning processes which are mediated via nausea, such as conditioned taste aversion. Although we observed perfect generalisation to the test context in the Context-Change group, this does not mean that humans cannot exhibit contextually conditioned nausea. A more sensitive discrimination that assesses the context specificity of conditioned nausea in humans could involve using two different experimenters in two different rooms, one in which bipolar stimulation is always delivered and one in which monopolar stimulation is always delivered. In such a design, the device itself would not predict nausea, but the context would. This would reduce any overshadowing by the device itself and therefore provide a test of whether any nausea can be conditioned to the context at all. It is also possible that overlapping features of the two contexts may have facilitated generalisation rather than purely the device overshadowing the context, as both contexts had the same experimenter and were located in the same university with basic structural features in common. Experimenting with more salient context changes could also be one avenue through which future studies can shed light on the context specificity of human nausea learning. It should also be noted that monopolar GVS has been found to lead to very small postural sways, potentially due to a small net pitch effect [24]. However, since our participants were seated, it is unlikely that any small net pitch effect would induce even minor postural sways and certainly none sufficient to produce nausea. This is supported by none of the participants in the control group being classified as nausea responders. However, one at least theoretical confound, would be that the experimental groups may have experienced sensitisation to GVS across the two bipolar GVS training days, which could lead to differential responsiveness to monopolar GVS on test, compared with controls who never experienced bipolar GVS. This appears unlikely, as during training participants reported weaker nausea in response to bipolar GVS on the second training day than the first (data not shown), suggesting that, if anything, there was habituation to GVS. One other potential limitation of the present study was the reliance on self-report assessments of nausea, as nauseating stimulation during training could have caused a response shift in participants in the two experimental groups. Although demand characteristics are also possible, the failure of any participants to discern the true aims of the experiment may mean that this cover story offers a suitable method of examining placebo nausea in the absence of perceived demands. The problems inherent with using self-report assessments could be resolved in future designs using the GVS by including a behavioural assessment such as stimulation tolerance, or physiological assessments such as electrogastrogram. However, these types of outcomes are not without their own limitations [4,25].
488
V.F. Quinn et al. / Journal of Psychosomatic Research 78 (2015) 484–488
In summary, we found compelling evidence that GVS can be used to examine placebo related nausea that resolves methodological limitations of rotation chairs and optokinetic drums. Further, the current study demonstrates perfect generalizability of placebo-induced nausea across contexts, suggesting the nauseating agent or device may be the most important cue in producing placebo-induced nausea. This may be particularly important when trying to reduce nausea in the clinic. Acknowledgments We would like to thank Luana Colloca for her insights into the manuscript. This project was supported by an Australian Post Graduate Award. References [1] Sanger GJ, Andrews PLR. Treatment of nausea and vomiting: gaps in our knowledge. Auton Neurosci 2006;129:3–16. http://dx.doi.org/10.1016/j.autneu.2006.07.009. [2] Colagiuri B, Zachariae R. Patient expectancy and post-chemotherapy nausea: a metaanalysis. Ann Behav Med 2010;40:3–14. http://dx.doi.org/10.1007/s12160-0109186-4. [3] Montgomery GH, Schnur JB, Erblich J, Diefenbach MA, Bovbjerg DH. Pre-surgery psychological factors predict pain, nausea and fatigue one week following breast cancer surgery. J Pain Symptom Manage 2010;39:1043–52. http://dx.doi.org/10.1016/j. jpainsymman.2009.11.318. [4] Quinn VF, Colagiuri B. Placebo Interventions for Nausea: a Systematic Review. Ann Behav Med 2014;1–14. http://dx.doi.org/10.1007/s12160-014-9670-3. [5] Horing B, Weimer K, Schrade D, Muth ER, Scisco JL, Enck P, et al. Reduction of motion sickness with an enhanced placebo instruction: an experimental study with healthy participants. Psychosom Med 2013;75:497–504. http://dx.doi.org/10.1097/PSY. 0b013e3182915ee7. [6] Klosterhalfen S, Kellermann S, Braun S, Kowalski A, Schrauth M, Zipfel S, et al. Gender and the nocebo response following conditioning and expectancy. J Psychosom Res 2009;66:323–8. http://dx.doi.org/10.1016/j.jpsychores.2008.09.019. [7] Levine ME, Stern RM, Koch KL. The effects of manipulating expectations through placebo and nocebo administration on gastic tachyarrhythmia and motioninduced nausea. Psychosom Med 2006;68:478–86. http://dx.doi.org/10.1097/01. psy.0000221377.52036.50. [8] Williamson MJ, Thomas MJ, Stern RM. The contribution of expectations to motion sickness symptoms and gastric activity. J Psychosom Res 2004;56:721–6. http://dx.doi.org/10.1016/S0022-3999(03)00130-2. [9] Klosterhalfen S, Kellermann S, Stockhorst U, Wolf J, Kirschbaum C, Hall G, et al. Latent inhibition of rotation chair-induced nausea in healthy male and female
[10]
[11] [12]
[13]
[14]
[15] [16]
[17]
[18]
[19]
[20] [21]
[22] [23] [24] [25]
volunteers. Psychosom Med 2005;67:335–40. http://dx.doi.org/10.1097/01.psy. 0000156930.00201.e0. Boakes RA, Westbrook RF, Elliot M, Swinbourne AL. Context dependency of conditioned aversions to water and sweet tastes. J Exp Psychol Anim Behav Process 1997;23:56–67. http://dx.doi.org/10.1037/0097-7403.23.1.56. Fitzpatrick RC, Day BL. Probing the human vestibular system with galvanic stimulation. J Appl Physiol 2004;96:2301–16. Grewal T, James C, Macefield VG. Frequency-dependent modulation of muscle sympathetic nerve activity by sinusoidal galvanic vestibular stimulation in human subjects. Exp Brain Res 2009;197:379–86. http://dx.doi.org/10.1007/s00221-0091926-y. Hammam E, Dawood T, Macefield VG. Low-frequency galvanic vestibular stimulation evokes two peaks of modulation in skin sympathetic nerve activity. Exp Brain Res 2012;219:441–6. http://dx.doi.org/10.1007/s00221-012-3090-z. Stephan T, Deutschlander A, Nolte A, Schneider E, Wiesmann M, Brandt T, et al. Functional MRI of galvanic vestibular stimulation with alternating currents at different frequencies. Neuroimage 2005;26:721–32. http://dx.doi.org/10.1016/j. neuroimage.2005.02.049. Wardman DL, Day BL, Fitzpatrick RC. Position and velocity responses to galvanic vestibular stimulation in human subjects during standing. J Physiol 2003;547:293–9. Karlberg M, McGarvie L, Magnsson M, Aw ST, Halmagyi GM. The effects of galvanic vestibular stimulation on the human vestibulo-ocular reflex. Audit Vesti Syst 2000; 11:3897–901. Watson SRD, Brizuela AE, Curthoys IS, Colebatch JG, MacDougall HG, Halmagyi GM. Maintained ocular tension produced by bilateral and unilateral galvanic (DC) vestibular stimulation in humans. Exp Brain Res 1998;122:453–8. MacDougall HG, Moore ST, Curthoys IS, Black FO. Modeling postural instability with galvanic vestibular stimulation. Exp Brain Res 2006;172:208–20. http://dx.doi.org/ 10.1007/s00221-005-0329-y. Klosterhalfen S, Ruttgers A, Krumrey E, Otto B, Stockhorst U, Riepl RL, et al. Pavlovian conditioning of taste aversion using a motion sickness paradigm. Psychosom Med 2000;62:671–7 [doi: 0033-3174/00/6205-0671]. Rodriguez M, Lopez M, Symonds M, Hall G. Lithium-induced context aversion in rats as a model of anticipatory nausea in humans. Physiol Behav 2000;71:571–9. Roscoe JA, Morrow GR, Aapro MS, Molassiotis A, Olver I. Anticipatory nausea and vomiting. Support Care Cancer 2011;19:1533–8. http://dx.doi.org/10.1007/s00520010-0980-0. Colloca L, Benedetti F. How prior experience shapes placebo analgesia. Pain 2006; 124:126–33. http://dx.doi.org/10.1016/j.pain.2006.04.005. Au Yeung S, Colagiuri B, Lovibond P, Colloca L. Partial reinforcement, extinction, and placebo analgesia. Pain 2014;155:1110–7. Magnusson M, Johansson R, Wiklund J. Galvanically induced body sway in the anterior–posterior plane. Acta Otolaryngol 1990;110:11–7. Colagiuri B, Lovibond P. Psychological processes that can bias responses to placebo treatment for pain. In: Colloca L, Flaten M, Meissner K, editors. Placebo and pain: from bench to bedside. London: Academic Press; 2013. p. 175–82.
