Negative emotions are commonly experienced in fibromyalgia and may affect pain. This study examined the effects of anger and sadness on clinical pain reports and on pain threshold and tolerance in response to electrical stimulation in women with and without fibromyalgia.
In an experimental study, 62 women with fibromyalgia and 59 women without fibromyalgia recalled a neutral situation, followed by recalling both an anger-inducing and a sadness-inducing situation, in counterbalanced order. The effect of these emotions on pain responses (non-induced clinical pain and experimentally-induced sensory threshold, pain threshold, and pain tolerance) was analyzed with a repeated-measures analysis of variance.
Clinical pain reports increased (P < 0.001) in women with fibromyalgia, and pain threshold (P < 0.001) and tolerance (P < 0.001) decreased in both groups in response to anger and sadness induction. Sadness reactivity predicted clinical pain responses. Anger reactivity predicted both clinical and electrically-stimulated pain responses.
The experience of both anger and sadness amplifies pain in women with and without fibromyalgia. A stronger emotion-induced pain response was associated with more emotional reactivity. No convincing evidence was found for a larger sensitivity to anger and sadness in women with fibromyalgia than in women without fibromyalgia, or for a larger sensitivity to anger than to sadness in fibromyalgia. The occurrence of anger and sadness appears to be a general risk factor for pain amplification. Emotion regulation techniques may attenuate emotional pain sensitization in patients with fibromyalgia.
Fibromyalgia is a chronic widespread pain disorder of unknown etiology (1). Enhanced pain sensitivity to a variety of psychophysical stimuli has been found (2). In both healthy and chronic pain samples, emotions have also been observed to increase pain (3–5). Neural, physiologic, and psychological mechanisms may explain this emotion–pain link. Neurally, there is a close anatomic relationship between pain and emotion circuits in the brain (6, 7), with emotional and physical pain sharing a common neural alarm system (8–10). Physiologically, increased sympathetic activation, decreased opioid pain inhibition, and muscular reactivity in response to negative emotions can increase nociception (7, 11). Psychologically, negative emotions may increase attention to and bias the processing of stimuli (7, 12–14), leading to hypervigilance, symptom worry, misattribution of arousal to somatic sources, and avoidance behavior (14). Although under limited circumstances negative emotions may inhibit pain (7, 14), under most circumstances, negative emotions are pain amplifying (5, 15–17).
Negative emotions are experienced more strongly in patients with fibromyalgia than in the general population (18), and stressful interpersonal events (19), general negative affect (20), and anger in daily life (21) increase pain sensitivity. Anger may be particularly key to pain (3, 22, 23) because anger increases pain perception and may trigger pain-relevant physiologic reactivity and muscle tension (24–26). Two studies found enhanced pain in response to anger, but not to sadness or anxiety (14, 25). It remains unclear, however, whether general negative affect versus anger specifically is important to chronic pain.
Our aim was to examine the effects of experimentally-induced anger and sadness on self-reported clinical and experimentally-induced pain in a sample of women with and without fibromyalgia. If general negative affect exacerbates pain, then similar pain increases to anger and sadness will be observed, and both anger and sadness reactivity will predict pain severity. Alternatively, if anger specifically is pathogenic, then induced anger and anger reactivity will predict pain severity more than induced sadness and sadness reactivity. Analogous to the enhanced sensitivity for physical stimuli, we expect that negative emotions will increase pain more in women with fibromyalgia than in comparison women.
SUBJECTS AND METHODS
Participants consisted of 62 women with fibromyalgia and 59 women from the general population without fibromyalgia. The patients were classified with fibromyalgia by a rheumatologist of 1 of 3 hospitals in Utrecht and Almere, The Netherlands, according to established criteria (27). A comparison group consisted of women from the general community who did not have fibromyalgia, but could have other health problems found in the general population. Both women with and without fibromyalgia were recruited from a larger sample that previously participated in a questionnaire study (18). For this laboratory study, we excluded people who were receiving beta-blockers or angiotensin-converting enzyme inhibitors for high blood pressure, were pregnant, or had had a thyroidectomy. There were no other exclusion criteria applied so that the results would have greater generalizability to the respective populations.
Of 329 women invited to participate in the current study (188 women with and 141 women without fibromyalgia), 64 (19%) did not respond to repeated requests by letter, and 127 (39%) declined or were unable to participate. Stated reasons for not participating included: lack of time or transportation or scheduling difficulties (33%), not wanting to recollect emotional experiences (13%), personal circumstances (13%), physical problems (13%), not being interested in or supporting the content of the study (12%), and other reasons (15%). A total of 138 women (42% of those invited) were interested in participating and were screened for eligibility; 13 (9%) were excluded because of high blood pressure medication, 3 (2%) because of pregnancy, and 1 (1%) because of a thyroidectomy. The women participating in the current study were significantly older than the women who did not participate (average age 46 versus 43 years; P = 0.008), but did not differ on marital status, employment status, education, comorbidity, medication use, and (for women with fibromyalgia) the duration since first symptoms and diagnosis of fibromyalgia (all P values >0.07).
Table 1 shows the demographic and health-related characteristics of the participating women, who ranged in age from 21 to 72 years. Women with fibromyalgia had a mean symptom duration of 10 years (range 1.5–50 years) and were first diagnosed with fibromyalgia on average 3.5 years (range 4 months to 25 years) before the study. Women with and without fibromyalgia did not differ with regard to most demographic characteristics, except that women with fibromyalgia were less likely to be fully employed (χ2 = 40.99, P < 0.001). As expected, the groups differed on health-related characteristics; more women with fibromyalgia had been diagnosed with a psychological or psychiatric condition (χ2 = 6.49, P = 0.01), used prescribed (χ2 = 25.00, P < 0.001) or over the counter (χ2 = 10.83, P = 0.001) medications, and received nonpharmacologic treatment (χ2 = 24.23, P < 0.001). On the day before or the day of the experiment, women with and without fibromyalgia did not statistically differ in their use of antidiuretic medications (3% versus 5%), glucocorticoids (3% versus 2%), or thyroid medication (8% versus 2%), but women with fibromyalgia were more likely to use analgesics (52% versus 10%; χ2 = 24.10, P < 0.001), antidepressants (27% versus 3%; χ2 = 13.19, P < 0.001), or other medications (e.g., cholesterol or allergy medication, 29% versus 10%; χ2 = 6.77, P = 0.009).
Table 1. Demographic and health-related characteristics of women with fibromyalgia and women without fibromyalgia*
Fibromyalgia (n = 62)
No fibromyalgia (n = 59)
Values are the number (percentage) unless otherwise indicated. Group differences were tested by chi-square tests, except for age (independent-samples t-test).
Nonpharmacologic treatments (e.g., physiotherapy, psychotherapy, alternative medicine)†
The participants came to the laboratory between December 2006 and January 2008. The protocol of the study was approved by the research and ethics committee of the University Medical Center Utrecht, and all of the participants provided written informed consent.
The design allowed the study of both between-person (with or without fibromyalgia) and within-person (neutral emotion followed by anger and sadness inductions) comparisons. The participants first had a practice session with the pain apparatus (see below) to get accustomed to the assessment protocol. Each emotion induction condition started with the participant viewing a 2-minute relaxing nature video. Next, an autobiographical recall procedure was used to induce the targeted emotion (see below). All of the participants started with the neutral condition, which provided a baseline assessment of pain and emotion, after which either anger was elicited first followed by sadness (60 participants: 31 women with and 29 women without fibromyalgia), or sadness was elicited first followed by anger (61 participants: 31 women with and 30 women without fibromyalgia). The order of negative emotion induction was assigned randomly, stratified by fibromyalgia status.
We used an emotion induction procedure in which the participant was asked to recall an event in which she felt neutral (a general everyday event, such as taking a walk or cooking), angry (an event that still makes one angry when recalling it), or sad (an event that still makes one sad, such as a death of a family member). This mood induction procedure has high ecological validity and strongly elicits the targeted emotion (28, 29). The participants were asked to describe every detail and to continue talking until they felt the emotion strongly, and then to think silently about their emotions for 2 minutes, which allowed us to obtain physiologic measures not confounded by talking (the physiologic data are not shown here). The experimenter sat behind the participant and did not respond in any way. Immediately after these 2 minutes, the emotion and pain questionnaires were completed, and then 4 trials of the experimentally-induced pain assessment were conducted. After the pain assessments, participants counted back from 25 to 0 to distract themselves from the induced emotional experience and pain assessments, after which they read a neutral magazine for 5 minutes before the next condition started with presentation of the nature video. After the final condition, the participants were debriefed, received reimbursement for travel expenses but were not otherwise paid, and filled out an evaluation questionnaire. The experimental session lasted approximately 2 hours.
To assess anger and sadness “now, at this moment” in response to the neutral, anger, and sadness induction conditions, the 2 items most closely related to the constructs of anger (“angry” and “irritable”) and sadness (“sad” and “blue”) of the Positive and Negative Affect Schedule–Expanded Form (30) were used. Items were rated from 1 (very slightly or not at all) to 5 (extremely). In a previous study, each pair of items showed large intercorrelations (r = 0.74 for both) and excellent internal consistency (α = 0.85 for both scales) (21). Ratings for the 2 items for each emotion were summed, yielding a score ranging from 2–10.
Clinical pain and experimentally-induced pain were assessed. Self-reported clinical pain always preceded the experimentally-induced pain assessments and consisted of reporting current pain levels (“now, at this moment”) on a 100-mm visual analog scale, with anchors of “no pain at all” to “intolerable pain.” Clinical pain reports were analyzed in women with fibromyalgia only. For experimentally-induced pain measures, electrical pain induction was chosen because we wanted to assess pain repeatedly in the same person, and pain through electrical stimulation is short lived. Pain was induced by electrical stimulation with a Tursky concentric electrode attached to the inner side of the nondominant forearm (31). Over 40 seconds, an electric current gradually was increased from 0 mA to a maximum of 6 mA. The summarized instructions were: “A slowly increasing stimulus will be delivered to your forearm. You are asked to press a button three times as indicated on the screen. First, when you feel the signal. Second, when the tingling becomes painful. Third, when the tingling becomes so painful that you want it to stop. Then, after pressing the button, the stimulus stops and the pain is gone.” Therefore, the participants pressed a button when they felt the current (sensory threshold) and when it became painful (pain threshold) and intolerable (pain tolerance). Four pain assessments were conducted per condition, and very high internal consistencies were obtained (α = 0.94–0.99). Median values instead of means were analyzed to deal with occasional outliers that sometimes occurred because the participants accidentally pushed the button too soon or too late.
A single clinical pain rating was missing for one woman with fibromyalgia. Experimental pain assessments were missing for 5 women without fibromyalgia due to technical problems with the pain stimulus device. Therefore, analyses were conducted with data of a minimum of 61 women with and 54 women without fibromyalgia.
The neutral condition was used as the contrast condition to control for nonspecific factors. No meaningful changes in emotions or pain were found in response to the neutral induction as compared with watching the nature video at baseline. In order to compare the pain response to anger versus sadness, change scores were created for the emotion and pain ratings by subtracting the values at neutral from the sadness and anger induction conditions (conducting the analyses on final values instead of the change scores did not impact the results). To examine whether the emotion induction procedures were successful in inducing anger and sadness, a repeated-measures analysis of variance was conducted on anger and sadness change scores with emotion induced (anger or sadness) as the within-subjects factor, and participant group (with or without fibromyalgia) and induction order (anger before sadness or sadness before anger) as the between-subjects factors. To assess the impact of induced anger or sadness on pain, similar analyses were conducted. To express the magnitude of effects, partial eta squared (ηp2) effect sizes were used, with values of 0.01, 0.06, and 0.14 being considered small, moderate, and large effects, respectively (32). Because medication use could impact pain reactivity, the analyses were repeated while entering 6 dummy-coded (0 = no, 1 = yes) scores for medication taken on the day of or before the experiment: glucocorticoids, analgesics, antidiuretic medications, thyroid medication, antidepressants, and other medications. All of the significant effects of emotion induction remained significant after co-varying medication use; therefore, these results are not shown.
To determine associations of anger and sadness reactivity with emotion-induced pain responses, Pearson's correlation coefficients were calculated.
Group comparisons during the neutral condition.
Table 2 shows the means and SDs of the emotion and pain measures at the neutral induction condition for both groups. Women with fibromyalgia reported more sadness than women without fibromyalgia (t = −3.29, P = 0.001). The two groups did not differ with respect to their level of anger (P = 0.43), sensory threshold (P = 0.60), pain threshold (P = 0.10), and pain tolerance (P = 0.08) after the neutral condition.
Table 2. Mean ± SD of the emotions and pain measures at the neutral recall condition for women with and without fibromyalgia
Figures 1A and B show that both groups had highly significant anger and sadness increases from the neutral condition in response to the induction of anger and sadness (anger overall effect: F[1,117] = 497.55, P < 0.001, ηp2 = 0.81; sadness overall effect: F[1,117] = 571.71, P < 0.001, ηp2 = 0.83). Anger increased from a mean ± SD of 2.36 ± 0.80 in the neutral condition to 7.10 ± 2.16 in the anger condition and 4.38 ± 2.18 in the sadness condition. Sadness increased from a mean ± SD of 2.50 ± 1.01 in the neutral condition to 5.31 ± 2.41 in the anger condition and 7.50 ± 1.97 in the sadness condition. As intended, there was specificity to the targeted emotion: the anger response was stronger in the anger than in the sadness condition, and the sadness response was stronger in the sadness than in the anger condition (anger condition effect: F[1,117] = 135.42, P < 0.001, ηp2 = 0.54; sadness condition effect: F[1,117] = 105.16, P < 0.001, ηp2 = 0.47). There were no time (anger: P = 0.53; sadness: P = 0.98) or order effects (anger: P = 0.80; sadness: P = 0.95). Both of the groups showed a similar anger response (group effect: P = 0.56). Of note, the women without fibromyalgia showed a larger sadness response than the women with fibromyalgia (F[1,117] = 7.54, P = 0.007, ηp2 = 0.06).
Impact of anger and sadness on pain.
Clinical pain (fibromyalgia group only).
Clinical pain reports significantly increased with a large effect (overall effect: F[1,59] = 29.25, P < 0.001, ηp2 = 0.33), from a mean ± SD of 32 ± 18 in the neutral condition to 41 ± 21 in the anger condition and 42 ± 19 in the sadness condition. The magnitude of increase in clinical pain did not differ between the anger and sadness conditions (condition effect: P = 0.77), between the first and second negative emotion inductions (time effect: P = 0.08), and as a function of the order in which negative emotions were induced (order effect: P = 0.09).
The experimentally-induced sensory threshold did not change significantly for the sample overall after the induction of anger (mean ± SD 0.82 ± 0.47) or sadness (mean ± SD 0.85 ± 0.51) as compared with neutral (mean ± SD 0.83 ± 0.47; overall effect: P = 0.82), and the change did not differ between the women with and the women without fibromyalgia (group effect: P = 0.90), the anger and sadness conditions (condition effect: P = 0.15), the first and second negative emotion inductions (time effect: P = 0.59), or emotion order (order effect: P = 0.87).
Pain threshold and tolerance.
The experimentally-induced pain threshold decreased with a moderate effect size from a mean ± SD of 2.35 ± 1.22 in the neutral condition to 2.21 ± 1.17 in the anger condition and 2.21 ± 1.25 in the sadness condition (overall effect: F[1,112] = 13.77, P < 0.001, ηp2 = 0.11) (Figure 1C). Similarly, pain tolerance decreased from a mean ± SD of 3.14 ± 1.38 in the neutral condition to 3.04 ± 1.41 in the anger condition and 2.95 ± 1.42 in the sadness condition (overall effect: F[1,112] = 13.08, P < 0.001, ηp2 = 0.11) (Figure 1D). The overall decrease in pain threshold and pain tolerance did not differ between the women with and without fibromyalgia (group effect: P = 0.06 and 0.16, respectively), between the anger or sadness conditions (condition effect: P = 0.96 and 0.10, respectively), or between the first and second negative emotion inductions (time effect: P = 0.67 and 0.42, respectively). Although the change in pain thresholds did not differ as a function of the order in which emotions were evoked (order effect: P = 0.11), pain tolerance was decreased more to both anger and sadness when anger was the emotion first induced (F[1,112] = 5.88, P = 0.02).
Anger and sadness reactivity.
Among women with fibromyalgia, both anger and sadness reactivity to the emotion inductions were associated with greater increases in clinical pain responses (anger: r = 0.36, P = 0.004; sadness: r = 0.36, P = 0.004).
In the total sample, anger reactivity was associated with a larger decrease in pain threshold in response to the sadness (r = −0.22, P = 0.02), but not to the anger induction (r = −0.03). Anger reactivity was also associated with a larger decrease in pain tolerance, irrespective of group or emotion induction condition (r = −0.20, P = 0.04). Sadness reactivity was not predictive of pain threshold (P = 0.60) or pain tolerance (P = 0.89) responses.
Both women with and women without fibromyalgia manifested increased pain in response to the induction of both anger and sadness, and greater emotional reactivity was associated with a greater pain response. No convincing evidence was found for a larger pain response to anger and sadness in women with than without fibromyalgia, or for a larger pain response to anger than to sadness in women with fibromyalgia.
Although emotions reflecting imminent threats such as fear have shown pain-inhibiting effects (33), negative emotions generally have been shown to amplify pain in both healthy (17, 34) and chronic pain samples (25). Our study confirms this pain-amplifying effect for an externalizing emotion, i.e., anger, and an internalizing emotion, i.e., sadness, in women with and without fibromyalgia. Both clinical pain reports (large effect size) and pain sensitivity to electrical stimulation (moderate effect size) increased after these negative emotions. Sensory threshold was not affected, which may be due to different neural areas being involved in the sensory perception of a signal than in experiencing a signal as aversive (35). Whether the pain-sensitizing effect of anger and sadness is explained by neural, physiologic, or psychological mechanisms is a meaningful area of further research. Regardless of the exact mechanism, our findings suggest that addressing anger and sadness may be a meaningful therapeutic target to attenuate pain sensitivity.
Our results provide only limited support for the hypothesis that anger is a particularly strong pain inducer (3, 25). The pain response to electrical stimulation was positively related to the degree to which the experience of anger, but not sadness, increased in response to the emotion inductions. However, women with fibromyalgia did not show a stronger pain response to anger than to sadness. This finding and the observation that both anger and sadness reactivity augmented clinical pain are consistent with the notion that general negative affect or a common element such as aversion rather than specific emotions contributes to the amplification of pain.
Fibromyalgia has been described as a sensitivity syndrome (1, 2) because patients may even respond with pain to stimuli that do not typically cause pain, termed allodynia. We hypothesized that women with fibromyalgia also experience what might be called “emotional allodynia,” i.e., increased pain in response to negative emotions. We found only partial support for this. Women with fibromyalgia did indeed experience increased clinical and experimental pain in response to induced anger and sadness. However, unlike a study that found increased pain responses to anger in patients with chronic low back pain but not healthy controls (25), in our study, both women with and without fibromyalgia showed emotion-related pain responses to electrical stimulation. This finding raises questions about the possible uniqueness of emotionally-induced pain in fibromyalgia. Nonetheless, it is a clinically relevant finding that pain in the women with fibromyalgia was increased above an already high baseline level when anger and sadness were induced.
The experience of more intense negative emotions may trigger greater pain because stronger emotions enhance neurophysiologic pain circuits (17). We examined whether pain responsiveness was indeed “caused by” emotional intensity. Greater anger reactivity, and to a lesser extent sadness reactivity, were associated with a stronger pain response, which supports the hypothesis that emotions can amplify pain. Given the high level of pain in patients with fibromyalgia (27), the frequency and intensity of negative emotions in this group (18), and the unavoidability of experiencing anger or sadness on a regular basis in daily life (21), suggestions are needed on how to prevent anger-induced and sadness-induced pain amplification. Perhaps mindfulness-based and cognitive–behavioral stress reduction interventions (36, 37) are helpful in reducing emotional intensity.
With respect to methodologic strengths and weaknesses, to our knowledge, our study is the first to test the causal effects of 2 separate negative emotions on pain in a rather large sample of women with and without fibromyalgia. Our design combined rigorous experimental testing and generalization to situations in everyday life by using ecologically valid recall of emotions. An inherent problem with emotion recall procedures is the difficulty of inducing a single unique emotion while minimizing the experience of other emotions. Therefore, our procedure may have obfuscated the specific effects of anger versus sadness on pain. Also, people likely varied in how intensively and accurately they experienced and expressed the targeted emotion. Patients with fibromyalgia may be relatively poor at recognizing emotions and giving feelings genuine, non-conflicted verbal expression (18), as was expected of them in this study. It is noteworthy that the comparison women in our study actually reported more sadness than did the women with fibromyalgia during the sadness induction, and a similar, although nonsignificant, trend occurred for anger during the anger induction. It is possible, therefore, that an autobiographical memory approach, i.e., asking people to recall, experience, and then verbally express angry and sad events, was the most difficult and least successful for women with fibromyalgia. If so, then this might account for the failure to demonstrate that the women with fibromyalgia showed a greater emotion-induced pain response than the control women. It is also possible that it is not the generation and experience of particular emotions that increase pain, but rather the inability to distinguish emotions (38) or the inhibition of an emotion such as anger (39). Future research might provide more insight into this by recording and analyzing the emotional expression or by selecting and comparing patients high and low on alexithymia and anger inhibition. Our findings may neither generalize to fear (33) nor to major depression (40).
Another issue relates to our control group. Given the high prevalence of various symptoms and pain problems in middle-aged women, selecting a control group that was completely free from pain and health problems would lead to overestimation of differences between the groups and not adequately represent the general population. Therefore, we purposefully identified a comparison group of women who were free from fibromyalgia, but had whatever physical and mental health symptoms or concerns that are typically found in a normative population. As a result, this comparison group may be more like the fibromyalgia group than in studies that had healthier, pain-free control groups. Furthermore, although various methods to experimentally induce pain are strongly correlated (41), our findings of electrically-induced pain do not necessarily translate to other kinds of pain such as pressure pain or randomly occurring natural pain. Finally, the participants were allowed to use pain and other medications before the experiment. We allowed this because discontinuing regular medications may lead to withdrawal or otherwise affect responses to the experimental procedures. Although allowing medication use increased the variance in our study, controlling for medication effects did not change the results.
There is increasing appreciation of the importance of emotions in fibromyalgia and other chronic pain populations (42, 43), and this includes both emotional states such as anger, depression, and anxiety, and emotional processes such as awareness, processing, and expression. The present study demonstrates emotional pain sensitization, although this was not limited to the women with fibromyalgia. However, emotional sensitization of pain may be especially detrimental in people who already have high pain levels. Research should test techniques to facilitate better emotion regulation, including not only interventions to down-regulate emotional experience such as relaxation, cognitive reappraisal, and exercise, but also techniques to facilitate emotional awareness, experiencing, and processing.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. van Middendorp had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Van Middendorp, Lumley, Jacobs, Bijlsma, Geenen.
Acquisition of data. Van Middendorp, Jacobs, Bijlsma, Geenen.
Analysis and interpretation of data. Van Middendorp, Lumley, Geenen.
We thank Paulien Vermaas for her valuable contributions to the project; Nienke Braam, Ligaya Butalid, Madelon Cremers, Saskia Mulder, Marijn van Oers, and Jessie Smulders for help in data collection and data entry; Leslie Beks of the University Medical Center Utrecht and the rheumatologists of the University Medical Center Utrecht, Diakonessenhuis Utrecht, and Flevoziekenhuis Almere for their help in the recruitment of patients; Ton Aalbers of Tilburg University for providing the experimental pain device; and Irene Klugkist from the Department of Methodology and Statistics at Utrecht University for statistical advice.