- Email: [email protected]
Effects of stress and relaxation on capsaicin-induced pain
Effects of Stress and Relaxation on Capsaicin-Induced Pain Henrietta Logan, Susan Lutgendorf, Pierre Rainville, David Sheffield, Kurt Iverson, and David Lubaroff Abstract: A sizable body of research has been devoted to understanding the relationship between pain sensitivity and the psychological state of the individual. Considerable disagreement as to the direction of the association still exists. This study examines the effects of 2 experimental manipulations, cognitive/emotional stress and relaxation, on capsaicin-induced pain. Subjects were pretrained in relaxation and then randomized to experimental stress produced by a 20-minute Stroop test, relaxation (tape), or a control condition (neutral video), followed by a capsaicin injection in the forearm. Cardiovascular measures were taken at regular intervals, and cortisol, norepinephrine (NE), and selfreports of arousal (relaxation index) were taken immediately before and after the experimental task. The manipulation significantly interacted with sex to predict capsaicin-induced maximum pain. Women in the stress condition reported greater pain than both men in the stress condition and women in the relaxation condition. Pain was correlated negatively with task-induced changes in NE and cortisol and positively with self-reported arousal (decreased relaxation). However, separate analyses showed that some physiologic indexes of heightened arousal (increased blood pressure and NE) predicted lower pain only in men, whereas subjective increases in arousal predicted higher pain only in women. Multiple hierarchical regression analyses confirmed that physiologic and selfreported arousal predicted pain independently and in opposite directions, and a model including both accounted for 56% of the overall variance. These findings suggest that a unidimensional model of arousal may be insufficient to explain the effects of stress on pain and that these effects operate differently in men and women. © 2001 by the American Pain Society Key words: Pain, capsaicin, stress, relaxation, arousal, sympathetic nervous system.
D
uring the past decade, a substantial body of research has been devoted to understanding the relationship between pain sensitivity and the psychological state of the individual.1 However, there is still considerable disagreement as to the direction of the association between cognitive/emotional stress and pain. Studies have shown emotional stress and anxiety increase pain and that further stressing already stressed patients can produce greater pain sensitivity.2-5 For example, Logan et al6 reported that atypically stressed dental patients anticipate, experience, and remember more pain after invasive dental procedures. In addition, relaxation and biofeedback interventions that reduce Received May 1, 2000; Revised August 25, 2000; Accepted September 6, 2000. From the Division of Public Health Services and Research and Department of Medicine, University of Florida, Gainesville, FL, and Department of Psychology, Department of Neurology, Dows Institute for Dental Research, and Departments of Urology and Microbiology, University of Iowa, Iowa City, IA. Supported in part by Grant RR00059 from the General Clinical Research Centers Program, NCRR, NIH. Dr Rainville is a postdoctoral research fellow supported by the Human Frontier Science Program organization. Address reprint requests to Henrietta Logan, PhD, Division of Public Health Services and Research, Box 100404, Gainesville, FL 32610. E-mail: [email protected] © 2001 by the American Pain Society 1526-5900/01/0203-0001$35.00/0 doi:10.1054/jpai.2001.21597
stress-induced arousal have been shown to reduce pain sensitivity.7,8 However, other studies found that anxiety and stress decrease pain sensitivity.9,10 In addition, animal and human laboratory studies have shown an analgesic effect associated with stress-induced arousal.11-14 These inconsistencies question the possibility to generalize experimental findings to clinical settings. Some studies suggest that it is the psychological characteristics of the stressor—how it is perceived—that influence pain sensitivity, whereas others have focused on the effect of physiologic arousal. In this respect, the study by Rhudy and Meagher15 is particularly informative. This study suggests that fear induced by exposure to brief shock decreases pain sensitivity, whereas anxiety elicited by the threat of shock is associated with the reverse. Furthermore, results indicate that the exposure to the shock was associated with high levels of physiologic arousal, whereas the threat only produced moderate arousal, leading these investigators to conclude that pain modulation could be explained by physiologic arousal. According to these investigators, the “immediate alarm reaction to present threat” (fear) is “characterized by impulses to escape” (fight/flight), increased arousal, and decreased pain sensitivity. In contrast, “future-oriented emotion characterized by negative affect and apprehensive anticipation of potential
The Journal of Pain, Vol 2, No 3 (June), 2001: pp 160-170
160
ORIGINAL REPORT/Logan et al threats (anxiety) […] result in hypervigilance” and increased pain sensitivity. Bandura et al12 showed that low perceived self-efficacy in coping with a stressor may lead to both high levels of autonomic arousal and an opiate-mediated decrease in pain sensitivity possibly similar to the analgesia associated with fear in the study by Rhudy and Meagher.15 These effects underscore the complementary nature of psychological and physiologic interpretations and suggest that stressinduced arousal may account for pain modulation. Sex-related differences in physiologic or psychological responses might partly contribute to some of the discrepant findings on the effects of cognitive/emotional stress on pain.16-18 Sex differences in neuroendocrine response to foot shock and immobilization have been reported, with the rat sympathoadrenal system showing more reactivity in females than males.19,20 Men have higher resting blood pressure and larger blood pressure and neuroendocrine responses to stress than women.21,22 In addition, sex differences in thermal pain report are reduced after adjusting for resting blood pressure.23,24 These sex differences in pain- and stressinduced physiologic activity raise the possibility that men and women may display different patterns of pain modulation in response to interventions that alter physiologic arousal. Furthermore, different psychological characteristics are related to pain sensitivity in men and in women. For instance, Fillingim et al25 found that higher scores on perceived efficacy and control beliefs were associated with lowered pain sensitivity in women but not men, whereas lower anxiety was related to decreased pain sensitivity in men but not women. Thus, both physiologic and psychological variables may be differentially relevant to the stress-pain perception relationship in men and women. We conducted the following investigation to test the effect of pain-unrelated cognitive stress/arousal and relaxation on pain induced by a subcutaneous injection of capsaicin. We focus on the maximum pain report using the capsaicin model because of its clinical relevance.26 Capsaicin is the irritant agent found in hot chili peppers that, when injected into the skin of human subjects, produces neurogenic inflammation and evokes the sensation of intense burning (similar to a wasp sting), hyperalgesia, and allodynia that is nearly impossible to ignore.27-29 Capsaicin-sensitive primary afferent fibers play a critical role in the initiation and maintenance of pathologic pain including neuropathies.30 Capsaicin-induced pain is associated with central neurogenic mechanisms of sensitization and has been used to model chronic pain states.28,31 These recent reports increase our confidence that capsaicininduced pain models important mechanisms that are relevant to the study of clinical pain conditions, a criticism of many laboratory studies.32 The level of arousal was manipulated by using wellaccepted laboratory methodologies; a Stroop task was used to produce psychological stress/arousal, and guided imagery combined with passive progressive
161 relaxation was used to produce relaxation. Differences in cardiovascular and neuroendocrine responses may reflect potential explanatory mechanisms for the effect of pain-unrelated stress on pain. Norepinephrine and cortisol were measured because they have been associated with pain modulation.33-36 Attention was given in this prospective study to control for hormonal fluctuations on pain levels by including only women in the follicular phase of their menstrual cycle (days 3 to 10).37 Changes in both the subjective experience of relaxation and in physiologic response to the task were measured and used as potential predictors of pain perception/modulation. This design allowed us to verify whether changes in pain perception, in response to interventions not directly related to pain, can be explained by nonspecific changes in the level of physiologic activity reflecting a “single arousal dimension” and to evaluate whether these effects were associated with changes in subjective levels of arousal/relaxation. The effect of this experimental manipulation on the flare response has been the object of a separate report.38
Method Subjects Subjects were healthy men and premenopausal women between the ages of 19 and 48 years recruited from advertising in local newspapers and screened by telephone to determine eligibility as previously described.38 Screening criteria were designed to exclude subjects with inflammatory or immunomodulatory conditions and those taking medications that potentially could affect the physiologic outcome parameters. Subjects with conditions such as diabetes, multiple sclerosis, cancer, rheumatoid arthritis, autoimmune conditions, fibromyalgia, chronic fatigue syndrome, cancer, organ transplant, lupus, eczema, allergy to bee stings, pregnancy, or cardiac or respiratory conditions that would put the patient at potential risk after a capsaicin injection were excluded. Use of birth control pills or other hormonal medication, immunomodulatory medications (eg, corticosteroids), β-adrenergic receptor antagonists, psychotropic medication, chronic use of antihistamines, and cigarette consumption also were grounds for exclusion. Subjects with current or past history of major psychiatric illness or anxiety disorder, hospitalization within the past 6 months, or an acute infectious illness within the last month were excluded. All women participants were premenopausal. Sixty-six subjects met initial screening criteria to be included in the relaxation training part of the study. Ten subjects (6 men and 4 women) did not meet the relaxation criteria (see following). One subject dropped out before the start of the study, and 5 dropped out during the relaxation training for reasons of inconvenience or scheduling difficulties. Of the remaining 50 participants, 44 had complete data relevant to this report. (Three aliquots of blood were hemolyzed at baseline and thus were unavailable for
162 norepinephrine measure. One of these was from the relaxation group and 2 were from the stress group. There were 2 subjects in the stress group from whom blood could not be drawn after baseline, and blood of 1 control subject was hemolyzed and not available to be tested for norepinephrine.) Subjects were assigned randomly to condition as follows: stress (n = 16 [9 male, 7 female]), relaxation (n = 19 [10 male, 9 female]), and control (n = 9 [6 male, 3 female]).
Procedure The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the Institutional Review Board of the University of Iowa.
Training Volunteers were screened by telephone to assess eligibility. Potentially eligible subjects completed an informed consent and attended 3 separate group relaxation training sessions held during a 1-week period. Subjects were trained in relaxation to ensure that those to be included in the relaxation group could indeed achieve a predetermined level of relaxation. This procedure was applied to all subjects before randomization (1) to avoid the sampling bias that the exclusion of subjects unable to relax only from the relaxation group would have introduced and (2) to provide homogenous pre-experimental conditions across groups and thereby prevent the introduction of group differences associated with the familiarity with the experimental context and the development of specific expectancies. At each session, a finger temperature sensor was attached to the tip of the middle finger of the nondominant hand. Temperature was measured every minute while participants listened to a 20-minute relaxation tape through individual tape recorders with headphones. In addition, they were told to keep their eyes closed. The tape included previously validated instructions for imagerybased passive progressive relaxation. Participants were accepted into the study if they were able to achieve a 2º increase in finger temperature during at least 2 of the 3 relaxation training sessions.39 Potential participants who did not meet the temperature change criterion were offered 2 additional training sessions and were included in the study if they were able to achieve a 2º increase in finger temperature twice across the 5 sessions.
Experimental Session Participants were scheduled for a study appointment and asked to refrain from exercise, consumption of alcohol, and ingestion of nonprescription medication for 24 hours and caffeine for 15 hours before the study. Women were scheduled during the follicular phase of their menstrual cycle (days 3 to 10). The experimental session was held in the early evening (starting at 5:30 PM). Women were given a pregnancy test before initiation of the experiment and excluded if pregnant. Subjects were then seated in a comfortable chair,
Stress Effects on Capsaicin-Induced Pain instructions given, and a catheter was inserted. Using previously published methodology, the site for the capsaicin injection was marked on the volar forearm of the subjects.29 Subjects completed questionnaires and read magazines for a total of 22 minutes before the baseline period started. The experimental session was divided into 3 stages: baseline, experimental task, and capsaicin pain. The baseline consisted of 8 minutes during which subjects continued to read a magazine quietly before the experimental tasks. At the end of the baseline period, subjects completed a relaxation/tension inventory, blood was sampled from the catheter, and subjects began the experimental task. The stress group was administered a 20-minute computerized version of the Stroop colorword test,40 which has been shown to reliably produce sympathetic activation.41 Participants were urged to improve their performance by the examiner at 5minute intervals. Those in the relaxation group listened to the relaxation tape used previously in the relaxation training sessions, using a hand-held portable tape recorder and large headphones to eliminate any distracting noise. Subjects in this condition were instructed to keep their eyes closed. Those in the control group watched a 20-minute neutral video of a nonfiction historical account of “The Bridges and Ferries of Iowa City” narrated by a local historian. After the completion of each task, participants completed the relaxation inventory, had a second blood sample drawn, and received the capsaicin injection. Less than 30 seconds elapsed from completion of the task to the capsaicin injection. Capsaicin (100 µg) was injected into the volar surface of the nondominant forearm at the point previously marked.
Capsaicin The capsaicin preparation consisted of 100 µg of capsaicin (8-methyl N vanillyl 6-nonamide) suspended in 10 µL of a polyoxyethylene (20) sorbetan mono-oleate (Tween 80) saline vehicle. A stock solution of capsaicin was prepared containing 10% capsaicin in ethanol (wt by volume) by using previously published methodology.29 Two-milliliter aliquots were withdrawn and diluted to 1% with ethanol. The ethanol then was removed by vacuum, and the capsaicin was dissolved in 0.14 mL of 7.5% Tween 80 by weight in saline. This mixture was brought into a colloidal mixture with 1.86 mL of 0.9% saline by sonication and then injected through a Millipore filter (0.2 mm pore size) (Millipore Corporation, Bedford, MA) into a sterile injection vial. The vial was frozen at –20°C until use. Before using, the vial was warmed to room temperature and shaken by hand for 30 seconds.
Measures Cardiovascular Activity Cardiovascular measures were acquired every 2 minutes for the 8 minutes in the baseline period and every
ORIGINAL REPORT/Logan et al 4 minutes for the 20 minutes of the experimental task (5 measures starting 2 minutes after task initiation). Heart rate (HR), systolic blood pressure (SBP), and diastolic blood pressure (DBP) were obtained by using a Critikon Dinamap Plus (Johnson & Johnson, New Brunswick, NJ) vital signs monitor applied to the nondominant arm.
Cortisol Blood was sampled at the end of the baseline period and at the end of the experimental task immediately before the injection of capsaicin. Cortisol was measured as an indicator of the hypothalamic pituitary adrenocortical (HPA) response to the stressor.42 Aliquots of serum were separated from blood samples and assayed by competitive radioimmunoassay (EURO/DPC’s Double Antibody Cortisol RIA; Diagnostic Products Corporation, Los Angeles, CA). Approximate sensitivity of the assay is 0.3 µg/dL. Interassay coefficients of variability range from 5.8% to 7.8%, and intra-assay coefficients of variability range from 2.4% to 4.5%.
Norepinephrine Norepinephrine was measured from the same blood samples as cortisol as an indicator of sympathetic nervous system activity.43 Plasma norepinephrine (NE) was measured in the Cardiovascular Center Core Laboratory by using high-performance liquid chromatography with electrochemical detection (HPLC-EC). Catecholamines in the sample were adsorbed onto acid-washed alumina and eluted with 0.1 mol/L perchloric acid. After microfiltration, the eluate was buffered and underwent chromatography on a Biophase ODS (C-18) Keystone Scientific Inc, Belafonte, PA) HPLC column (5 µm, 250 mm × 4.6 mm) with a mobile phase of 0.1 mol/L KH2PO4, 0.1 mmol/L ethylenediamine tetra-acetic acid, 7.8% methanol, and 4 mmol/L heptane sulfonic acid as the ion-pairing agent. Catecholamines were detected with a BAS LC-4B (Bioanalytical Systems, Lafayette, IN) electrochemical detector at an applied potential of +650 mV by using a glassy carbon-working electrode (BAS). Peaks were quantitated on a Shimadzu CR 3-A (Shimadzu Corporation, Columbia, MD) integrator. Sensitivity of the HPLC-EC system was 50 pg for norepinephrine.
163 pain on a 100-mm visual analog scale (VAS) anchored with “no pain” to “pain as bad as you can imagine.”1
Data Reduction and Analytic Strategy Data analysis used the Statistical Package for the Social Sciences (SPSS, Chicago, IL) version 7.5 for PC and SAS version 7.0 (SAS Institute, Cary, NC) for PC. The 4 cardiovascular measures obtained in the baseline period before the experimental task were averaged. Mean levels of cardiovascular measures during the experimental task were calculated for each group by averaging all 5 measures obtained during the experimental task. Delta values were calculated as the change from baseline to task for all measures.45 Equivalence of groups at baseline was examined by using 2-way analyses of variance (ANOVA) to test the effects of task, sex, and sex by task. Next, the adequacy of the experimental manipulation was checked by comparing task-related changes in the 3 groups (ANOVA). Tukey’s post hoc tests were performed on significant ANOVAs to determine pair-wise differences between conditions. A second set of analyses examined the effects of the experimental manipulation on the maximum pain to examine group differences, sex differences, and group by sex interaction. Clinicians and patients tend to be most concerned about maximum pain in any clinical pain setting; thus, the maximum pain rating observed within the first 3 minutes after capsaicin injection was used as the dependent measure in these analyses. These analyses were repeated by using initial pain at 7 seconds after injection and the area under the curve (AUC) for the first 3 minutes of pain measurements. Pearson correlation analyses were conducted on self-report measures (RI), physiologic measures of arousal, and maximum pain. Delta mean scores of postexperimental arousal task and pretask were used in these analyses. Z tests on Fisher transformed Pearson correlation coefficients were used to test for differences between correlations.46 Finally, a series of hierarchical regression analyses were conducted to determine potential mediators responsible for any differences or interactions on maximum spontaneous pain.
Results
Relaxation
Sample Characteristics
The subjective level of relaxation was assessed by using the relaxation inventory (RI) at the end of the baseline period and immediately after the task before the capsaicin injection.44 Items were summed to obtain an overall relaxation score, with higher scores indicating more relaxation/less subjective arousal. Test-retest reliability for this measure is between 0.87 and 0.97.
Subjects had a mean age of 30.20 (SD = 9.02) years and a median income in the range of $20,000 to $30,000. The majority of subjects (93.7%) were white. A large percentage of subjects (88%) had at least some college education. There were no significant differences among the 3 groups (stress, relax, or control) on any demographic variables (eg, age, sex, income, education) (all P values > .42). There were no sex, task, or sex by task effects on baseline measures of SBP, DBP, HR, NE, cortisol, and RI with one exception. There was a trend for men (mean [SD] = 117.8 [11.4]) to have higher SBP than women (mean [SD] = 108.8 [12.2]; F [2, 38] = 3.83, P = .06).
Pain Measures of spontaneous pain were taken every minute beginning at 7 seconds after the injection of capsaicin for 20 minutes and every 5 minutes thereafter until 60 minutes. Participants were asked to mark their
164
Stress Effects on Capsaicin-Induced Pain
Table 1. Mean (SEM) Task-Induced Changes in SBP, DBP, HR, NE, Cortisol, and RI Scores in the Stress, Relaxation, and Control Groups, With the Corresponding F Value for the Main Effect of Task (ANOVA) and the Results of the Post Hoc Comparisons (Tukey)
∆SBP ∆DBP ∆HR ∆NE ∆Cortisol ∆RI
STRESS (S)
RELAX (R)
CONTROL (C)
F [2, 42]
11.8 (2.8) 19.8 (1.5) 8.3 (1.3) –2.9 (19.3) 1.6 (1.0) –27.2 (5.6)
–5.3 (1.2) 6.2 (1.0) –4.3 (1.1) –21.3 (10.1) –1.2 (0.75) 14.1 (4.8)
–1.7 (1.8 ) 10.2 (1.3) –1.7 (4.3) 19.3 (12.1) –1.4 (0.72) 1.4 (3.8)
19.58*** 32.12*** 33.36*** 2.44* 3.42** 18.80***
TUKEY S>R** S>R** S>R** All not significant S>R* S>R**
*P < .10. **P < .05. ***P < .01.
Manipulation Checks We performed separate 2-way (sex by task) analyses of variance on task-related changes for each of the arousal measures (SBP, DBP, HR, NE, cortisol, and RI). For each measure, there was an effect of task (although it was marginally significant for NE, P < .10) but no effect for sex or interaction between sex by task (Table 1). Post hoc analyses revealed that the stress condition resulted in greater arousal than either the relaxation or the control condition on SBP, DBP, HR, and RI (P < .05), but the relaxation condition did not differ from the control (all P values > .10). The same pattern was true for cortisol (P < .10), but the differences did not reach the criterion for statistical significance.
Spontaneous Pain The mean for spontaneous pain across the sample was 59.0 mm (SD = 19.6) on the 100-mm VAS, with 2 participants rating the pain as 100 (“pain as bad as you can imagine”). Examination of the profile of pain ratings over time confirmed the rapid decrease (within a few minutes) typically observed after capsaicin injections. After 3 minutes, mean and standard deviation for pain ratings were 36.6 (19.6). There was a significant interaction between sex and the experimental condition (F [2, 43] = 4.07, P = .02). (The interaction between sex and conditions was confirmed on the initial pain rating [P < .05] and was marginally significant on the AUC for the first 3 minutes of measurements [P = .06]. This effect did not reach significance when later points are included in the analysis of the AUC [P > .10]). Follow-up Tukey tests showed that stressed women showed significantly greater maximum pain than did stressed men (P < .01). Women in the relax condition showed significantly less pain than women in the stress condition (P < .05) (Fig 1). There were no other significant differences. Baseline cortisol, NE, DBP, SBP, and HR were not significantly correlated with maximum pain (all P values > .2). Delta RI was negatively correlated with maximum pain (r = –0.35), indicating that the more subjects reported being relaxed the less pain they described. In addition, ∆NE (r = –0.25) and ∆ cortisol (r = –0.28) were negatively correlated with maximum pain report. Delta
SBP, ∆DBP, and ∆HR were not significantly correlated with maximum pain report. Table 2 shows correlations between indexes of arousal (∆SBP, ∆DBP, ∆HR, ∆NE, ∆cortisol, and ∆RI) and maximum pain for the whole sample and for men and women separately. The ∆RI was inversely related to maximum pain (r = –0.63) in women but not in men (r = 0.13), a difference that reached statistical significance (Fisher transformation t = 7.9, P < .01). Although not significantly different from each other, ∆NE was inversely associated with maximum pain (r = –0.42) in men but not women (r = –0.06). Likewise, ∆DBP was inversely associated with pain in men (r = –0.46, P < .05) but not in women (r = 0.14, P > .1). These latter correlation coefficients were marginally different from each other (Fisher transformation t = 1.94, P > .1).
Regression Models To better understand the factors involved in the interaction between sex and the experimental task on maximum pain, multiple regression analysis was used. Based on prior research and our correlation analysis, variables were selected as predictors of pain
Fig 1. Mean (+SEM) maximum pain evoked by capsaicin after a cognitive stress, relaxation, or a control neutral procedure in men and women. P values are given for the significant post hoc comparisons examining the interaction between sex and tasks.
ORIGINAL REPORT/Logan et al
165
Table 2. Pearson Product Moment Correlations Between Maximum Pain and Task-Related Changes in SBP, DBP, NE, Cortisol, and RI Scores in the Whole Sample and for Men and Women, With the Corresponding Result of the Fisher Transformation Evaluating the Difference Between the Correlation Coefficients in Men and Women WHOLE SAMPLE (N = 44) ∆SBP ∆DBP ∆HR ∆NE ∆Cortisol ∆RI
–0.05 –0.15 0.11 –0.25** –0.28** –0.35**
MEN (N = ?) –0.24 –0.46** –0.10 –0.42** –0.29* 0.13
WOMEN (N = ?)
FISHER TRANSFORMATION (MEN
0.38* 0.14 0.34 –0.06 –0.31* –0.63***
V
WOMEN)
NS * NS * NS ***
Abbreviation: NS, Not significant. *P < .10. **P < .05. ***P < .01.
and separate models were built predicting maximum pain from the physiologic (NE and cortisol) and the self-report measures of arousal (RI). In each model, the variables were entered in the following manner. First, baseline measures and sex were entered as control variables. The delta scores of the arousal variables were then entered. The next step included the interaction term of sex by the delta arousal measure(s). The final step was the condition and sex by condition interaction term. This allowed us to determine whether the sex by condition interaction was explained by the arousal measures.
Model 1: Self-Report Measures The interaction term between sex and task-related ∆RI significantly predicted maximum pain (β = –1.29, P = .03). Although the main effects of baseline RI, sex, and ∆RI were entered into the model, they did not significantly contribute to predict pain. The sex by condition interaction was no longer a significant predictor of maximum pain after the interaction term sex by ∆RI was entered into the model. This model accounted for 35% of the variance in maximum pain.
Model 2: Physiologic Measures Task-induced changes in NE (β = –.44, P = .007) and cortisol (β = –.49, P = .002) were significant predictors of maximum pain, over and above the significant relationship of baseline cortisol with maximum pain (β = –.51, P = .002). However, the sex by condition interaction (change in R2 = 8%; β = –1.13, P = .03) was still a significant predictor of maximum pain after changes in NE and cortisol were entered into the model, indicating that physiologic indexes of arousal could not totally account for this interaction. This total model accounted for 43% of the variance in maximum pain. The interactions of sex by ∆cortisol and sex by ∆NE were forced into the model but did not predict maximum pain (P > .10) and thus were dropped from further analysis. We also tested cardiovascular measures (baseline and deltas) in a similar regression model; however, even when forced into the model, these values did not pre-
dict maximum pain and were also dropped from further analysis.
Model 3: Combined Model Table 3 shows that both physiologic and self-report measures predicted maximum pain. As before, baseline cortisol and changes in cortisol and NE were associated with maximum pain (model 2). Neither the change in relaxation score by sex nor the sex by condition interaction reached significance in this combined model. However, this combined model was better at predicting pain than either the physiologic model or the selfreport model alone (in both cases P < .01), accounting for 56% of the variance in maximum pain. The physiologic measures were not correlated with the self-report measures of arousal (P > .1), thereby explaining their independent effects on maximum pain.
Discussion There were 4 major findings from this study. First, the manipulation of arousal by the experimental tasks interacted with sex of subjects to predict pain. These results were consistent whether maximum or initial pain report was used, making these findings especially relevant to clinical settings in which acute painful procedures are used. Stressed women reported significantly greater pain than both men in the stress condition and women in the relaxation condition. A second major finding was that higher levels of physiologic arousal (∆DBP, ∆NE, and ∆cortisol) were related to lower maximum pain report, most consistently in men (∆DBP and ∆NE). Third, self-reported arousal (decrease in subjective relaxation; ∆RI) was associated positively with maximum pain only in women. Finally, physiologic and self-reported arousal independently predicted maximum pain, and a model including both accounted for 56% of the variance in maximum pain. Both self-report and physiologic indexes showed that the experimental manipulation was effective in so far as the stress condition produced more arousal than the control or relaxation condition. However, there were
166
Stress Effects on Capsaicin-Induced Pain
Results of the Regression Analysis Testing Model 3: Pain Regressed on Baseline Cortisol, NE, and R; ∆RI; ∆Cortisol; ∆NE; Sex; Sex by ∆RI; Condition; and Sex by Condition Table 3.
STANDARDIZED Cortisol baseline NE baseline RI baseline Sex ∆Cortisol ∆NE ∆RI Sex by ∆RI Condition Sex by condition
–0.45 –0.15 0.23 0.38 –0.42 –0.37 0.44 –0.72 0.42 –0.49
B
T
STATISTIC –3.16 –1.11 1.78 0.97 –2.79 –2.65 0.85 –1.36 1.01 –0.91
no differences in arousal indexes between the control and relaxation groups. There are 2 possible explanations for this lack of difference. First, the small number of subjects limited the statistical power to detect effects, and, second, the control group also was trained in relaxation and may have used this strategy while watching the neutral control video. The finding that stressed women reported more pain than stressed men is consistent with a large body of research showing that women experience more pain than men.47 Although a number of studies have looked at the association between stress and pain report, to our knowledge, only 2 other studies have examined sex differences in pain report after stress. Bragdon et al17 observed that women with low BP reported more thermal pain before stress than men with low BP, but there were no sex differences after the stress task. Rhudy and Meagher15 found no sex difference in heat pain threshold after a stress manipulation. It is possible that the observed differences between our findings and these previous studies are related to the use of different pain tasks and measures in each study. The pain produced by a discrete subcutaneous injection of capsaicin is inescapable—it peaks rapidly and thereafter decreases slowly. In contrast, pain produced by the sustained application of radiant or contact heat stimuli is escapable, and both threshold and tolerance tests depend on active motor/behavioral responses. The possibility or impossibility to escape pain and the differential motor/behavioral requirements of the pain measurements may lead to differences in the active/passive coping strategies adopted by men and women. The coping literature shows sex-specific differences in the use of specific cognitive strategies.48,49 For instance, women report greater use of catastrophizing than men, and this coping strategy has been associated with increased pain report.50 Both cognitive stress and inescapable pain may promote catastrophizing and may have contributed to the higher pain reported in women in the stress group. Sex differences in stressinduced pain modulation should be examined within the same study by using multiple pain tests and measurements to evaluate this possibility.
P VALUE .003 .280 .080 .340 .009 .010 .400 .180 .320 .370
R2 0.12
0.49
0.55 0.56
Another possibility to explain the difference between our findings and those of previous studies is that the relaxation training procedure used in the present study may have shifted the sex-related effects in pain. Because all subjects were familiar with the experimental setting and were trained in relaxation, subjects in all 3 groups might have been at an overall lower baseline stress level in the experimental phase than in other studies. This training may have contributed to mask the typical higher pain sensitivity of women in the relaxation and control groups. This effect would have been counteracted by the stress task lending to the expected sex difference in pain. We also found that women in the stress condition reported more pain than women in the relaxation condition. Given that the induction of stress was the most effective manipulation in this study, these results suggest that women displayed stress-induced hyperalgesia. However, the difference in pain between the stress and the control conditions did not reach significance and, therefore, our results cannot exclude the possibility that relaxation may have contributed to part of the difference in pain between women in the stress and relaxation groups. Only 1 other study has compared and found similar effects of both stress and relaxation on pain report, but it did not examine whether these effects were interactive with the sex of the subject.51 Consistent with our results, Houston and Jesurum52 examined the effects of relaxation on chest tube removal pain in a pilot study and found a trend toward less pain during relaxation in women but not in men. Our results and those of Houston and Jesurum are consistent with previous findings showing that stress or relaxation irrelevant to pain may modulate pain perception and further indicate that these effects differ in men and women. Physiologic measures of task-induced changes in arousal (∆DBP, ∆NE, and ∆cortisol) were related inversely to pain report. The relationship between pain and both DBP and NE only was seen in men, whereas the relationship with cortisol was equally strong in men and women. Previous studies have shown that NE has analgesic effects.35 For instance, NE has been implicated
ORIGINAL REPORT/Logan et al in opiate analgesic mechanisms,36 and sex differences in opiate analgesia are reported frequently.53 Thus, our data may support a sex-specific analgesic effect. Few previous studies, however, have examined this relationship, and the underlying mechanisms are not yet established.53 Other studies, including those we have conducted, found relationships between baseline blood pressure and pain.23,24 In this study, however, we do not find such a relationship in any of the subgroups by sex by condition or across the whole sample, which may be related to lack of statistical power. There is some indication that blood pressure reactivity, indicative of sympathetic arousal, is related to pain in a sex-specific fashion; however, these findings are equivocal.18,54 In our study, changes in DBP were associated significantly with pain in men but not in women, and these correlations were only marginally different from each other. In contrast to physiologic arousal, self-reported arousal (decreased RI) was associated positively with pain. This is consistent with other previously reported research.6 In our study, however, the relationship was only significant in women. Our data are consistent with Fillingim et al25 who also found significant associations between psychological measures and pain report in women but not men. Thus, physiologic and psychological measures of arousal are differentially related to pain reports in men and women. These sex-specific effects in the correlation analyses prompted further examinations of the factors that could account for the sex-by-condition interaction on pain. Regression analysis confirmed that this interaction was explained by changes in self-reported arousal (RI scores; model 1)—that is, the sex by condition interaction was no longer significant when this factor was entered into the model. In contrast, physiologic arousal did not explain the sex-by-condition interaction (model 2). Thus, the manipulation effects we observed were better explained by the self-report measure than the physiologic measures. Moreover, the physiologic and self-report measures were independently associated with pain report. The combined model that contained all variables of interest predicted 56% of the variance in pain report model 3). Because we are slightly below the convention of a 5:1 variable subject (or variable to subjects?) ratio on the combined model,55 results should be interpreted cautiously and should be replicated in future research. On the other hand, the rule of 5:1 usually is interpreted for independent variables, and several of our variables are control (baseline) variables, increasing our confidence in the validity of this model. Overall, these findings suggest that stress-induced pain modulation may not be explained fully by a unidimensional construct of physiologic arousal and that subjective measures of arousal reflect, at least in part, different processes. Studies of the effect of emotion on pain have focused on fear and anxiety and largely have ignored the possible contribution of other subjective emotional states to the experience of pain. For example, in the study by Rhudy and Meagher,15 negative
167 emotions were elicited by the administration of electric shocks or by the threat of impending shocks to evaluate the effects of fear and anxiety, respectively, on heat pain threshold. These manipulations led to decreased and increased pain thresholds, respectively, and the investigators interpreted those findings as reflecting an analgesic effect associated with states of high arousal (fear) and an hyperalgesic effect associated with states of moderate arousal (anxiety). However, this study also included subjective ratings of arousal and basic emotions, which revealed that the experimental manipulations produced distinct patterns of negative emotional feelings. The “fear condition” was reported to elicit experiences of fear, anger, surprise, and disgust, whereas the “anxiety condition” elicited fear and decreased surprise relative to a neutral condition. These reports suggest that the pain modulation observed may not reflect simply the level of arousal but may relate to the emotion experienced. Therefore, it is reasonable to suggest that the experiences of anger, surprise, and disgust, which were almost exclusively reported in the “fear” condition, may contribute to the observed emotion-induced analgesia (a possibility not assessed directly by Rhudy and Meagher). Although our experiment did not include subjective ratings of emotions, the stress task may have produced different sets of emotions in men and women. These possibilities need to be examined more closely in experimental paradigms that manipulate not only fear and anxiety but also other emotions and use subjective ratings of emotional experience as potential predictors of pain. The interaction between sex and task, the stronger association found between pain and subjective reports of relaxation in women and physiologic activity in men, and the independent portion of variance in pain that the subjective and physiologic measures accounted for in the regression models imply that different modulatory mechanisms may have been at work in men and women. Assuming that the subjective reports of relaxation reflect, at least in part, brain mechanisms involved in the “conscious perception of one’s own state,” whereas physiologic activation may occur largely subliminally, we speculate that pain modulation in women may be more proximal to central nervous system mechanisms associated with the subjective experience of being stressed or relaxed. In contrast, pain modulation in men was not related to their subjective reports of relaxation in this experiment and was more closely related to physiologic indexes of arousal. The combination of these effects suggests an association between brain mechanisms involved in the subjective perception of pain and stress (or reduced relaxation) in women and a possible dissociation of such processes in men. The investigation of central nervous system mechanisms of pain perception and modulation in humans is in its infancy, and although men and women present largely similar patterns of brain responses to experimental pain, some differences have been noted.56 Men and women may differ not only in pain-related activa-
168
Stress Effects on Capsaicin-Induced Pain
tion but also in the mobilization of central modulatory processes. Recent studies suggest important individual differences in the engagement of descending modulatory mechanisms,57 some of which have been associated with the sex of the individual.58,59 Our results may reflect the differential engagement of those mechanisms during stress in men and women. These results with maximum capsaicin-induced pain also may have clinical implications for patients, particularly women, undergoing painful procedures. Preprocedural arousal may influence the amount of pain that subjects experience. Minimization of high levels of arousal before procedures as adjuncts to conventional analgesics may aid pain management and potentially speed recovery, particularly in women.60,61 Cognitive tasks performed in the laboratory may approximate the effect of the numerous sources of stress unrelated to pain that are inherent to normal human environments including the clinic. Several important conclusions can be drawn from this study. First, the manipulation of stress/arousal levels modulated the pain experience, but the direction of the effect depended on the sex of the individual. Second, self-report and physiologic measures of arousal indepen-
dently predicted pain, and those effects were again different in men and women. Taken together, these findings suggest that a unidimensional model of arousal may be insufficient to explain individual and stress-related differences in pain. Future research on stress-induced pain modulation should consider the multidimensionality of stress (physiologic and subjective experience) and its impact on emotion and coping. These emotions and coping strategies may engage central modulatory processes differentially in men and women and may contribute to the frequently found sex differences in pain.
References
11. Olson GA, Olson RD, Kastin AJ: Endogenous opiates: 1996. Peptides 18:1651-1688, 1997
1. Price DD: Psychological Mechanisms of Pain and Analgesia, volume 15. Seattle, IASP Press, 1999, pp 6-9 2. Caceres C, Burns JW: Cardiovascular reactivity to psychological stress may enhance subsequent pain sensitivity. Pain 69:237-244, 1997 3. Logan H, Baron RS, Kohout F: Sensory focus and procedural information as therapeutic treatment for dental stress. Psychosom Med 57:475-484, 1995 4. Baron RS, Logan H, Hoppe S: Emotional and sensory focus as mediators of dental pain among patients differing in desired and felt dental control. Health Psychology 12:381-389, 1993 5. Cornwall A, Donderi DC: The effect of experimentally induced anxiety on the experience of pressure pain. Pain 35:105-113, 1988 6. Logan H, Baron R, Keeley K, Law A, Stein S: Desired and felt control as mediators of stress in a dental setting. Health Psychology 10:352-359, 1991 7. Ham LP, Packard RC: A retrospective, follow-up study of biofeedback-assisted relaxation therapy in patients with posttraumatic headache. Biofeedback & Self Regulation 21:93-104, 1996 8. Sherman RA, Gall N, Gormly J: Treatment of phantom limb pain with muscular relaxation training to disrupt the pain-anxiety-tension cycle. Pain 6:47-55, 1979 9. Dougher MJ, Goldstein D, Leight KA: Induced anxiety and pain. J Anxiety Disord 1:259-264, 1987 10. Weisenberg M, Aviram O, Wolf Y, Raphaeli N: Relevant and irrelevant anxiety in the reaction to pain. Pain 20:371383, 1984
Acknowledgments We would like to acknowledge Dr Erling Anderson for the generous use of equipment; Dr Winston Barcellos for medical advice; Mr Mark Morrison, Dr Carol Hodne, and Dr Robert Wade for assistance in data collection; Dr H. Lester Kirchner for advice on data analysis; Dr Lloyd Matheson for assistance with the capsaicin preparation; Donna Farley, Senior Research Assistant; and the director and staff of the Clinical Research Center at the University of Iowa.
12. Bandura A, Cioffi D, Taylor CB, Brouillard ME: Perceived self-efficacy in coping with cognitive stressors and opioid activation. J Pers Soc Psychol 55:479-488, 1988 13. Lichtman AH, Fanselow MS: Cats produce analgesia in rats on the tail-flick test: Naltrexone sensitivity is determined by the nociceptive test stimulus. Brain Res 12:9194, 1990 14. Fanselow MS: Shock-induced analgesia on the formalin test: Effects of shock severity, naloxone, hypophysectomy, and associative variables. Behav Neurosci 98:79-85, 1984 15. Rhudy JL, Meagher MW: Fear and anxiety: Divergent effects on human pain thresholds. Pain 84:65-75, 2000 16. Maixner W, Humphrey C: Gender differences in pain and cardiovascular responses to forearm ischemia. Clin J Pain 2:16-25, 1993 17. Bragdon EE, Light KC, Girdler SS, Maixner W: Blood pressure, gender and parental are factors in baseline and poststress pain sensitivity in normotensive adults. Int J Behav Med 4:17-38, 1997 18. Nyklicek I, Vingerhoets AJJM, Van Heck GL: Hypertension and pain sensitivity: Effects of gender and cardiovascular reactivity. Biol Psychol 50:127-142, 1999 19. Weinstock M, Razin M, Schoerer-Apelbaum D, Men D, McCarty R: Gender differences in sympathoadrenal activity in rats at rest and in response to footshock stress. Int J Dev Neurosci 16:289-295, 1998 20. Livezey GT, Miller JM, Vogel WH: Plasma norepinephrine, epinephrine and corticosterone stress responses to restraint in individual male and female rats, and their correlations. Neurosci Lett 62:51-56, 1985
ORIGINAL REPORT/Logan et al 21. Polefrone JM, Manuck SB: Gender differences in cardiovascular and neuroendocrine response to stressors, in Barnett RC, Biener L, Baruch GK (eds): Gender and Stress. New York, Macmillan, 1987, pp 13-38 22. Light KC, Turner JR, Hinderliter AL, Sherwood A: Race and gender comparisons: Hemodynamic responses to a series of stressors. Health Psychol 12:354-365, 1993 23. Fillingim RB, Maixner W: The influence of resting blood pressure and gender on pain perception. Psychosom Med 58:326-332, 1996 24. Sheffield D, Biles PL, Orom H, Maixner W, Sheps DS: Race and sex differences in cutaneous pain perception. Psychosom Med. 62:517-523, 2000 25. Fillingim RB, Keefe F, Light KC, Booker DK, Maxiner W: The influence of gender and psychological factors on pain perceptions. J Gender Culture Health 1:21-36, 1996 26. Gracely RH: Studies of pain in man, in PD Wall, Melzack R (eds): Textbook of Pain (ed 3). Edinburgh, Scotland, Churchill Livingstone, 1994, pp 303-313 27. Baumann TK, Simone DA, Shain CN, LaMotte RH: Neurogenic hyperalgesia: The search for the primary cutaneous afferent fibers that contribute to capsaicin-induced pain and hyperalgesia. J Neurophysiol 66:212-226, 1991 28. Iadarola MJ, Berman KF, Zeffiro TA, Byas-Smith MG, Gracely RH, Max MB, Bennett GJ: Neural activation during acute capsaicin-evoked pain and allodynia assessed with PET. Brain 121(pt 5):931-947, 1998 29. LaMotte RH, Shain DN, Simone DA, Tsai EF: Neurogenic hyperalgesia: Psychophysical studies of underlying mechanisms. J Neurophysiol 66:190-211, 1991 30. Jansco G, Dux M, Santha P: Role of capsaicin-sensitive afferent nerves in initiation and maintenance of pathological pain. Behav Brain Sci 20:454-455, 1997 31. Morris VH, Cruwts SC, Kidd BL: Characterization of capsaicin-induced mechanical hyperalgesia as a marker for altered nociceptive processing in patients with rheumatoid arthritis. Pain 71:179-186, 1997 32. Berkeley KJ: From psychophysics to the clinic. Pain Forum 4:225-227, 1995 33. Kuraishi Y, Harada Y, Takagi H: Noradrenaline regulation of pain-transmission in the spinal cord mediated by αadrenoceptors. Brain Res 174:333-336, 1979 34. Melzack R Katz J: Pain measurement in persons in pain, in Wall P, Melzack R (eds): Textbook of Pain (ed 3). Edinburgh, Scotland, Churchill Livingstone, 1994, pp 337-348 35. Furst S: Transmitters involved in antinociception in the spinal cord. Brain Res Bull 48:129-141, 1999 36. Yaksh TL: Pharmacology and mechanisms of opioid analgesic activity. Acta Anasthesiol Scand 41:94-111, 1997 37. Riley JL, Robinson ME, Wise EA, Price DD: A meta-analytic review of pain perception across the menstrual cycle. Pain 81:225-235, 1999 38. Lutgendorf S, Logan H, Kirchner HL, Rothrock N, Svengalis S, Iverson K, Lubaroff D: Effects of relaxation and stress on the capsaicin-induced local inflammatory response. J Psychom Med 62:524-534, 2000
169 39. Schwartz MS: Biofeedback: A Practitioner’s Guide. New York, Guilford Press, 1987 40. Klein GS: Semantic power measured through the interference of words with color naming. Am J Psychol 77:576588, 1964 41. Frankenhauser M, Mellis J, Rissler A, Bjorksell C, Patkar P: Catecholamine excretion as related to cognitive and emotional reaction patterns. Psychosom Med 30:109-120, 1968 42. Kuhn CM: Adrenocortical and gonadal steroids in behavioral cardiovascular medicine, in Scheneiderman N, Weiss S, Kaufmann P (eds): Handbook of Research Methods in Cardiovascular Behavioral Medicine. New York, Plenum, 1989, pp 185-204 43. Ziegler MC: Catecholamine measurement in behavioral research, in Scheneiderman N, Weiss S, Kaufmann P (eds): Handbook of Research Methods in Cardiovascular Behavioral Medicine. New York, Plenum, 1989, pp 167-183 44. Crist DA, Rickard HC, Prentice-Dunn S, Barker HR: The relaxation inventory: Self-report scales of relaxation training effects. J Pers Assess 53:716-726, 1989 45. Llabre NM, Spitzer SB, Saab PG, Ironson GH, Schneiderman N: The reliability and specificity of delta versus residualized change as measures of cardiovascular reactivity to behavioral challenges. Psychophysiology 28:701711, 1991 46. Hays WL: Statistics for the Social Scientist. New York, Holt Rinehart and Winston, 1973 47. Unruh AM, Ritchie J, Merskey H: Does gender affect appraisal of pain and pain coping strategies? Clin J Pain 15:31-40, 1999 48. Otto MW, Dougher MJ: Sex differences and personality factors in responsivity to pain. Percept Mot Skills 61:383390, 1985 49. Lang PJ, Greenwald MK, Bradley MM, Hamm AO: Looking at pictures: Affective, facial, visceral, and behavioral reactions. Psychophysiology 30:261-273, 1993 50. Robinson ME, Riley JL 3rd, Myers CD: Psychosocial contributions to sex-related differences in pain responses, in Fillingim RB (ed): Sex Gender and Pain: From the Benchtop to the Clinic. Seattle, IASP Press, 2000 51. Ford MJ, Camilleri M, Zinsmeister AR, Hanson RB: Psychosensory modulation of colonic sensation in the human transverse and sigmoid colon. Gastroenterology 109:1772-1780, 1995 52. Houston S, Jesurum J: The quick relaxation technique: Effect on pain associated with chest tube removal. Appl Nurs Res 12:196-205, 1999 53. Berkley KJ: Sex differences in pain. Behav Brain Sci 20:371-380, 1997 54. Maixner W, Humphrey C: Gender differences in pain and cardiovascular responses to forearm ischemia. Clin J Pain 2:16-25, 1993 55. Hair JF Jr, Anderson RE, Tatham RL, Black RC: Multivariate Data Analysis (ed 5). Upper Saddle River, NJ, Prentice Hall, 1998, p 166 56. Paulson PE, Minoshima S, Morrow TJ, Casey KL: Gender differences in pain perception and patterns of cerebral
170 activation during noxious heat stimulation in humans. Pain 76:223-229, 1998 57. Danziger N, Fournier E, Bouhassira D, Michaud D, De Broucker T, Santarcangelo E, Carli G, Chertock L, Willer JC: Different strategies of modulation can be operative during hypnotic analgesia: A neurophysiological study. Pain 75:8592, 1998 58. France CR, Suchowiecki S: A comparison of diffuse noxious inhibitory controls in men and women. Pain 81:77-84, 1999 59. Kiernan BD, Dane JR, Philips LH, Price DD: Hypnotic
Stress Effects on Capsaicin-Induced Pain analgesia reduces R-III nociceptive reflex: Further evidence concerning the multifactorial nature of hypnotic analgesia. Pain 60:39-47, 1995 60. NIH Technology Assessment Panel: Integration of behavioral and relaxation approaches into the treatment of chronic pain and insomnia. JAMA 276:313318, 1996 61. Lang EV, Benotsch EG, Fick LJ, Lutgendorf S, Berbaum ML, Berbaum KS, Logan HL, Spiegel D: Adjunct nonpharmacologic analgesia for invasive medical procedures: A randomized trial. Lancet 355:14861490, 2000
D
uring the past decade, a substantial body of research has been devoted to understanding the relationship between pain sensitivity and the psychological state of the individual.1 However, there is still considerable disagreement as to the direction of the association between cognitive/emotional stress and pain. Studies have shown emotional stress and anxiety increase pain and that further stressing already stressed patients can produce greater pain sensitivity.2-5 For example, Logan et al6 reported that atypically stressed dental patients anticipate, experience, and remember more pain after invasive dental procedures. In addition, relaxation and biofeedback interventions that reduce Received May 1, 2000; Revised August 25, 2000; Accepted September 6, 2000. From the Division of Public Health Services and Research and Department of Medicine, University of Florida, Gainesville, FL, and Department of Psychology, Department of Neurology, Dows Institute for Dental Research, and Departments of Urology and Microbiology, University of Iowa, Iowa City, IA. Supported in part by Grant RR00059 from the General Clinical Research Centers Program, NCRR, NIH. Dr Rainville is a postdoctoral research fellow supported by the Human Frontier Science Program organization. Address reprint requests to Henrietta Logan, PhD, Division of Public Health Services and Research, Box 100404, Gainesville, FL 32610. E-mail: [email protected] © 2001 by the American Pain Society 1526-5900/01/0203-0001$35.00/0 doi:10.1054/jpai.2001.21597
stress-induced arousal have been shown to reduce pain sensitivity.7,8 However, other studies found that anxiety and stress decrease pain sensitivity.9,10 In addition, animal and human laboratory studies have shown an analgesic effect associated with stress-induced arousal.11-14 These inconsistencies question the possibility to generalize experimental findings to clinical settings. Some studies suggest that it is the psychological characteristics of the stressor—how it is perceived—that influence pain sensitivity, whereas others have focused on the effect of physiologic arousal. In this respect, the study by Rhudy and Meagher15 is particularly informative. This study suggests that fear induced by exposure to brief shock decreases pain sensitivity, whereas anxiety elicited by the threat of shock is associated with the reverse. Furthermore, results indicate that the exposure to the shock was associated with high levels of physiologic arousal, whereas the threat only produced moderate arousal, leading these investigators to conclude that pain modulation could be explained by physiologic arousal. According to these investigators, the “immediate alarm reaction to present threat” (fear) is “characterized by impulses to escape” (fight/flight), increased arousal, and decreased pain sensitivity. In contrast, “future-oriented emotion characterized by negative affect and apprehensive anticipation of potential
The Journal of Pain, Vol 2, No 3 (June), 2001: pp 160-170
160
ORIGINAL REPORT/Logan et al threats (anxiety) […] result in hypervigilance” and increased pain sensitivity. Bandura et al12 showed that low perceived self-efficacy in coping with a stressor may lead to both high levels of autonomic arousal and an opiate-mediated decrease in pain sensitivity possibly similar to the analgesia associated with fear in the study by Rhudy and Meagher.15 These effects underscore the complementary nature of psychological and physiologic interpretations and suggest that stressinduced arousal may account for pain modulation. Sex-related differences in physiologic or psychological responses might partly contribute to some of the discrepant findings on the effects of cognitive/emotional stress on pain.16-18 Sex differences in neuroendocrine response to foot shock and immobilization have been reported, with the rat sympathoadrenal system showing more reactivity in females than males.19,20 Men have higher resting blood pressure and larger blood pressure and neuroendocrine responses to stress than women.21,22 In addition, sex differences in thermal pain report are reduced after adjusting for resting blood pressure.23,24 These sex differences in pain- and stressinduced physiologic activity raise the possibility that men and women may display different patterns of pain modulation in response to interventions that alter physiologic arousal. Furthermore, different psychological characteristics are related to pain sensitivity in men and in women. For instance, Fillingim et al25 found that higher scores on perceived efficacy and control beliefs were associated with lowered pain sensitivity in women but not men, whereas lower anxiety was related to decreased pain sensitivity in men but not women. Thus, both physiologic and psychological variables may be differentially relevant to the stress-pain perception relationship in men and women. We conducted the following investigation to test the effect of pain-unrelated cognitive stress/arousal and relaxation on pain induced by a subcutaneous injection of capsaicin. We focus on the maximum pain report using the capsaicin model because of its clinical relevance.26 Capsaicin is the irritant agent found in hot chili peppers that, when injected into the skin of human subjects, produces neurogenic inflammation and evokes the sensation of intense burning (similar to a wasp sting), hyperalgesia, and allodynia that is nearly impossible to ignore.27-29 Capsaicin-sensitive primary afferent fibers play a critical role in the initiation and maintenance of pathologic pain including neuropathies.30 Capsaicin-induced pain is associated with central neurogenic mechanisms of sensitization and has been used to model chronic pain states.28,31 These recent reports increase our confidence that capsaicininduced pain models important mechanisms that are relevant to the study of clinical pain conditions, a criticism of many laboratory studies.32 The level of arousal was manipulated by using wellaccepted laboratory methodologies; a Stroop task was used to produce psychological stress/arousal, and guided imagery combined with passive progressive
161 relaxation was used to produce relaxation. Differences in cardiovascular and neuroendocrine responses may reflect potential explanatory mechanisms for the effect of pain-unrelated stress on pain. Norepinephrine and cortisol were measured because they have been associated with pain modulation.33-36 Attention was given in this prospective study to control for hormonal fluctuations on pain levels by including only women in the follicular phase of their menstrual cycle (days 3 to 10).37 Changes in both the subjective experience of relaxation and in physiologic response to the task were measured and used as potential predictors of pain perception/modulation. This design allowed us to verify whether changes in pain perception, in response to interventions not directly related to pain, can be explained by nonspecific changes in the level of physiologic activity reflecting a “single arousal dimension” and to evaluate whether these effects were associated with changes in subjective levels of arousal/relaxation. The effect of this experimental manipulation on the flare response has been the object of a separate report.38
Method Subjects Subjects were healthy men and premenopausal women between the ages of 19 and 48 years recruited from advertising in local newspapers and screened by telephone to determine eligibility as previously described.38 Screening criteria were designed to exclude subjects with inflammatory or immunomodulatory conditions and those taking medications that potentially could affect the physiologic outcome parameters. Subjects with conditions such as diabetes, multiple sclerosis, cancer, rheumatoid arthritis, autoimmune conditions, fibromyalgia, chronic fatigue syndrome, cancer, organ transplant, lupus, eczema, allergy to bee stings, pregnancy, or cardiac or respiratory conditions that would put the patient at potential risk after a capsaicin injection were excluded. Use of birth control pills or other hormonal medication, immunomodulatory medications (eg, corticosteroids), β-adrenergic receptor antagonists, psychotropic medication, chronic use of antihistamines, and cigarette consumption also were grounds for exclusion. Subjects with current or past history of major psychiatric illness or anxiety disorder, hospitalization within the past 6 months, or an acute infectious illness within the last month were excluded. All women participants were premenopausal. Sixty-six subjects met initial screening criteria to be included in the relaxation training part of the study. Ten subjects (6 men and 4 women) did not meet the relaxation criteria (see following). One subject dropped out before the start of the study, and 5 dropped out during the relaxation training for reasons of inconvenience or scheduling difficulties. Of the remaining 50 participants, 44 had complete data relevant to this report. (Three aliquots of blood were hemolyzed at baseline and thus were unavailable for
162 norepinephrine measure. One of these was from the relaxation group and 2 were from the stress group. There were 2 subjects in the stress group from whom blood could not be drawn after baseline, and blood of 1 control subject was hemolyzed and not available to be tested for norepinephrine.) Subjects were assigned randomly to condition as follows: stress (n = 16 [9 male, 7 female]), relaxation (n = 19 [10 male, 9 female]), and control (n = 9 [6 male, 3 female]).
Procedure The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the Institutional Review Board of the University of Iowa.
Training Volunteers were screened by telephone to assess eligibility. Potentially eligible subjects completed an informed consent and attended 3 separate group relaxation training sessions held during a 1-week period. Subjects were trained in relaxation to ensure that those to be included in the relaxation group could indeed achieve a predetermined level of relaxation. This procedure was applied to all subjects before randomization (1) to avoid the sampling bias that the exclusion of subjects unable to relax only from the relaxation group would have introduced and (2) to provide homogenous pre-experimental conditions across groups and thereby prevent the introduction of group differences associated with the familiarity with the experimental context and the development of specific expectancies. At each session, a finger temperature sensor was attached to the tip of the middle finger of the nondominant hand. Temperature was measured every minute while participants listened to a 20-minute relaxation tape through individual tape recorders with headphones. In addition, they were told to keep their eyes closed. The tape included previously validated instructions for imagerybased passive progressive relaxation. Participants were accepted into the study if they were able to achieve a 2º increase in finger temperature during at least 2 of the 3 relaxation training sessions.39 Potential participants who did not meet the temperature change criterion were offered 2 additional training sessions and were included in the study if they were able to achieve a 2º increase in finger temperature twice across the 5 sessions.
Experimental Session Participants were scheduled for a study appointment and asked to refrain from exercise, consumption of alcohol, and ingestion of nonprescription medication for 24 hours and caffeine for 15 hours before the study. Women were scheduled during the follicular phase of their menstrual cycle (days 3 to 10). The experimental session was held in the early evening (starting at 5:30 PM). Women were given a pregnancy test before initiation of the experiment and excluded if pregnant. Subjects were then seated in a comfortable chair,
Stress Effects on Capsaicin-Induced Pain instructions given, and a catheter was inserted. Using previously published methodology, the site for the capsaicin injection was marked on the volar forearm of the subjects.29 Subjects completed questionnaires and read magazines for a total of 22 minutes before the baseline period started. The experimental session was divided into 3 stages: baseline, experimental task, and capsaicin pain. The baseline consisted of 8 minutes during which subjects continued to read a magazine quietly before the experimental tasks. At the end of the baseline period, subjects completed a relaxation/tension inventory, blood was sampled from the catheter, and subjects began the experimental task. The stress group was administered a 20-minute computerized version of the Stroop colorword test,40 which has been shown to reliably produce sympathetic activation.41 Participants were urged to improve their performance by the examiner at 5minute intervals. Those in the relaxation group listened to the relaxation tape used previously in the relaxation training sessions, using a hand-held portable tape recorder and large headphones to eliminate any distracting noise. Subjects in this condition were instructed to keep their eyes closed. Those in the control group watched a 20-minute neutral video of a nonfiction historical account of “The Bridges and Ferries of Iowa City” narrated by a local historian. After the completion of each task, participants completed the relaxation inventory, had a second blood sample drawn, and received the capsaicin injection. Less than 30 seconds elapsed from completion of the task to the capsaicin injection. Capsaicin (100 µg) was injected into the volar surface of the nondominant forearm at the point previously marked.
Capsaicin The capsaicin preparation consisted of 100 µg of capsaicin (8-methyl N vanillyl 6-nonamide) suspended in 10 µL of a polyoxyethylene (20) sorbetan mono-oleate (Tween 80) saline vehicle. A stock solution of capsaicin was prepared containing 10% capsaicin in ethanol (wt by volume) by using previously published methodology.29 Two-milliliter aliquots were withdrawn and diluted to 1% with ethanol. The ethanol then was removed by vacuum, and the capsaicin was dissolved in 0.14 mL of 7.5% Tween 80 by weight in saline. This mixture was brought into a colloidal mixture with 1.86 mL of 0.9% saline by sonication and then injected through a Millipore filter (0.2 mm pore size) (Millipore Corporation, Bedford, MA) into a sterile injection vial. The vial was frozen at –20°C until use. Before using, the vial was warmed to room temperature and shaken by hand for 30 seconds.
Measures Cardiovascular Activity Cardiovascular measures were acquired every 2 minutes for the 8 minutes in the baseline period and every
ORIGINAL REPORT/Logan et al 4 minutes for the 20 minutes of the experimental task (5 measures starting 2 minutes after task initiation). Heart rate (HR), systolic blood pressure (SBP), and diastolic blood pressure (DBP) were obtained by using a Critikon Dinamap Plus (Johnson & Johnson, New Brunswick, NJ) vital signs monitor applied to the nondominant arm.
Cortisol Blood was sampled at the end of the baseline period and at the end of the experimental task immediately before the injection of capsaicin. Cortisol was measured as an indicator of the hypothalamic pituitary adrenocortical (HPA) response to the stressor.42 Aliquots of serum were separated from blood samples and assayed by competitive radioimmunoassay (EURO/DPC’s Double Antibody Cortisol RIA; Diagnostic Products Corporation, Los Angeles, CA). Approximate sensitivity of the assay is 0.3 µg/dL. Interassay coefficients of variability range from 5.8% to 7.8%, and intra-assay coefficients of variability range from 2.4% to 4.5%.
Norepinephrine Norepinephrine was measured from the same blood samples as cortisol as an indicator of sympathetic nervous system activity.43 Plasma norepinephrine (NE) was measured in the Cardiovascular Center Core Laboratory by using high-performance liquid chromatography with electrochemical detection (HPLC-EC). Catecholamines in the sample were adsorbed onto acid-washed alumina and eluted with 0.1 mol/L perchloric acid. After microfiltration, the eluate was buffered and underwent chromatography on a Biophase ODS (C-18) Keystone Scientific Inc, Belafonte, PA) HPLC column (5 µm, 250 mm × 4.6 mm) with a mobile phase of 0.1 mol/L KH2PO4, 0.1 mmol/L ethylenediamine tetra-acetic acid, 7.8% methanol, and 4 mmol/L heptane sulfonic acid as the ion-pairing agent. Catecholamines were detected with a BAS LC-4B (Bioanalytical Systems, Lafayette, IN) electrochemical detector at an applied potential of +650 mV by using a glassy carbon-working electrode (BAS). Peaks were quantitated on a Shimadzu CR 3-A (Shimadzu Corporation, Columbia, MD) integrator. Sensitivity of the HPLC-EC system was 50 pg for norepinephrine.
163 pain on a 100-mm visual analog scale (VAS) anchored with “no pain” to “pain as bad as you can imagine.”1
Data Reduction and Analytic Strategy Data analysis used the Statistical Package for the Social Sciences (SPSS, Chicago, IL) version 7.5 for PC and SAS version 7.0 (SAS Institute, Cary, NC) for PC. The 4 cardiovascular measures obtained in the baseline period before the experimental task were averaged. Mean levels of cardiovascular measures during the experimental task were calculated for each group by averaging all 5 measures obtained during the experimental task. Delta values were calculated as the change from baseline to task for all measures.45 Equivalence of groups at baseline was examined by using 2-way analyses of variance (ANOVA) to test the effects of task, sex, and sex by task. Next, the adequacy of the experimental manipulation was checked by comparing task-related changes in the 3 groups (ANOVA). Tukey’s post hoc tests were performed on significant ANOVAs to determine pair-wise differences between conditions. A second set of analyses examined the effects of the experimental manipulation on the maximum pain to examine group differences, sex differences, and group by sex interaction. Clinicians and patients tend to be most concerned about maximum pain in any clinical pain setting; thus, the maximum pain rating observed within the first 3 minutes after capsaicin injection was used as the dependent measure in these analyses. These analyses were repeated by using initial pain at 7 seconds after injection and the area under the curve (AUC) for the first 3 minutes of pain measurements. Pearson correlation analyses were conducted on self-report measures (RI), physiologic measures of arousal, and maximum pain. Delta mean scores of postexperimental arousal task and pretask were used in these analyses. Z tests on Fisher transformed Pearson correlation coefficients were used to test for differences between correlations.46 Finally, a series of hierarchical regression analyses were conducted to determine potential mediators responsible for any differences or interactions on maximum spontaneous pain.
Results
Relaxation
Sample Characteristics
The subjective level of relaxation was assessed by using the relaxation inventory (RI) at the end of the baseline period and immediately after the task before the capsaicin injection.44 Items were summed to obtain an overall relaxation score, with higher scores indicating more relaxation/less subjective arousal. Test-retest reliability for this measure is between 0.87 and 0.97.
Subjects had a mean age of 30.20 (SD = 9.02) years and a median income in the range of $20,000 to $30,000. The majority of subjects (93.7%) were white. A large percentage of subjects (88%) had at least some college education. There were no significant differences among the 3 groups (stress, relax, or control) on any demographic variables (eg, age, sex, income, education) (all P values > .42). There were no sex, task, or sex by task effects on baseline measures of SBP, DBP, HR, NE, cortisol, and RI with one exception. There was a trend for men (mean [SD] = 117.8 [11.4]) to have higher SBP than women (mean [SD] = 108.8 [12.2]; F [2, 38] = 3.83, P = .06).
Pain Measures of spontaneous pain were taken every minute beginning at 7 seconds after the injection of capsaicin for 20 minutes and every 5 minutes thereafter until 60 minutes. Participants were asked to mark their
164
Stress Effects on Capsaicin-Induced Pain
Table 1. Mean (SEM) Task-Induced Changes in SBP, DBP, HR, NE, Cortisol, and RI Scores in the Stress, Relaxation, and Control Groups, With the Corresponding F Value for the Main Effect of Task (ANOVA) and the Results of the Post Hoc Comparisons (Tukey)
∆SBP ∆DBP ∆HR ∆NE ∆Cortisol ∆RI
STRESS (S)
RELAX (R)
CONTROL (C)
F [2, 42]
11.8 (2.8) 19.8 (1.5) 8.3 (1.3) –2.9 (19.3) 1.6 (1.0) –27.2 (5.6)
–5.3 (1.2) 6.2 (1.0) –4.3 (1.1) –21.3 (10.1) –1.2 (0.75) 14.1 (4.8)
–1.7 (1.8 ) 10.2 (1.3) –1.7 (4.3) 19.3 (12.1) –1.4 (0.72) 1.4 (3.8)
19.58*** 32.12*** 33.36*** 2.44* 3.42** 18.80***
TUKEY S>R** S>R** S>R** All not significant S>R* S>R**
*P < .10. **P < .05. ***P < .01.
Manipulation Checks We performed separate 2-way (sex by task) analyses of variance on task-related changes for each of the arousal measures (SBP, DBP, HR, NE, cortisol, and RI). For each measure, there was an effect of task (although it was marginally significant for NE, P < .10) but no effect for sex or interaction between sex by task (Table 1). Post hoc analyses revealed that the stress condition resulted in greater arousal than either the relaxation or the control condition on SBP, DBP, HR, and RI (P < .05), but the relaxation condition did not differ from the control (all P values > .10). The same pattern was true for cortisol (P < .10), but the differences did not reach the criterion for statistical significance.
Spontaneous Pain The mean for spontaneous pain across the sample was 59.0 mm (SD = 19.6) on the 100-mm VAS, with 2 participants rating the pain as 100 (“pain as bad as you can imagine”). Examination of the profile of pain ratings over time confirmed the rapid decrease (within a few minutes) typically observed after capsaicin injections. After 3 minutes, mean and standard deviation for pain ratings were 36.6 (19.6). There was a significant interaction between sex and the experimental condition (F [2, 43] = 4.07, P = .02). (The interaction between sex and conditions was confirmed on the initial pain rating [P < .05] and was marginally significant on the AUC for the first 3 minutes of measurements [P = .06]. This effect did not reach significance when later points are included in the analysis of the AUC [P > .10]). Follow-up Tukey tests showed that stressed women showed significantly greater maximum pain than did stressed men (P < .01). Women in the relax condition showed significantly less pain than women in the stress condition (P < .05) (Fig 1). There were no other significant differences. Baseline cortisol, NE, DBP, SBP, and HR were not significantly correlated with maximum pain (all P values > .2). Delta RI was negatively correlated with maximum pain (r = –0.35), indicating that the more subjects reported being relaxed the less pain they described. In addition, ∆NE (r = –0.25) and ∆ cortisol (r = –0.28) were negatively correlated with maximum pain report. Delta
SBP, ∆DBP, and ∆HR were not significantly correlated with maximum pain report. Table 2 shows correlations between indexes of arousal (∆SBP, ∆DBP, ∆HR, ∆NE, ∆cortisol, and ∆RI) and maximum pain for the whole sample and for men and women separately. The ∆RI was inversely related to maximum pain (r = –0.63) in women but not in men (r = 0.13), a difference that reached statistical significance (Fisher transformation t = 7.9, P < .01). Although not significantly different from each other, ∆NE was inversely associated with maximum pain (r = –0.42) in men but not women (r = –0.06). Likewise, ∆DBP was inversely associated with pain in men (r = –0.46, P < .05) but not in women (r = 0.14, P > .1). These latter correlation coefficients were marginally different from each other (Fisher transformation t = 1.94, P > .1).
Regression Models To better understand the factors involved in the interaction between sex and the experimental task on maximum pain, multiple regression analysis was used. Based on prior research and our correlation analysis, variables were selected as predictors of pain
Fig 1. Mean (+SEM) maximum pain evoked by capsaicin after a cognitive stress, relaxation, or a control neutral procedure in men and women. P values are given for the significant post hoc comparisons examining the interaction between sex and tasks.
ORIGINAL REPORT/Logan et al
165
Table 2. Pearson Product Moment Correlations Between Maximum Pain and Task-Related Changes in SBP, DBP, NE, Cortisol, and RI Scores in the Whole Sample and for Men and Women, With the Corresponding Result of the Fisher Transformation Evaluating the Difference Between the Correlation Coefficients in Men and Women WHOLE SAMPLE (N = 44) ∆SBP ∆DBP ∆HR ∆NE ∆Cortisol ∆RI
–0.05 –0.15 0.11 –0.25** –0.28** –0.35**
MEN (N = ?) –0.24 –0.46** –0.10 –0.42** –0.29* 0.13
WOMEN (N = ?)
FISHER TRANSFORMATION (MEN
0.38* 0.14 0.34 –0.06 –0.31* –0.63***
V
WOMEN)
NS * NS * NS ***
Abbreviation: NS, Not significant. *P < .10. **P < .05. ***P < .01.
and separate models were built predicting maximum pain from the physiologic (NE and cortisol) and the self-report measures of arousal (RI). In each model, the variables were entered in the following manner. First, baseline measures and sex were entered as control variables. The delta scores of the arousal variables were then entered. The next step included the interaction term of sex by the delta arousal measure(s). The final step was the condition and sex by condition interaction term. This allowed us to determine whether the sex by condition interaction was explained by the arousal measures.
Model 1: Self-Report Measures The interaction term between sex and task-related ∆RI significantly predicted maximum pain (β = –1.29, P = .03). Although the main effects of baseline RI, sex, and ∆RI were entered into the model, they did not significantly contribute to predict pain. The sex by condition interaction was no longer a significant predictor of maximum pain after the interaction term sex by ∆RI was entered into the model. This model accounted for 35% of the variance in maximum pain.
Model 2: Physiologic Measures Task-induced changes in NE (β = –.44, P = .007) and cortisol (β = –.49, P = .002) were significant predictors of maximum pain, over and above the significant relationship of baseline cortisol with maximum pain (β = –.51, P = .002). However, the sex by condition interaction (change in R2 = 8%; β = –1.13, P = .03) was still a significant predictor of maximum pain after changes in NE and cortisol were entered into the model, indicating that physiologic indexes of arousal could not totally account for this interaction. This total model accounted for 43% of the variance in maximum pain. The interactions of sex by ∆cortisol and sex by ∆NE were forced into the model but did not predict maximum pain (P > .10) and thus were dropped from further analysis. We also tested cardiovascular measures (baseline and deltas) in a similar regression model; however, even when forced into the model, these values did not pre-
dict maximum pain and were also dropped from further analysis.
Model 3: Combined Model Table 3 shows that both physiologic and self-report measures predicted maximum pain. As before, baseline cortisol and changes in cortisol and NE were associated with maximum pain (model 2). Neither the change in relaxation score by sex nor the sex by condition interaction reached significance in this combined model. However, this combined model was better at predicting pain than either the physiologic model or the selfreport model alone (in both cases P < .01), accounting for 56% of the variance in maximum pain. The physiologic measures were not correlated with the self-report measures of arousal (P > .1), thereby explaining their independent effects on maximum pain.
Discussion There were 4 major findings from this study. First, the manipulation of arousal by the experimental tasks interacted with sex of subjects to predict pain. These results were consistent whether maximum or initial pain report was used, making these findings especially relevant to clinical settings in which acute painful procedures are used. Stressed women reported significantly greater pain than both men in the stress condition and women in the relaxation condition. A second major finding was that higher levels of physiologic arousal (∆DBP, ∆NE, and ∆cortisol) were related to lower maximum pain report, most consistently in men (∆DBP and ∆NE). Third, self-reported arousal (decrease in subjective relaxation; ∆RI) was associated positively with maximum pain only in women. Finally, physiologic and self-reported arousal independently predicted maximum pain, and a model including both accounted for 56% of the variance in maximum pain. Both self-report and physiologic indexes showed that the experimental manipulation was effective in so far as the stress condition produced more arousal than the control or relaxation condition. However, there were
166
Stress Effects on Capsaicin-Induced Pain
Results of the Regression Analysis Testing Model 3: Pain Regressed on Baseline Cortisol, NE, and R; ∆RI; ∆Cortisol; ∆NE; Sex; Sex by ∆RI; Condition; and Sex by Condition Table 3.
STANDARDIZED Cortisol baseline NE baseline RI baseline Sex ∆Cortisol ∆NE ∆RI Sex by ∆RI Condition Sex by condition
–0.45 –0.15 0.23 0.38 –0.42 –0.37 0.44 –0.72 0.42 –0.49
B
T
STATISTIC –3.16 –1.11 1.78 0.97 –2.79 –2.65 0.85 –1.36 1.01 –0.91
no differences in arousal indexes between the control and relaxation groups. There are 2 possible explanations for this lack of difference. First, the small number of subjects limited the statistical power to detect effects, and, second, the control group also was trained in relaxation and may have used this strategy while watching the neutral control video. The finding that stressed women reported more pain than stressed men is consistent with a large body of research showing that women experience more pain than men.47 Although a number of studies have looked at the association between stress and pain report, to our knowledge, only 2 other studies have examined sex differences in pain report after stress. Bragdon et al17 observed that women with low BP reported more thermal pain before stress than men with low BP, but there were no sex differences after the stress task. Rhudy and Meagher15 found no sex difference in heat pain threshold after a stress manipulation. It is possible that the observed differences between our findings and these previous studies are related to the use of different pain tasks and measures in each study. The pain produced by a discrete subcutaneous injection of capsaicin is inescapable—it peaks rapidly and thereafter decreases slowly. In contrast, pain produced by the sustained application of radiant or contact heat stimuli is escapable, and both threshold and tolerance tests depend on active motor/behavioral responses. The possibility or impossibility to escape pain and the differential motor/behavioral requirements of the pain measurements may lead to differences in the active/passive coping strategies adopted by men and women. The coping literature shows sex-specific differences in the use of specific cognitive strategies.48,49 For instance, women report greater use of catastrophizing than men, and this coping strategy has been associated with increased pain report.50 Both cognitive stress and inescapable pain may promote catastrophizing and may have contributed to the higher pain reported in women in the stress group. Sex differences in stressinduced pain modulation should be examined within the same study by using multiple pain tests and measurements to evaluate this possibility.
P VALUE .003 .280 .080 .340 .009 .010 .400 .180 .320 .370
R2 0.12
0.49
0.55 0.56
Another possibility to explain the difference between our findings and those of previous studies is that the relaxation training procedure used in the present study may have shifted the sex-related effects in pain. Because all subjects were familiar with the experimental setting and were trained in relaxation, subjects in all 3 groups might have been at an overall lower baseline stress level in the experimental phase than in other studies. This training may have contributed to mask the typical higher pain sensitivity of women in the relaxation and control groups. This effect would have been counteracted by the stress task lending to the expected sex difference in pain. We also found that women in the stress condition reported more pain than women in the relaxation condition. Given that the induction of stress was the most effective manipulation in this study, these results suggest that women displayed stress-induced hyperalgesia. However, the difference in pain between the stress and the control conditions did not reach significance and, therefore, our results cannot exclude the possibility that relaxation may have contributed to part of the difference in pain between women in the stress and relaxation groups. Only 1 other study has compared and found similar effects of both stress and relaxation on pain report, but it did not examine whether these effects were interactive with the sex of the subject.51 Consistent with our results, Houston and Jesurum52 examined the effects of relaxation on chest tube removal pain in a pilot study and found a trend toward less pain during relaxation in women but not in men. Our results and those of Houston and Jesurum are consistent with previous findings showing that stress or relaxation irrelevant to pain may modulate pain perception and further indicate that these effects differ in men and women. Physiologic measures of task-induced changes in arousal (∆DBP, ∆NE, and ∆cortisol) were related inversely to pain report. The relationship between pain and both DBP and NE only was seen in men, whereas the relationship with cortisol was equally strong in men and women. Previous studies have shown that NE has analgesic effects.35 For instance, NE has been implicated
ORIGINAL REPORT/Logan et al in opiate analgesic mechanisms,36 and sex differences in opiate analgesia are reported frequently.53 Thus, our data may support a sex-specific analgesic effect. Few previous studies, however, have examined this relationship, and the underlying mechanisms are not yet established.53 Other studies, including those we have conducted, found relationships between baseline blood pressure and pain.23,24 In this study, however, we do not find such a relationship in any of the subgroups by sex by condition or across the whole sample, which may be related to lack of statistical power. There is some indication that blood pressure reactivity, indicative of sympathetic arousal, is related to pain in a sex-specific fashion; however, these findings are equivocal.18,54 In our study, changes in DBP were associated significantly with pain in men but not in women, and these correlations were only marginally different from each other. In contrast to physiologic arousal, self-reported arousal (decreased RI) was associated positively with pain. This is consistent with other previously reported research.6 In our study, however, the relationship was only significant in women. Our data are consistent with Fillingim et al25 who also found significant associations between psychological measures and pain report in women but not men. Thus, physiologic and psychological measures of arousal are differentially related to pain reports in men and women. These sex-specific effects in the correlation analyses prompted further examinations of the factors that could account for the sex-by-condition interaction on pain. Regression analysis confirmed that this interaction was explained by changes in self-reported arousal (RI scores; model 1)—that is, the sex by condition interaction was no longer significant when this factor was entered into the model. In contrast, physiologic arousal did not explain the sex-by-condition interaction (model 2). Thus, the manipulation effects we observed were better explained by the self-report measure than the physiologic measures. Moreover, the physiologic and self-report measures were independently associated with pain report. The combined model that contained all variables of interest predicted 56% of the variance in pain report model 3). Because we are slightly below the convention of a 5:1 variable subject (or variable to subjects?) ratio on the combined model,55 results should be interpreted cautiously and should be replicated in future research. On the other hand, the rule of 5:1 usually is interpreted for independent variables, and several of our variables are control (baseline) variables, increasing our confidence in the validity of this model. Overall, these findings suggest that stress-induced pain modulation may not be explained fully by a unidimensional construct of physiologic arousal and that subjective measures of arousal reflect, at least in part, different processes. Studies of the effect of emotion on pain have focused on fear and anxiety and largely have ignored the possible contribution of other subjective emotional states to the experience of pain. For example, in the study by Rhudy and Meagher,15 negative
167 emotions were elicited by the administration of electric shocks or by the threat of impending shocks to evaluate the effects of fear and anxiety, respectively, on heat pain threshold. These manipulations led to decreased and increased pain thresholds, respectively, and the investigators interpreted those findings as reflecting an analgesic effect associated with states of high arousal (fear) and an hyperalgesic effect associated with states of moderate arousal (anxiety). However, this study also included subjective ratings of arousal and basic emotions, which revealed that the experimental manipulations produced distinct patterns of negative emotional feelings. The “fear condition” was reported to elicit experiences of fear, anger, surprise, and disgust, whereas the “anxiety condition” elicited fear and decreased surprise relative to a neutral condition. These reports suggest that the pain modulation observed may not reflect simply the level of arousal but may relate to the emotion experienced. Therefore, it is reasonable to suggest that the experiences of anger, surprise, and disgust, which were almost exclusively reported in the “fear” condition, may contribute to the observed emotion-induced analgesia (a possibility not assessed directly by Rhudy and Meagher). Although our experiment did not include subjective ratings of emotions, the stress task may have produced different sets of emotions in men and women. These possibilities need to be examined more closely in experimental paradigms that manipulate not only fear and anxiety but also other emotions and use subjective ratings of emotional experience as potential predictors of pain. The interaction between sex and task, the stronger association found between pain and subjective reports of relaxation in women and physiologic activity in men, and the independent portion of variance in pain that the subjective and physiologic measures accounted for in the regression models imply that different modulatory mechanisms may have been at work in men and women. Assuming that the subjective reports of relaxation reflect, at least in part, brain mechanisms involved in the “conscious perception of one’s own state,” whereas physiologic activation may occur largely subliminally, we speculate that pain modulation in women may be more proximal to central nervous system mechanisms associated with the subjective experience of being stressed or relaxed. In contrast, pain modulation in men was not related to their subjective reports of relaxation in this experiment and was more closely related to physiologic indexes of arousal. The combination of these effects suggests an association between brain mechanisms involved in the subjective perception of pain and stress (or reduced relaxation) in women and a possible dissociation of such processes in men. The investigation of central nervous system mechanisms of pain perception and modulation in humans is in its infancy, and although men and women present largely similar patterns of brain responses to experimental pain, some differences have been noted.56 Men and women may differ not only in pain-related activa-
168
Stress Effects on Capsaicin-Induced Pain
tion but also in the mobilization of central modulatory processes. Recent studies suggest important individual differences in the engagement of descending modulatory mechanisms,57 some of which have been associated with the sex of the individual.58,59 Our results may reflect the differential engagement of those mechanisms during stress in men and women. These results with maximum capsaicin-induced pain also may have clinical implications for patients, particularly women, undergoing painful procedures. Preprocedural arousal may influence the amount of pain that subjects experience. Minimization of high levels of arousal before procedures as adjuncts to conventional analgesics may aid pain management and potentially speed recovery, particularly in women.60,61 Cognitive tasks performed in the laboratory may approximate the effect of the numerous sources of stress unrelated to pain that are inherent to normal human environments including the clinic. Several important conclusions can be drawn from this study. First, the manipulation of stress/arousal levels modulated the pain experience, but the direction of the effect depended on the sex of the individual. Second, self-report and physiologic measures of arousal indepen-
dently predicted pain, and those effects were again different in men and women. Taken together, these findings suggest that a unidimensional model of arousal may be insufficient to explain individual and stress-related differences in pain. Future research on stress-induced pain modulation should consider the multidimensionality of stress (physiologic and subjective experience) and its impact on emotion and coping. These emotions and coping strategies may engage central modulatory processes differentially in men and women and may contribute to the frequently found sex differences in pain.
References
11. Olson GA, Olson RD, Kastin AJ: Endogenous opiates: 1996. Peptides 18:1651-1688, 1997
1. Price DD: Psychological Mechanisms of Pain and Analgesia, volume 15. Seattle, IASP Press, 1999, pp 6-9 2. Caceres C, Burns JW: Cardiovascular reactivity to psychological stress may enhance subsequent pain sensitivity. Pain 69:237-244, 1997 3. Logan H, Baron RS, Kohout F: Sensory focus and procedural information as therapeutic treatment for dental stress. Psychosom Med 57:475-484, 1995 4. Baron RS, Logan H, Hoppe S: Emotional and sensory focus as mediators of dental pain among patients differing in desired and felt dental control. Health Psychology 12:381-389, 1993 5. Cornwall A, Donderi DC: The effect of experimentally induced anxiety on the experience of pressure pain. Pain 35:105-113, 1988 6. Logan H, Baron R, Keeley K, Law A, Stein S: Desired and felt control as mediators of stress in a dental setting. Health Psychology 10:352-359, 1991 7. Ham LP, Packard RC: A retrospective, follow-up study of biofeedback-assisted relaxation therapy in patients with posttraumatic headache. Biofeedback & Self Regulation 21:93-104, 1996 8. Sherman RA, Gall N, Gormly J: Treatment of phantom limb pain with muscular relaxation training to disrupt the pain-anxiety-tension cycle. Pain 6:47-55, 1979 9. Dougher MJ, Goldstein D, Leight KA: Induced anxiety and pain. J Anxiety Disord 1:259-264, 1987 10. Weisenberg M, Aviram O, Wolf Y, Raphaeli N: Relevant and irrelevant anxiety in the reaction to pain. Pain 20:371383, 1984
Acknowledgments We would like to acknowledge Dr Erling Anderson for the generous use of equipment; Dr Winston Barcellos for medical advice; Mr Mark Morrison, Dr Carol Hodne, and Dr Robert Wade for assistance in data collection; Dr H. Lester Kirchner for advice on data analysis; Dr Lloyd Matheson for assistance with the capsaicin preparation; Donna Farley, Senior Research Assistant; and the director and staff of the Clinical Research Center at the University of Iowa.
12. Bandura A, Cioffi D, Taylor CB, Brouillard ME: Perceived self-efficacy in coping with cognitive stressors and opioid activation. J Pers Soc Psychol 55:479-488, 1988 13. Lichtman AH, Fanselow MS: Cats produce analgesia in rats on the tail-flick test: Naltrexone sensitivity is determined by the nociceptive test stimulus. Brain Res 12:9194, 1990 14. Fanselow MS: Shock-induced analgesia on the formalin test: Effects of shock severity, naloxone, hypophysectomy, and associative variables. Behav Neurosci 98:79-85, 1984 15. Rhudy JL, Meagher MW: Fear and anxiety: Divergent effects on human pain thresholds. Pain 84:65-75, 2000 16. Maixner W, Humphrey C: Gender differences in pain and cardiovascular responses to forearm ischemia. Clin J Pain 2:16-25, 1993 17. Bragdon EE, Light KC, Girdler SS, Maixner W: Blood pressure, gender and parental are factors in baseline and poststress pain sensitivity in normotensive adults. Int J Behav Med 4:17-38, 1997 18. Nyklicek I, Vingerhoets AJJM, Van Heck GL: Hypertension and pain sensitivity: Effects of gender and cardiovascular reactivity. Biol Psychol 50:127-142, 1999 19. Weinstock M, Razin M, Schoerer-Apelbaum D, Men D, McCarty R: Gender differences in sympathoadrenal activity in rats at rest and in response to footshock stress. Int J Dev Neurosci 16:289-295, 1998 20. Livezey GT, Miller JM, Vogel WH: Plasma norepinephrine, epinephrine and corticosterone stress responses to restraint in individual male and female rats, and their correlations. Neurosci Lett 62:51-56, 1985
ORIGINAL REPORT/Logan et al 21. Polefrone JM, Manuck SB: Gender differences in cardiovascular and neuroendocrine response to stressors, in Barnett RC, Biener L, Baruch GK (eds): Gender and Stress. New York, Macmillan, 1987, pp 13-38 22. Light KC, Turner JR, Hinderliter AL, Sherwood A: Race and gender comparisons: Hemodynamic responses to a series of stressors. Health Psychol 12:354-365, 1993 23. Fillingim RB, Maixner W: The influence of resting blood pressure and gender on pain perception. Psychosom Med 58:326-332, 1996 24. Sheffield D, Biles PL, Orom H, Maixner W, Sheps DS: Race and sex differences in cutaneous pain perception. Psychosom Med. 62:517-523, 2000 25. Fillingim RB, Keefe F, Light KC, Booker DK, Maxiner W: The influence of gender and psychological factors on pain perceptions. J Gender Culture Health 1:21-36, 1996 26. Gracely RH: Studies of pain in man, in PD Wall, Melzack R (eds): Textbook of Pain (ed 3). Edinburgh, Scotland, Churchill Livingstone, 1994, pp 303-313 27. Baumann TK, Simone DA, Shain CN, LaMotte RH: Neurogenic hyperalgesia: The search for the primary cutaneous afferent fibers that contribute to capsaicin-induced pain and hyperalgesia. J Neurophysiol 66:212-226, 1991 28. Iadarola MJ, Berman KF, Zeffiro TA, Byas-Smith MG, Gracely RH, Max MB, Bennett GJ: Neural activation during acute capsaicin-evoked pain and allodynia assessed with PET. Brain 121(pt 5):931-947, 1998 29. LaMotte RH, Shain DN, Simone DA, Tsai EF: Neurogenic hyperalgesia: Psychophysical studies of underlying mechanisms. J Neurophysiol 66:190-211, 1991 30. Jansco G, Dux M, Santha P: Role of capsaicin-sensitive afferent nerves in initiation and maintenance of pathological pain. Behav Brain Sci 20:454-455, 1997 31. Morris VH, Cruwts SC, Kidd BL: Characterization of capsaicin-induced mechanical hyperalgesia as a marker for altered nociceptive processing in patients with rheumatoid arthritis. Pain 71:179-186, 1997 32. Berkeley KJ: From psychophysics to the clinic. Pain Forum 4:225-227, 1995 33. Kuraishi Y, Harada Y, Takagi H: Noradrenaline regulation of pain-transmission in the spinal cord mediated by αadrenoceptors. Brain Res 174:333-336, 1979 34. Melzack R Katz J: Pain measurement in persons in pain, in Wall P, Melzack R (eds): Textbook of Pain (ed 3). Edinburgh, Scotland, Churchill Livingstone, 1994, pp 337-348 35. Furst S: Transmitters involved in antinociception in the spinal cord. Brain Res Bull 48:129-141, 1999 36. Yaksh TL: Pharmacology and mechanisms of opioid analgesic activity. Acta Anasthesiol Scand 41:94-111, 1997 37. Riley JL, Robinson ME, Wise EA, Price DD: A meta-analytic review of pain perception across the menstrual cycle. Pain 81:225-235, 1999 38. Lutgendorf S, Logan H, Kirchner HL, Rothrock N, Svengalis S, Iverson K, Lubaroff D: Effects of relaxation and stress on the capsaicin-induced local inflammatory response. J Psychom Med 62:524-534, 2000
169 39. Schwartz MS: Biofeedback: A Practitioner’s Guide. New York, Guilford Press, 1987 40. Klein GS: Semantic power measured through the interference of words with color naming. Am J Psychol 77:576588, 1964 41. Frankenhauser M, Mellis J, Rissler A, Bjorksell C, Patkar P: Catecholamine excretion as related to cognitive and emotional reaction patterns. Psychosom Med 30:109-120, 1968 42. Kuhn CM: Adrenocortical and gonadal steroids in behavioral cardiovascular medicine, in Scheneiderman N, Weiss S, Kaufmann P (eds): Handbook of Research Methods in Cardiovascular Behavioral Medicine. New York, Plenum, 1989, pp 185-204 43. Ziegler MC: Catecholamine measurement in behavioral research, in Scheneiderman N, Weiss S, Kaufmann P (eds): Handbook of Research Methods in Cardiovascular Behavioral Medicine. New York, Plenum, 1989, pp 167-183 44. Crist DA, Rickard HC, Prentice-Dunn S, Barker HR: The relaxation inventory: Self-report scales of relaxation training effects. J Pers Assess 53:716-726, 1989 45. Llabre NM, Spitzer SB, Saab PG, Ironson GH, Schneiderman N: The reliability and specificity of delta versus residualized change as measures of cardiovascular reactivity to behavioral challenges. Psychophysiology 28:701711, 1991 46. Hays WL: Statistics for the Social Scientist. New York, Holt Rinehart and Winston, 1973 47. Unruh AM, Ritchie J, Merskey H: Does gender affect appraisal of pain and pain coping strategies? Clin J Pain 15:31-40, 1999 48. Otto MW, Dougher MJ: Sex differences and personality factors in responsivity to pain. Percept Mot Skills 61:383390, 1985 49. Lang PJ, Greenwald MK, Bradley MM, Hamm AO: Looking at pictures: Affective, facial, visceral, and behavioral reactions. Psychophysiology 30:261-273, 1993 50. Robinson ME, Riley JL 3rd, Myers CD: Psychosocial contributions to sex-related differences in pain responses, in Fillingim RB (ed): Sex Gender and Pain: From the Benchtop to the Clinic. Seattle, IASP Press, 2000 51. Ford MJ, Camilleri M, Zinsmeister AR, Hanson RB: Psychosensory modulation of colonic sensation in the human transverse and sigmoid colon. Gastroenterology 109:1772-1780, 1995 52. Houston S, Jesurum J: The quick relaxation technique: Effect on pain associated with chest tube removal. Appl Nurs Res 12:196-205, 1999 53. Berkley KJ: Sex differences in pain. Behav Brain Sci 20:371-380, 1997 54. Maixner W, Humphrey C: Gender differences in pain and cardiovascular responses to forearm ischemia. Clin J Pain 2:16-25, 1993 55. Hair JF Jr, Anderson RE, Tatham RL, Black RC: Multivariate Data Analysis (ed 5). Upper Saddle River, NJ, Prentice Hall, 1998, p 166 56. Paulson PE, Minoshima S, Morrow TJ, Casey KL: Gender differences in pain perception and patterns of cerebral
170 activation during noxious heat stimulation in humans. Pain 76:223-229, 1998 57. Danziger N, Fournier E, Bouhassira D, Michaud D, De Broucker T, Santarcangelo E, Carli G, Chertock L, Willer JC: Different strategies of modulation can be operative during hypnotic analgesia: A neurophysiological study. Pain 75:8592, 1998 58. France CR, Suchowiecki S: A comparison of diffuse noxious inhibitory controls in men and women. Pain 81:77-84, 1999 59. Kiernan BD, Dane JR, Philips LH, Price DD: Hypnotic
Stress Effects on Capsaicin-Induced Pain analgesia reduces R-III nociceptive reflex: Further evidence concerning the multifactorial nature of hypnotic analgesia. Pain 60:39-47, 1995 60. NIH Technology Assessment Panel: Integration of behavioral and relaxation approaches into the treatment of chronic pain and insomnia. JAMA 276:313318, 1996 61. Lang EV, Benotsch EG, Fick LJ, Lutgendorf S, Berbaum ML, Berbaum KS, Logan HL, Spiegel D: Adjunct nonpharmacologic analgesia for invasive medical procedures: A randomized trial. Lancet 355:14861490, 2000