Study design

This experimental study was conducted as a randomized controlled trial in which healthy, pain-free participants were randomly divided (counterbalanced) into two intervention groups and one control group. Both intervention groups received either a hypoalgesic or a hyperalgesic suggestion related to the pattern of the subsequently applied heat pain within an OA paradigm. The control group received no suggestion. All participants received the identical information about the exact temperature course of the heat stimuli beforehand. In order to perform the suggestions as authentically as possible, a cover story was told to all participants at the beginning of the study. All participants were blinded to the true purpose of the study. The study was previously approved by the ethics committee of the University of Lübeck (file number: 21–028) and pre-registered in the Open Science Framework (https://osf.io/69eyp). All participants provided oral and written informed consent. An overview of the study design is provided in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Study design.

Before randomization, participants were instructed and a cover story was provided. Participants were told that in this study, changes in electrodermal activity (EDA) would be assessed as a measure of the autonomic nervous system during experimental heat stimuli and used to predict the perception of pain (cover story). Participants were assigned to either i) the hypoalgesic group with suggestion towards profound hypoalgesia following the temperature reduction, ii) the hyperalgesic group with verbal suggestion towards hyperalgesia following the temperature reduction (see the example above), or iii) the control group with no intervention. Regardless of the group assignment participants were exposed to two offset and two constant trials provided in a counterbalanced manner.

https://doi.org/10.1371/journal.pone.0280579.g001

Study population

Healthy, pain-free participants aged 18 to 65 years were recruited on the campus of the University of Lübeck. All participants had to subjectively confirm that they were healthy. Furthermore, participant had no cardiovascular, systemic, psychiatric or neurological disease. Furthermore, all participants were excluded if they had a history of chronic pain (> 3 months) within the last 2 years or had experienced pain (including headache, toothache, muscle soreness, etc.) within the last week prior to study participation. In addition, the use of medication, excluding contraceptives, in the last 48 hours was an exclusion criterion. Furthermore, participants were asked not to drink alcohol, exercise, or take pain medication for 24 hours prior to participation in the study and not to drink coffee or smoke cigarettes for 4 hours prior to study participation.

Equipment

A Pathway CHEPS (Contact Heat -Evoked Potential Stimulator) with a contact area of 27mm diameter was used for the application of the heat stimuli (Medoc, Ramat Yishai, Israel). The thermode was attached to the non-dominant volar forearm approximately 10 cm below the elbow using a blood pressure cuff with a pressure of 25 mmHg. A computerized visual analog scale hardware device (COVAS; with the range 0 = "no pain" to 100 = "most tolerable pain") was used for continuous assessment of pain intensity (Medoc, Ramat Yishai, Israel). Furthermore, during heat stimuli, electrodermal activity (EDA; respiBAN Professional, Plux, Lisbon, Portugal) was measured using two Ag/AgCl hydrogel electrodes (Covidien / Kendall, Dublin, Ireland) at the medial phalanx of index and middle finger of the non-dominant arm (sampling rate of 1000 Hz, PLUX Wireless Biosignals, S.A., Portugal). Electrodermal activity depends on sweat secretion, which is closely related to autonomic nervous system activity [27]. EDA was used to test, if verbal suggestion influences physiological responses and if EDA can be used as an objective marker for OA.

Experimental heat stimulation

Two constant trials (CT) and two offset trials (OT) were performed on the non-dominant volar forearm, so that the participants were presented with a total of four heat stimuli. The order in which trials were presented was randomized in a counterbalanced fashion. A two-minute pause was kept between each stimulus, during which the thermode was moved on the forearm by a few centimeters. The temperature’s rise and fall rate for all heat stimuli was 15°C/second. During CT, the temperature increased from a baseline level of 35°C to 46°C and remained constant for 40 seconds before returning to the baseline level. During OT, the temperature first increased to 46°C (T1) for 10 seconds, then increased to 47°C (T2) for 10 seconds, and finally decreased again to 46°C (T3) lasting 20 seconds. The temperature pattern of the two trials can be seen in Fig 1. These figures were shown to the participants before the application of the heat stimuli. During the application of the heat stimuli, participants were asked to rate perceived pain continuously and as precisely using the COVAS.

Suggestion

The participants were provided with a cover story. It was explained that the aim of the study was to find out whether the subjective sensation of pain could be "read out" from the physiological reaction of the body (skin conductance) and thus be predicted. A cover story was necessary to also justify the introduction of suggestion as credibly as possible without participants becoming skeptical or biased.

The hypoalgesic or hyperalgesic groups received suggestions about the expected pain pattern and the pain intensity of the applied heat stimuli. The hypoalgesic group received a suggestion to enhance the effect of OA and adaptation to CT, i.e., to reduce pain perception. The hyperalgesic group, on the other hand, received a suggestion, which was intended to reduce the effect of OA and adaptation to CT, i.e., to increase pain perception. The expected pain pattern was manipulated verbally (S1 File) and supported with the graphical presentation of the assumed pain pattern (S1 Fig) and took place directly after the explanation of the temperature gradients, i.e. immediately before application of the respective heat stimulus.

Questionnaires

Before starting the heat application, participants were asked to complete several questionnaires: The Patient Health Questionnaire (PHQ-9) includes nine questions about depression [28]. While the Pain Vigilance and Awareness Questionnaire (PVAQ) measures pain perception and pain awareness [29], the Pain Sensitivity Questionnaire (PSQ) can be used to determine the subject’s pain sensitivity [30]. The State-Trait Anxiety Inventory-SKD (STAI-SKD) was also collected to determine the participant’s current state anxiety before the experiment [31]. The Social Desirability Scale-17 (SDS-17) was used to measure the participant’s social desirability [32]. Furthermore, the Mindful Attention and Awareness Scale (MAAS) was used to measure dispositional mindfulness [33] and the Life-Orientation Test (LOT-R) was used to measure individual differences between optimism and pessimism based on personality traits [34].

Manipulation check

To assess the effect of suggestion on pain perception during the OA paradigm, a manipulation check was performed immediately after pain assessment. The following was asked separately for OT and CT: “Please try to recall the moment immediately after receiving heat stimuli. Did you perceive the pain as in the previously displayed figures?” Participants provided a binary (yes vs. no) response.

Statistical analysis

In the absence of studies investigating OA and verbal suggestion, a meta-analysis examining the effect of verbal suggestion on general pain perception was used to calculate the sample size [22]. With the lowest reported effect size of 0.66 (Cohen’s d), a power of 80%, and an alpha of 0.05, a total number of 30 participants in each group (total = 90) was required (G*Power, University of Düsseldorf [35]) to demonstrate a significant difference between experimental and control groups.

COVAS data from the Medoc software and the EDA signals were synchronized. The time-series data were down-sampled to a frequency of 1 Hz by using the “resample” function of the python package “Pandas” (Python 3.9.7, pandas 1.4.2). Here, multiple data points are aggregated and replaced by their average. No further preprocessing steps were performed. All other statistical analyses were performed using the IBM Statistical Package for Social Science (SPSS version 26, Armonk, NY). The three groups were tested for group differences using age, BMI, sex, dominant hand, and questionnaire data. One-way analysis of variance (ANOVA), Kruskal-Wallis tests, or chi-square tests were used accordingly.

Differences between the groups in their initial pain response at the beginning of the heat stimuli were examined. For this purpose, pain ratings were averaged from the last 5 seconds of the T1 and T2 interval. Separately for OTs and CTs (mean value of the two CTs and two OTs) one-way ANOVAs were used to analyze differences between the groups. The primary outcome in this study was the magnitude of the pain response to the T3 interval (OT). To ensure that the magnitude of the OA effect was not under- or overestimated, the mean of 10 seconds centered in the T3 interval (secs. 25–34) were extracted. The 5 seconds at the beginning of the interval were not included, because the pain may still decrease during this time, and the 5 seconds at the end of the interval were not included, because the analgesic effect of OA usually decreases after approximately 15 seconds [36]. OT and CT (again mean of the two CTs and OTs) were analyzed separately, as both trials were also separately attempted to be influenced by suggestion. However, dependent t-tests were used to demonstrate whether the pain response and EDA signal from CT were significantly different from OT in each of the groups and thus whether there was an OA effect. To examine the effect of suggestion on OT and CT, a one-way ANOVA was conducted comparing the T3 interval pain response of the three groups as described. An additional method of analysis calculating the percentage difference of the maximum pain ratings for T2 and the minimum pain rating for T3 is included in the supporting information (S2 File).

The EDA data were analyzed according to identical principles as the pain response. As part of a secondary analysis of the EDA data in the control group, an ANOVA was performed with the within factor “interval” (T1, T2, T3) and the within factor “trial” (OT, CT). Furthermore, a dependent t-test was used to examine whether the magnitude of the EDA magnitude differed within the temperature increase/decrease (i.e. comparing the EDA magnitude from the difference of T2 and T1 with the difference of T2 and T3).

If statistically significant main or interaction effects were detected, Bonferroni corrected post-hoc t tests were conducted. The correlations between the pain response of the T3 (OT) and the previously described questionnaires were calculated using the Spearman coefficient. No data were missing at the time of analysis. A p value of less than 0.05 was considered significant for all comparisons.

Expectancy mechanism

To the best of our knowledge, this is the first study that has attempted to influence OA using suggestion. However, similar results have already been reported for studies that attempted to influence outcomes using other paradigms to quantify endogenous pain modulation by using suggestion. For example, a similar conclusion was reported by Vaegter et al. (2020) which attempted to influence exercise-induced hypoalgesia (EIH) using suggestion [37]. In that study, EIH was defined as an increased pain threshold and pain tolerance induced by performing a single exercise routine. It was found that volunteers who received a negative suggestion prior to exercise, experienced hyperalgesia instead of EIH. Furthermore, studies found that CPM can also be influenced by suggestions and thereby altered expectation [26, 38]. In CPM, the pain response to a painful test stimulus is inhibited by the application of a distant painful conditioning stimulus [39]. Goffaux et al. (2007) studied 20 healthy volunteers regarding their pain perception during the CPM paradigm while they were given different verbal suggestions about the expected pain process [38]. While the hypoalgesic group experienced profound analgesia, this was absent in the hyperalgesic group. Moreover, Bjørkedal and Flaten (2012) also found an effect of verbal suggestion on pain perception in the CPM paradigm [26]. It should be noted, however, that OA is not based on the same mechanisms as CPM and EIH, making them not directly comparable. For example, CPM, unlike OA, can be influenced by ketamine [40], and the two paradigms have underlying distinct brain mechanisms [41]. In addition, other studies have shown that there is no correlation between OA and EIH [42] or OA and CPM [41, 43], also suggesting individual mechanisms of these pain modulation phenotypes. However, based on these similar results for CPM, EIH, and OA, it is reasonable to assume that altered expectations manipulated by suggestion influence endogenous pain modulation processes as quantified via various paradigms.

In general, it can be assumed that suggestion influences OA, since brain activity during OA overlaps with the activity during placebo analgesia [8, 44]. Previous studies have shown that especially the activation of the rostral anterior cingulate cortex (rACC) and the dorsolateral prefrontal cortex (DLPC) play a major role in placebo analgesia [45, 46]. They are both functionally connected with the periaqueductal gray and the rostral ventromedial medulla (PAG-RVM system), which can send inhibitory projections to the spine and thereby elicit a diffuse analgesic response [47]. Increased activation of the DLPC and PAG-RVM circuits were also found during OA [13, 15, 41], suggesting that the mechanisms of placebo analgesia and OA may be similar. However, this is contradicted by the results that hypoalgesic suggestion did not produce increased pain reduction in OA in this study. This can be explained by the fact, that previous studies have shown that placebo analgesia is mediated primarily by increased release of endogenous opiates [48], whereas OA has been shown to be opioid-independent [49]. For example, placebo analgesia can be blocked by the opioid antagonist naloxone [50], whereas naloxone, on the other hand, has no effect on the magnitude of OA [49]. One study has found that nocebo hyperalgesia is mediated by the neurotransmitter cholecystokinin (CCK) [51]. However, the effect of CCK on OA has not yet been studied, although it is relevant since the hyperalgesic manipulation could have influenced the OA.

Interestingly, no significant effect of the suggestion was found on CT in this study. Pain perception in the T3 interval of the CT was neither increased nor decreased. This result is in contrasts with other reports because, in principle, both hypo- or hyperalgesic suggestion have been found in previous studies to influence a wide variety of noxious stimuli [21, 52]. For example, the study by van Laarhoven et al. (2011) found an effect of hyperalgesic verbal suggestion on pain perception in healthy women. However, this study did not use tonic heat stimuli, but mechanical and electrical stimuli [23]. The methodology of other studies also differed in many ways from the present study. For example, studies often used other stimulus modalities (e.g., cold, electric shocks, ischemic pain), or did not use verbal and visual suggestion but used either conditioning alone or a combination of conditioning and suggestion to influence pain perception [21, 52]. Furthermore, no study was found that investigated the effect of verbal and visual suggestion on a constant (tonic) heat stimulus, as done in this study. However, one explanation for the differences in influences on CT versus OT could be the difference in physiological processing. It is suggested that pain adaptation to a moderate, constant heat stimulus is primarily mediated by peripheral mechanisms [53–55]. In comparison, both peripheral and central mechanisms are known to shape OA [8]. Since peripheral mechanisms cannot be influenced by suggestion, this could be a possible explanation for the lack of influence on CT. At the same time, the shown suggestibility during OT could support the assumption that OA is primarily a central phenomenon, as pain perception was modulated by expectancy, here. Further studies comparing the suggestibility of responses to OT and CT are needed to draw further conclusions about the underlying mechanisms.

As a limitation, suggestions might not have been fully successful, as shown by the results of the manipulation check. Although the majority (70.1%) of the subjects reported that they perceived a pain response during OT according to the prior given suggestion. However, in contrast to the hypoalgesic group (90.6%), only 21.2% in the hyperalgesic group confirmed a pain response as suggested. Thus, it can be assumed that one reason for this may be the exceptionally robust analgesia during the OA paradigm. Thus, these results show that OA can be influenced only unidirectionally, being more likely to be enhanced but more difficult to be inhibited.

Physiological mechanisms

So far, physiological measurements taken during OA included functional magnetic resonance imaging [12, 13, 15, 17, 41], electroencephalography [56] and functional near-infrared spectroscopy [57]. To the best of our knowledge, this is the first experiment which recorded EDA during OA. This was done for the following reasons: Firstly, OA has been shown to be mediated by the activation of brain areas associated with the regulation of autonomic reactivity [13, 15, 58, 59], thus we aimed to capture this variable continuously to test if OA measurements behaviorally overlap with physiological responses. Secondly, we aimed to investigate if verbal suggestion alters both, subjective and objective outcomes during an OA paradigm.

Interestingly, results of this study showed that, contrary to our prediction, EDA responses to OT were not decreased alongside pain perception. In turn, the increase of the temperature during a T2 interval significantly elevated the EDA level which persisted during the T3 interval, whereas the pain response was reduced. Indeed, a lowered pain intensity overlapped with higher EDA level in offset compared to constant trial. It can be suggested that the higher stimulus in T2 of the OT activates the descending pain inhibition pathways and therefore inhibits pain. In fact, the EDA level during CT gradually decreased over-time which was, in general, associated with higher pain compared to OTs. During OA, endogenous modulatory mechanisms have been shown to be activated [9], which are believed to be driven by PAG activation [13]. PAG is an anatomic structure with multiple nociceptive projections and it plays a crucial role in control of autonomic functions [58, 59]. Whether the enhanced activation in PAG explains reversed offset in EDA needs to be determined.

However, whether this EDA response represents a physiological correlate of the OA via pain response is not clear. Indeed, no differences were found for the EDA data within the OT (as is usual for the pain response) when the time intervals T1 and T3 were considered. One can therefore assume that the increased EDA response within T3, may related to an (still) ongoing EDA response caused by a higher stimulation during the T2 interval. Thus, it should be further investigated whether the EDA within an OA paradigm may also reflect the pain predictions that may involve perception of both temperature decreases (OA) and temperature increases (onset hyperalgesia) as shown by Alter et al. 2020 [60].

Our psychological manipulation did not influence the EDA signal. Although EDA has been used extensively to capture reported pain intensity and has been shown to be a potent biomarker in pain prediction [61], we could not observe that autonomic reactivity was influenced by verbal suggestion. Similar results have been reported in some of the previous experiments [62], in which verbal suggestions towards analgesia or hyperalgesia were provided [63]. It cannot be, however, excluded that this is a result of the relatively small effect size observed at the behavioral level (pain).

Psychological mechanisms

The results of this study could serve as an explanatory approach to describe why OA is reduced in chronic pain patients. Various studies showed that a large proportion of chronic pain patients have dysfunctional beliefs about their condition and dysfunctional coping strategies in dealing with their condition [64]. It can be hypothesized that because of these dysfunctional beliefs and coping strategies, chronic pain patients have a fundamentally more negative expectancy toward pain. In this study, a negative expectancy was also evoked in the nocebo group by a suggestion in healthy participants. These participants also subsequently showed reduced OA. Thus, the expectation towards pain could have a decisive influence on the magnitude of OA. For the reduced OA in the nocebo group, the optimism of a person seems to play a role. According to the results, it can be assumed that a more pronounced optimism reduces the effect of suggestion of the nocebo group on OA. Thus, the individual could protect from a more pronounced hyperalgesia as a result of the received suggestion. In contrast, no significance of other psychological factors considered for the effect of the suggestions could be observed.