ann. behav. med. DOI 10.1007/s12160-015-9686-3
ORIGINAL ARTICLE
Social Support Attenuates Physiological Stress Responses and Experimental Pain Sensitivity to Cold Pressor Pain Matthew H. Roberts, B.S. & Rebecca R. Klatzkin, Ph.D. & Beth Mechlin, Ph.D.
# The Society of Behavioral Medicine 2015
Abstract Background Social support improves health and has been shown to attenuate stress and pain. The precise characteristics of social support responsible for these effects, however, remain elusive. Purpose The purpose of this study is to examine the relative efficacy of social support versus a neutral non-verbal social presence to attenuate stress and pain. Methods Seventy-six participants provided pain ratings and task assessments during a cold pressor task (CPT) in one of three conditions: verbal social support, neutral non-support, or alone. Reactivity to the CPT was assessed via cardiovascular measures, cortisol, and subjective ratings. Results Participants receiving social support showed attenuated blood pressure, heart rate, and cortisol reactivity, as well as reduced pain ratings, task difficulty, tension, and effort compared to neutral non-support and alone conditions. Conclusions Social support, not the mere presence of another individual, attenuated stress and pain during a CPT. Given the negative health consequences of stress and pain, clinical studies incorporating social support into medical procedures and treatments are warranted.
Keywords Social support . Pain . Stress . Cardiovascular . Cortisol Electronic supplementary material The online version of this article (doi:10.1007/s12160-015-9686-3) contains supplementary material, which is available to authorized users. M. H. Roberts : R. R. Klatzkin (*) Department of Psychology, Rhodes College, 2000 North Parkway, Memphis, TN 38112, USA e-mail: [email protected] B. Mechlin Department of Psychology, Earlham College, Richmond, IN, USA
Introduction Heightened levels of psychological and physiological stress have widespread, deleterious effects on patient health and recovery [1–4]. The negative health consequences of stress are mitigated by social support, defined by Cobb (1976) as “information leading the subject to believe that he is cared for and loved, esteemed, and a member of a network of mutual obligations” [5, p. 300]. High levels of social support have been linked to decreased morbidity and mortality [6, 7], facilitated wound healing [1], decreased inflammation [8], greater health-related quality of life for cancer patients [9], and numerous positive health outcomes associated with aspects of the cardiovascular, endocrine, and immune systems [10]. The ability of social support to augment various health indices may be mediated by the buffering effect of social support on the physiological stress response [11, 12]. The stressreducing effects of social support have been demonstrated in laboratory settings, with studies reporting attenuated cardiovascular and cortisol reactivity to mental stress for participants receiving social support compared to participants in alone conditions [13–15], particularly under conditions of high social threat [16]. These findings have been further substantiated by a meta-analytical report concluding that, across 22 independent studies, social support had moderately strong to strong effect sizes in its ability to diminish both cardiovascular and cortisol stress responses [17]. Reductions in pain perception may also be a means by which social support improves health and well-being [18]. Social support administered in a laboratory setting has been shown to reduce pain intensity and unpleasantness ratings, as well as increase pain thresholds during cold pressor and thermal pain tasks [19–21]. Furthermore, studies have demonstrated that simply viewing a photograph of a significant other serves to attenuate thermal pain ratings compared to a photograph of a stranger, acquaintance, or object [19, 21, 22]. Given
ann. behav. med.
that pain sensations activate autonomic and neuroendocrine stress systems [23], pain reduction via social support may blunt activation in these stress systems, potentially enhancing health outcomes. The ability of social support to remain efficacious across a variety of experimental manipulations demonstrates the pervasiveness of its stress- and pain-reducing effects [17, 18, 24, 25]. Support from both romantic partners as well as close friends have been reported to attenuate physiological stress [17, 26], while social support from friends, strangers, and significant others have each been shown to reduce pain perception in the laboratory [20, 27], regardless of whether the support is active (i.e., verbal) or passive (i.e., non-verbal). There are, however, confounds inherent to friends or partners as supporters (e.g., inconsistency in the effectiveness and nature of the support as well as length and strength of the relationship to the supporter) that threaten the internal validity of studies utilizing this particular form of social support [17]. Utilizing trained confederates as supporters eliminates these particular confounds and is as efficacious as support from a close friend in reducing cardiovascular stress responses [28] and pain ratings [27]. Moreover, studies have demonstrated that female confederates providing social support attenuated physiological stress reactivity compared to non-support [29, 30]. Consequently, the present study utilized trained female confederates in an attempt to reduce potential confounding variables and isolate the effects of social support on stress and pain. Another methodological issue in social support research is the use of an appropriate control group(s). Typically, an active social support group is compared to a single alone condition, but additional control groups enable studies to tease apart the precise mechanisms of the stress- and pain-reducing effects of social support. For instance, both active and passive social support have been shown to attenuate pain perception compared to alone conditions [15]. These results indicate that the verbal nature of the support may not be critical to the painreducing effects and that either the perception of being supported or the mere social presence of an individual may be critical to pain attenuation. In an attempt to further understand the nature of social support’s effects, Christenfeld and colleagues compared social support from a friend or confederate to a non-support group in which participants were exposed to a neutral, non-verbal confederate from whom they did not expect to receive support [28]. Participants in the latter group exhibited heightened cardiovascular reactivity to mental stress compared to the social support conditions [28], indicating that the simple presence of an individual may not be sufficient to attenuate stress reactivity. The lack of an alone control group in this study, however, limits the strength of the argument. Furthermore, a systematic review of the literature on social support and pain determined that a non-verbal social presence inconsistently affected pain according to individual
differences and yielded mixed results [18]. To our knowledge, no study to date has compared a non-verbal, neutral social presence condition to both verbal social support and alone conditions in order to determine whether it is the social support itself or the mere social presence of another individual responsible for attenuating stress and pain. Consequently, the primary aim of the current study was to compare psychophysiological stress reactivity and pain perception during cold pressor pain task across three groups: social support, neutral non-support, and alone. Perceived access to social support (i.e., the support a person feels he or she receives throughout day-to-day life [31]) has also been associated with attenuated stress reactivity and pain perception [25, 26, 32–35]. Because perceived access to social support can positively affect psychological health, anxiety, hypertension, and immune function [36, 37], it may have the potential to interact with the stress- and pain-reducing effects of social support in the laboratory. Due to the scarcity of research as well as inconsistent evidence for the relationship between these factors [38–40], the current study investigated whether the influence of perceived access to social support on stress and pain perception differs by support condition. Lastly, although the stress- and pain-reducing effects of social support have been well documented independently, no study to date has demonstrated the ability of social support to concurrently attenuate stress and pain. Due to the methodological issues described above as well as the complex interplay between stress and nociception, the reduction of both stress and pain in response to social support is simply an assumption without adequate empirical support. The concept of stress-induced analgesia (SIA), in which heightened stress levels are associated with decreased pain perception [41, 42], argues against this prediction, although great individual variability in SIA exists [41] and recent evidence suggests that SIA is inconsistently induced in laboratory settings [43]. Thus, a further purpose of the present study was to bridge pain and stress research in order to determine whether experimental manipulation of social support could reduce both stress reactivity and pain perception. We hypothesized that participants in a social support condition would show attenuated cardiovascular and cortisol reactivity as well as decreased pain perception during a cold pressor pain task compared to both neutral non-supportive and alone conditions.
Methods Participants Female undergraduates (n=76) between the ages of 18 and 21 responded to an advertisement for research investigating the influence of the menstrual cycle on stress and pain in order to disguise the true purpose of the experiment. Participants were
ann. behav. med.
excluded from study participation if they reported a history of any of the following in the past year: high blood pressure, a chronic pain condition, any cardiovascular, seizure, neuroendocrine, respiratory, or gastrointestinal disorder, hepatic or renal impairment, or Raynaud’s disease. Any participants who reported suicidality, severe depression or anxiety, regular use of tobacco, neural stimulants (e.g., for ADHD), blood pressure medication, or psychotropic medication for anxiety or depression were also excluded. Although a policy was in place to refer prospective participants to the on-campus counseling center if psychological screening questionnaires deemed them at risk, no prospective participant reported suicidality or scored in the severe range for depression or anxiety (see below). The research was approved by the college’s Institutional Review Board, and all participants received partial course credit for their time. Experimental Protocol All prospective participants completed a preliminary screening process prior to the laboratory testing session. This consisted of preliminary online questionnaires (see below) containing basic health and demographic information relevant to exclusionary criteria as well as questionnaires related to subjective psychological states and perceived access to social support. Eligible participants were scheduled for a 90-min laboratory testing session starting between 1:30 and 4:30 pm and randomly assigned to one of three support conditions: alone (n=25), neutral non-support (n=25), or support (n= 26). In the alone condition, the participant completed the testing session with only the experimenter present. Participants in the neutral non-support and support conditions completed the testing session alongside a female confederate whom participants believed was also taking part in the study. On the day of testing, participants refrained from exercising strenuously; drinking more than a single caffeinated beverage in the morning; eating or drinking (except water) 1 h prior to the study; consuming alcohol 12 h prior to the study; or taking any antihistamines, pain medication, and neural stimulants. On the day of their laboratory visit, the experimenter obtained verbal confirmation that participants had followed each of these requirements. We tested participants only during the first 14 days (non-luteal phase) of their menstrual cycle due to previous findings showing menstrual cycle phase differences in stress-responsive salivary cortisol levels [44]. Upon arrival to the laboratory, the experimenter guided those participants in the support and neutral non-support conditions to a waiting room and informed them that a second participant (i.e., confederate) had not yet arrived. No more than 2 min later, the confederate entered the waiting room and both individuals completed informed consent forms. The experimenter then asked the participant and confederate to draw cards to determine who would complete the pain task
first. The card draw was rigged to ensure that the participant always went first. The experimenter then escorted the participant to the testing room and the confederate to a separate, outer room, adjacent to the testing room. In the alone condition, participants provided consent in the waiting area alone before being escorted to the testing room. Once in the testing room, participants then completed baseline questionnaires (see below) before a 10-min baseline rest period during which an automatic blood pressure cuff collected cardiovascular measures (see below). At the end of the rest period, participants provided a baseline salivary sample. Wait Period Following an explanation of the cold pressor task (CPT), participants underwent a 5-min wait period either alone (alone condition) or with the confederate present (support and neutral non-support conditions) while the experimenter prepared the cold water bath (for further details, see Electronic Supplementary Material (ESM)). The experimenter asked the participant to familiarize themselves with the numerical pain rating scales during this time. In the support and neutral non-support conditions, the confederate read her magazine silently during the wait period. In the support condition, however, she also briefly engaged the participant to tell her that she is there to support her and to wish her good luck. Cold Pressor Pain Task Participants underwent a CPT by submerging their dominant hand up to the wrist in a circulating tank of water at 4.0± 0.1 °C until they were no longer willing or able to tolerate the pain (i.e., pain tolerance) or until they reached the 3-min time limit. During the CPT, participants non-verbally rated their pain every 20 s by pointing to a numerical rating scale from 0 (no pain) to 100 (strongest pain imaginable). Ratings were completed out of view of the confederate to eliminate evaluation effects, given that the negative feeling of being evaluated by the supporter can confound the beneficial buffering effects of social support [45]. Following each pain rating, the experimenter immediately recorded the value. At the end of the 3 min or at pain tolerance, whichever came first, participants rated the intensity and unpleasantness of their pain on a numerical rating scale from 0 (no pain intensity or unpleasantness) to 100 (most intense or unpleasant pain imaginable) immediately before removing their hand from the water. These final ratings were obtained in order to distinguish between the overall pain experienced over the course of the task (i.e., mean CPT pain) and the intensity or unpleasantness of the pain experienced immediately following the task. The pain task instructions prior to testing included a thorough description of the difference between pain intensity (i.e., physical aspect of pain) and unpleasantness (i.e., emotional aspect
ann. behav. med.
of pain). For those participants who kept their hand in the water bath for the maximum time, the time to tolerance was recorded as 3 min. After drying their hand, participants began their post-task questionnaires (see below). Support Manipulation During the pre-task wait period and the CPT, confederates in the support condition periodically provided scripted statements of verbal emotional support (e.g., “You’re doing great! Remember you’re not alone”) and congratulated participants upon completion of the task (see ESM). In the neutral nonsupport condition, the confederate was also in the same room as the participant during both the wait period and the CPT but were verbally instructed (in front of the participant) to read a magazine and refrain from interacting with the participant while they were waiting for their turn to undergo the pain task. All research assistants were thoroughly trained for their roles as confederates in order to maintain consistency. Recovery Period In the support and neutral non-support conditions, the experimenter escorted the confederate out of the testing area when the participant had completed the post-task questionnaires (15 min after the start of the CPT) under the pretense that it was her turn to begin her baseline rest period. During the subsequent recovery period, participants sat quietly in the testing area until all salivary cortisol samples had been obtained (see below), signaling the completion of the study. Before debriefing the participants, the researcher asked follow-up questions to determine if the participant realized the true purpose of the study. In the support and neutral non-support conditions, the experimenter also asked participants about the second participant (i.e., confederate) and what role that they believed that she played in the study. No participant was aware of the deception or reported feeling evaluated in any way by the confederate. Questionnaires Preliminary Screening Questionnaire Due to the effects of depression [46, 47], trait anxiety [48, 49], and perceived stress [49] on laboratory pain and stress, each of these psychological variables was assessed via an online questionnaire prior to the laboratory visit. Severity of self-reported depressive symptoms was quantified using the Beck Depression Inventory (BDI), a 21-question multiple choice survey [50]. This scale comprehensively assesses dysphoric symptoms, including affective, cognitive, somatic, overt behavior, and interpersonal symptoms of depression. Each forcedchoice question has a set of at least four possible answer
choices, with increasing severity of depressive symptoms from 0 to 3. Scores may range from 0 to 53 with higher scores indicating increased depressive symptoms (>29 indicates severe depressive symptoms). Trait anxiety levels were measured using the Spielberger State-Trait Anxiety Inventory (STAI trait), a 20-question Likert-scale questionnaire assessing chronic feelings of anxiety [51]. Participants rate how they generally feel on a scale from 1 (not at all) to 4 (very much so). Scores may range from 20 to 80 with higher scores indicating greater levels of trait anxiety (>60 indicates severe trait anxiety). Self-perceptions of life stress during the last month were assessed using the Perceived Stress Scale (PSS), a 10-question multiple choice inventory [52]. Items assess how unpredictable, uncontrollable, and overloaded respondents view their lives and directly inquire about levels of experienced stress in the past month with answer choices ranging from 0 (never) to 4 (very often). Scores range from 0 to 40 with higher scores indicating greater overall perceived life stress. Finally, participants completed the Interpersonal Support Evaluation List (ISEL), a 40question inventory that measures perceived access to social support across four subscales: tangible, belonging, self-esteem, and appraisal support [53]. Scores on each subscale may range from 0 to 30, and each subscale was averaged to yield the reported measure of perceived access to social support. To control for the effects of oral contraception use on salivary cortisol levels, participants identified in the prescreening questionnaire whether or not they used birth control methods, and if so, what type. Baseline Questionnaires Given that previous studies reported affect [54], state anxiety [54, 55], and sleep quality [56] to affect pain and stress, participants completed questionnaires assessing each of these measures prior to the baseline rest period. Affect was quantified with the Positive and Negative Affect Schedule (PANAS), a 20-item multiple choice survey [57]. For each word describing a different feeling or emotion felt at the present moment, participants had a choice of ratings from 1 (very slightly or not at all) to 5 (extremely). The positive subscale consisted of ten words and a possible range from 10 to 50, with higher scores indicating more positive affect. The negative subscale consisted of ten words and a possible range from 10 to 50, with higher scores indicating more negative affect. State, or transitory, anxiety levels were assessed with the state portion of the Spielberger State-Trait Anxiety Inventory (STAI state), a 20-question inventory utilizing Likert scales [51]. The questionnaire contains 20 statements describing how an individual may feel at the present moment and participants rate how they currently feel on a scale from 1 (not at all) to 4 (very much so). Possible scores range from 20 to 80, with
ann. behav. med.
higher scores indicating greater levels of state anxiety. The Pittsburg Sleep Quality Index (PSQI), a 19-question variedresponse survey, measured recent self-rated sleeping habits and sleep quality [58]. The 19 items are combined to create seven component scores, each with a range of 0 (no difficulty) to 3 (severe difficulty). The seven component scores are then added to yield one global score, with a range of 0 to 21, and higher scores indicating more severe sleep difficulty. Post-Task Questionnaires Participants completed a CPT assessment in which they had to rate the difficulty, tension, effort, and concentration levels perceived during the CPT on a numerical rating scale from 0 to 10. Participants also completed a second iteration of the PANAS and STAI state. Additionally, participants completed a brief manipulation check of how supported they felt during the CPT on a series of seven-point Likert scales across six different axes: supportive-unsupportive, friendly-unfriendly, accepting-rejecting, close-distant, warm-cold, and helpfulunhelpful. Physiological Measurements Cardiovascular measurements were taken at 0, 5, and 10 min during the baseline rest period; 0 and 5 min of the wait period; and subsequently every 5 min starting immediately after the participant removed her hand from the water until the end of the experiment (30 min following the start of the CPT). All cardiovascular measurements were obtained using a programmable automatic blood pressure cuff (OSCAR 2, SunTech Medical). Saliva was collected in 1.5-mL Eppendorf tubes at the end of the baseline rest period and 20-, 25-, and 30-min post-CPT. Participants passively drooled into the tube for a maximum of 2 min per sample. Saliva samples were frozen within 30 min of collection at −20 °C until assayed. On the day of testing, all samples were centrifuged at 3000 rpm for 15 min to remove mucins and were analyzed in duplicate by enzyme immunoassay (Salimetrics, State College, PA). The mean intra-assay coefficient of variation was 8.7 %, and the inter-assay coefficient was 12.1 %.
Data Analysis Group differences in demographic factors and subjective measures (perceived stress, depression, sleep quality, perceived access to social support, baseline state anxiety, trait anxiety, and affect) were examined using a multivariate ANOVA for continuous variables and chi-square analyses for dichotomous variables as appropriate. Group differences in baseline
cardiovascular measures and cortisol were also investigated using a multivariate ANOVA. Where significant results emerged, post hoc analyses with Bonferroni corrections were conducted. In order to determine the subjective measures that should serve as covariates for all analyses assessing group differences in pain and stress measures, Pearson productmoment (Pearson’s r) correlations were performed. Due to group differences in baseline perceived stress levels and results indicating that trait anxiety, baseline positive affect, and perceived access to social support were correlated with various measures of pain and stress indices, these four subjective measures were included as covariates in all analyses of group differences in pain and stress. Multivariate analysis of covariances (ANCOVAs) were used to analyze group differences in pain intensity, pain unpleasantness, mean CPT pain (average pain rating during the CPT), and time to tolerance, as well as perceived social support during testing, difficulty, tension, concentration, and effort during the CPT. Separate ANCOVAs were also used to analyze delta scores for systolic blood pressure (SBP), diastolic blood pressure (DBP), heart rate (HR), negative affect, positive affect, and state anxiety. For cardiovascular measurements, delta scores were calculated by subtracting the average baseline measurement from the measurement taken immediately following the pain task. Delta scores for the subjective measurements were calculated by subtracting the baseline scores from the post-task scores. Where group differences emerged, Sidak-corrected post hoc comparisons were performed. Analysis of salivary cortisol levels over time was accomplished by calculating the formula for area under the curve with respect to ground (AUCg), an analysis that takes into account the change over time (20-, 25-, and 30-min postCPT) and the distance of these measures from ground, or baseline [59], and subsequently performing an ANCOVA. The AUCg was employed due to the lack of group differences in salivary cortisol concentrations at baseline. In order to determine if perceived access to social support impacted the stress- and pain-attenuating effects of social support in the laboratory, Pearson product-moment (Pearson’s r) correlations were performed in the entire sample and in each of the three groups separately. In order to reduce type I error, this analysis was limited to dependent variables which significantly differed as a function of support condition: pain intensity, pain unpleasantness, mean pain ratings, delta SBP, delta HR, and cortisol AUCg. Wherever significant main effects were found, omegasquared (ω2) calculations were performed to determine the strength of the effect sizes. Omega squared is less biased and more conservative than eta-squared and takes into account overall population variance [60]. Values of 0.01, 0.06, and 0.14 represent small, medium, and large effects, respectively [61].
ann. behav. med. Table 2 Intercorrelations among baseline subjective variables and pain and stress measures
Results Demographics and Subjective Measures No significant group differences emerged for oral contraceptive use or race (p>0.05 for all comparisons, Table 1). Furthermore, body mass index, age, trait anxiety, depression, sleep quality, perceived access to social support, and baseline state anxiety; affect; SBP; DBP; and HR did not differ between groups (p>0.05 for all comparisons, Table 1). There were, however, significant group differences in baseline perceived stress (F(2,72)=3.91, p=0.024), as the alone group rated higher than the neutral non-support group (p=0.021, Table 1). Correlational analyses among subjective measures and pain and stress variables revealed that trait anxiety was correlated with cortisol AUCg, r=0.23, p=0.05, while baseline positive affect was correlated with pain intensity, r= −0.28, p
ORIGINAL ARTICLE
Social Support Attenuates Physiological Stress Responses and Experimental Pain Sensitivity to Cold Pressor Pain Matthew H. Roberts, B.S. & Rebecca R. Klatzkin, Ph.D. & Beth Mechlin, Ph.D.
# The Society of Behavioral Medicine 2015
Abstract Background Social support improves health and has been shown to attenuate stress and pain. The precise characteristics of social support responsible for these effects, however, remain elusive. Purpose The purpose of this study is to examine the relative efficacy of social support versus a neutral non-verbal social presence to attenuate stress and pain. Methods Seventy-six participants provided pain ratings and task assessments during a cold pressor task (CPT) in one of three conditions: verbal social support, neutral non-support, or alone. Reactivity to the CPT was assessed via cardiovascular measures, cortisol, and subjective ratings. Results Participants receiving social support showed attenuated blood pressure, heart rate, and cortisol reactivity, as well as reduced pain ratings, task difficulty, tension, and effort compared to neutral non-support and alone conditions. Conclusions Social support, not the mere presence of another individual, attenuated stress and pain during a CPT. Given the negative health consequences of stress and pain, clinical studies incorporating social support into medical procedures and treatments are warranted.
Keywords Social support . Pain . Stress . Cardiovascular . Cortisol Electronic supplementary material The online version of this article (doi:10.1007/s12160-015-9686-3) contains supplementary material, which is available to authorized users. M. H. Roberts : R. R. Klatzkin (*) Department of Psychology, Rhodes College, 2000 North Parkway, Memphis, TN 38112, USA e-mail: [email protected] B. Mechlin Department of Psychology, Earlham College, Richmond, IN, USA
Introduction Heightened levels of psychological and physiological stress have widespread, deleterious effects on patient health and recovery [1–4]. The negative health consequences of stress are mitigated by social support, defined by Cobb (1976) as “information leading the subject to believe that he is cared for and loved, esteemed, and a member of a network of mutual obligations” [5, p. 300]. High levels of social support have been linked to decreased morbidity and mortality [6, 7], facilitated wound healing [1], decreased inflammation [8], greater health-related quality of life for cancer patients [9], and numerous positive health outcomes associated with aspects of the cardiovascular, endocrine, and immune systems [10]. The ability of social support to augment various health indices may be mediated by the buffering effect of social support on the physiological stress response [11, 12]. The stressreducing effects of social support have been demonstrated in laboratory settings, with studies reporting attenuated cardiovascular and cortisol reactivity to mental stress for participants receiving social support compared to participants in alone conditions [13–15], particularly under conditions of high social threat [16]. These findings have been further substantiated by a meta-analytical report concluding that, across 22 independent studies, social support had moderately strong to strong effect sizes in its ability to diminish both cardiovascular and cortisol stress responses [17]. Reductions in pain perception may also be a means by which social support improves health and well-being [18]. Social support administered in a laboratory setting has been shown to reduce pain intensity and unpleasantness ratings, as well as increase pain thresholds during cold pressor and thermal pain tasks [19–21]. Furthermore, studies have demonstrated that simply viewing a photograph of a significant other serves to attenuate thermal pain ratings compared to a photograph of a stranger, acquaintance, or object [19, 21, 22]. Given
ann. behav. med.
that pain sensations activate autonomic and neuroendocrine stress systems [23], pain reduction via social support may blunt activation in these stress systems, potentially enhancing health outcomes. The ability of social support to remain efficacious across a variety of experimental manipulations demonstrates the pervasiveness of its stress- and pain-reducing effects [17, 18, 24, 25]. Support from both romantic partners as well as close friends have been reported to attenuate physiological stress [17, 26], while social support from friends, strangers, and significant others have each been shown to reduce pain perception in the laboratory [20, 27], regardless of whether the support is active (i.e., verbal) or passive (i.e., non-verbal). There are, however, confounds inherent to friends or partners as supporters (e.g., inconsistency in the effectiveness and nature of the support as well as length and strength of the relationship to the supporter) that threaten the internal validity of studies utilizing this particular form of social support [17]. Utilizing trained confederates as supporters eliminates these particular confounds and is as efficacious as support from a close friend in reducing cardiovascular stress responses [28] and pain ratings [27]. Moreover, studies have demonstrated that female confederates providing social support attenuated physiological stress reactivity compared to non-support [29, 30]. Consequently, the present study utilized trained female confederates in an attempt to reduce potential confounding variables and isolate the effects of social support on stress and pain. Another methodological issue in social support research is the use of an appropriate control group(s). Typically, an active social support group is compared to a single alone condition, but additional control groups enable studies to tease apart the precise mechanisms of the stress- and pain-reducing effects of social support. For instance, both active and passive social support have been shown to attenuate pain perception compared to alone conditions [15]. These results indicate that the verbal nature of the support may not be critical to the painreducing effects and that either the perception of being supported or the mere social presence of an individual may be critical to pain attenuation. In an attempt to further understand the nature of social support’s effects, Christenfeld and colleagues compared social support from a friend or confederate to a non-support group in which participants were exposed to a neutral, non-verbal confederate from whom they did not expect to receive support [28]. Participants in the latter group exhibited heightened cardiovascular reactivity to mental stress compared to the social support conditions [28], indicating that the simple presence of an individual may not be sufficient to attenuate stress reactivity. The lack of an alone control group in this study, however, limits the strength of the argument. Furthermore, a systematic review of the literature on social support and pain determined that a non-verbal social presence inconsistently affected pain according to individual
differences and yielded mixed results [18]. To our knowledge, no study to date has compared a non-verbal, neutral social presence condition to both verbal social support and alone conditions in order to determine whether it is the social support itself or the mere social presence of another individual responsible for attenuating stress and pain. Consequently, the primary aim of the current study was to compare psychophysiological stress reactivity and pain perception during cold pressor pain task across three groups: social support, neutral non-support, and alone. Perceived access to social support (i.e., the support a person feels he or she receives throughout day-to-day life [31]) has also been associated with attenuated stress reactivity and pain perception [25, 26, 32–35]. Because perceived access to social support can positively affect psychological health, anxiety, hypertension, and immune function [36, 37], it may have the potential to interact with the stress- and pain-reducing effects of social support in the laboratory. Due to the scarcity of research as well as inconsistent evidence for the relationship between these factors [38–40], the current study investigated whether the influence of perceived access to social support on stress and pain perception differs by support condition. Lastly, although the stress- and pain-reducing effects of social support have been well documented independently, no study to date has demonstrated the ability of social support to concurrently attenuate stress and pain. Due to the methodological issues described above as well as the complex interplay between stress and nociception, the reduction of both stress and pain in response to social support is simply an assumption without adequate empirical support. The concept of stress-induced analgesia (SIA), in which heightened stress levels are associated with decreased pain perception [41, 42], argues against this prediction, although great individual variability in SIA exists [41] and recent evidence suggests that SIA is inconsistently induced in laboratory settings [43]. Thus, a further purpose of the present study was to bridge pain and stress research in order to determine whether experimental manipulation of social support could reduce both stress reactivity and pain perception. We hypothesized that participants in a social support condition would show attenuated cardiovascular and cortisol reactivity as well as decreased pain perception during a cold pressor pain task compared to both neutral non-supportive and alone conditions.
Methods Participants Female undergraduates (n=76) between the ages of 18 and 21 responded to an advertisement for research investigating the influence of the menstrual cycle on stress and pain in order to disguise the true purpose of the experiment. Participants were
ann. behav. med.
excluded from study participation if they reported a history of any of the following in the past year: high blood pressure, a chronic pain condition, any cardiovascular, seizure, neuroendocrine, respiratory, or gastrointestinal disorder, hepatic or renal impairment, or Raynaud’s disease. Any participants who reported suicidality, severe depression or anxiety, regular use of tobacco, neural stimulants (e.g., for ADHD), blood pressure medication, or psychotropic medication for anxiety or depression were also excluded. Although a policy was in place to refer prospective participants to the on-campus counseling center if psychological screening questionnaires deemed them at risk, no prospective participant reported suicidality or scored in the severe range for depression or anxiety (see below). The research was approved by the college’s Institutional Review Board, and all participants received partial course credit for their time. Experimental Protocol All prospective participants completed a preliminary screening process prior to the laboratory testing session. This consisted of preliminary online questionnaires (see below) containing basic health and demographic information relevant to exclusionary criteria as well as questionnaires related to subjective psychological states and perceived access to social support. Eligible participants were scheduled for a 90-min laboratory testing session starting between 1:30 and 4:30 pm and randomly assigned to one of three support conditions: alone (n=25), neutral non-support (n=25), or support (n= 26). In the alone condition, the participant completed the testing session with only the experimenter present. Participants in the neutral non-support and support conditions completed the testing session alongside a female confederate whom participants believed was also taking part in the study. On the day of testing, participants refrained from exercising strenuously; drinking more than a single caffeinated beverage in the morning; eating or drinking (except water) 1 h prior to the study; consuming alcohol 12 h prior to the study; or taking any antihistamines, pain medication, and neural stimulants. On the day of their laboratory visit, the experimenter obtained verbal confirmation that participants had followed each of these requirements. We tested participants only during the first 14 days (non-luteal phase) of their menstrual cycle due to previous findings showing menstrual cycle phase differences in stress-responsive salivary cortisol levels [44]. Upon arrival to the laboratory, the experimenter guided those participants in the support and neutral non-support conditions to a waiting room and informed them that a second participant (i.e., confederate) had not yet arrived. No more than 2 min later, the confederate entered the waiting room and both individuals completed informed consent forms. The experimenter then asked the participant and confederate to draw cards to determine who would complete the pain task
first. The card draw was rigged to ensure that the participant always went first. The experimenter then escorted the participant to the testing room and the confederate to a separate, outer room, adjacent to the testing room. In the alone condition, participants provided consent in the waiting area alone before being escorted to the testing room. Once in the testing room, participants then completed baseline questionnaires (see below) before a 10-min baseline rest period during which an automatic blood pressure cuff collected cardiovascular measures (see below). At the end of the rest period, participants provided a baseline salivary sample. Wait Period Following an explanation of the cold pressor task (CPT), participants underwent a 5-min wait period either alone (alone condition) or with the confederate present (support and neutral non-support conditions) while the experimenter prepared the cold water bath (for further details, see Electronic Supplementary Material (ESM)). The experimenter asked the participant to familiarize themselves with the numerical pain rating scales during this time. In the support and neutral non-support conditions, the confederate read her magazine silently during the wait period. In the support condition, however, she also briefly engaged the participant to tell her that she is there to support her and to wish her good luck. Cold Pressor Pain Task Participants underwent a CPT by submerging their dominant hand up to the wrist in a circulating tank of water at 4.0± 0.1 °C until they were no longer willing or able to tolerate the pain (i.e., pain tolerance) or until they reached the 3-min time limit. During the CPT, participants non-verbally rated their pain every 20 s by pointing to a numerical rating scale from 0 (no pain) to 100 (strongest pain imaginable). Ratings were completed out of view of the confederate to eliminate evaluation effects, given that the negative feeling of being evaluated by the supporter can confound the beneficial buffering effects of social support [45]. Following each pain rating, the experimenter immediately recorded the value. At the end of the 3 min or at pain tolerance, whichever came first, participants rated the intensity and unpleasantness of their pain on a numerical rating scale from 0 (no pain intensity or unpleasantness) to 100 (most intense or unpleasant pain imaginable) immediately before removing their hand from the water. These final ratings were obtained in order to distinguish between the overall pain experienced over the course of the task (i.e., mean CPT pain) and the intensity or unpleasantness of the pain experienced immediately following the task. The pain task instructions prior to testing included a thorough description of the difference between pain intensity (i.e., physical aspect of pain) and unpleasantness (i.e., emotional aspect
ann. behav. med.
of pain). For those participants who kept their hand in the water bath for the maximum time, the time to tolerance was recorded as 3 min. After drying their hand, participants began their post-task questionnaires (see below). Support Manipulation During the pre-task wait period and the CPT, confederates in the support condition periodically provided scripted statements of verbal emotional support (e.g., “You’re doing great! Remember you’re not alone”) and congratulated participants upon completion of the task (see ESM). In the neutral nonsupport condition, the confederate was also in the same room as the participant during both the wait period and the CPT but were verbally instructed (in front of the participant) to read a magazine and refrain from interacting with the participant while they were waiting for their turn to undergo the pain task. All research assistants were thoroughly trained for their roles as confederates in order to maintain consistency. Recovery Period In the support and neutral non-support conditions, the experimenter escorted the confederate out of the testing area when the participant had completed the post-task questionnaires (15 min after the start of the CPT) under the pretense that it was her turn to begin her baseline rest period. During the subsequent recovery period, participants sat quietly in the testing area until all salivary cortisol samples had been obtained (see below), signaling the completion of the study. Before debriefing the participants, the researcher asked follow-up questions to determine if the participant realized the true purpose of the study. In the support and neutral non-support conditions, the experimenter also asked participants about the second participant (i.e., confederate) and what role that they believed that she played in the study. No participant was aware of the deception or reported feeling evaluated in any way by the confederate. Questionnaires Preliminary Screening Questionnaire Due to the effects of depression [46, 47], trait anxiety [48, 49], and perceived stress [49] on laboratory pain and stress, each of these psychological variables was assessed via an online questionnaire prior to the laboratory visit. Severity of self-reported depressive symptoms was quantified using the Beck Depression Inventory (BDI), a 21-question multiple choice survey [50]. This scale comprehensively assesses dysphoric symptoms, including affective, cognitive, somatic, overt behavior, and interpersonal symptoms of depression. Each forcedchoice question has a set of at least four possible answer
choices, with increasing severity of depressive symptoms from 0 to 3. Scores may range from 0 to 53 with higher scores indicating increased depressive symptoms (>29 indicates severe depressive symptoms). Trait anxiety levels were measured using the Spielberger State-Trait Anxiety Inventory (STAI trait), a 20-question Likert-scale questionnaire assessing chronic feelings of anxiety [51]. Participants rate how they generally feel on a scale from 1 (not at all) to 4 (very much so). Scores may range from 20 to 80 with higher scores indicating greater levels of trait anxiety (>60 indicates severe trait anxiety). Self-perceptions of life stress during the last month were assessed using the Perceived Stress Scale (PSS), a 10-question multiple choice inventory [52]. Items assess how unpredictable, uncontrollable, and overloaded respondents view their lives and directly inquire about levels of experienced stress in the past month with answer choices ranging from 0 (never) to 4 (very often). Scores range from 0 to 40 with higher scores indicating greater overall perceived life stress. Finally, participants completed the Interpersonal Support Evaluation List (ISEL), a 40question inventory that measures perceived access to social support across four subscales: tangible, belonging, self-esteem, and appraisal support [53]. Scores on each subscale may range from 0 to 30, and each subscale was averaged to yield the reported measure of perceived access to social support. To control for the effects of oral contraception use on salivary cortisol levels, participants identified in the prescreening questionnaire whether or not they used birth control methods, and if so, what type. Baseline Questionnaires Given that previous studies reported affect [54], state anxiety [54, 55], and sleep quality [56] to affect pain and stress, participants completed questionnaires assessing each of these measures prior to the baseline rest period. Affect was quantified with the Positive and Negative Affect Schedule (PANAS), a 20-item multiple choice survey [57]. For each word describing a different feeling or emotion felt at the present moment, participants had a choice of ratings from 1 (very slightly or not at all) to 5 (extremely). The positive subscale consisted of ten words and a possible range from 10 to 50, with higher scores indicating more positive affect. The negative subscale consisted of ten words and a possible range from 10 to 50, with higher scores indicating more negative affect. State, or transitory, anxiety levels were assessed with the state portion of the Spielberger State-Trait Anxiety Inventory (STAI state), a 20-question inventory utilizing Likert scales [51]. The questionnaire contains 20 statements describing how an individual may feel at the present moment and participants rate how they currently feel on a scale from 1 (not at all) to 4 (very much so). Possible scores range from 20 to 80, with
ann. behav. med.
higher scores indicating greater levels of state anxiety. The Pittsburg Sleep Quality Index (PSQI), a 19-question variedresponse survey, measured recent self-rated sleeping habits and sleep quality [58]. The 19 items are combined to create seven component scores, each with a range of 0 (no difficulty) to 3 (severe difficulty). The seven component scores are then added to yield one global score, with a range of 0 to 21, and higher scores indicating more severe sleep difficulty. Post-Task Questionnaires Participants completed a CPT assessment in which they had to rate the difficulty, tension, effort, and concentration levels perceived during the CPT on a numerical rating scale from 0 to 10. Participants also completed a second iteration of the PANAS and STAI state. Additionally, participants completed a brief manipulation check of how supported they felt during the CPT on a series of seven-point Likert scales across six different axes: supportive-unsupportive, friendly-unfriendly, accepting-rejecting, close-distant, warm-cold, and helpfulunhelpful. Physiological Measurements Cardiovascular measurements were taken at 0, 5, and 10 min during the baseline rest period; 0 and 5 min of the wait period; and subsequently every 5 min starting immediately after the participant removed her hand from the water until the end of the experiment (30 min following the start of the CPT). All cardiovascular measurements were obtained using a programmable automatic blood pressure cuff (OSCAR 2, SunTech Medical). Saliva was collected in 1.5-mL Eppendorf tubes at the end of the baseline rest period and 20-, 25-, and 30-min post-CPT. Participants passively drooled into the tube for a maximum of 2 min per sample. Saliva samples were frozen within 30 min of collection at −20 °C until assayed. On the day of testing, all samples were centrifuged at 3000 rpm for 15 min to remove mucins and were analyzed in duplicate by enzyme immunoassay (Salimetrics, State College, PA). The mean intra-assay coefficient of variation was 8.7 %, and the inter-assay coefficient was 12.1 %.
Data Analysis Group differences in demographic factors and subjective measures (perceived stress, depression, sleep quality, perceived access to social support, baseline state anxiety, trait anxiety, and affect) were examined using a multivariate ANOVA for continuous variables and chi-square analyses for dichotomous variables as appropriate. Group differences in baseline
cardiovascular measures and cortisol were also investigated using a multivariate ANOVA. Where significant results emerged, post hoc analyses with Bonferroni corrections were conducted. In order to determine the subjective measures that should serve as covariates for all analyses assessing group differences in pain and stress measures, Pearson productmoment (Pearson’s r) correlations were performed. Due to group differences in baseline perceived stress levels and results indicating that trait anxiety, baseline positive affect, and perceived access to social support were correlated with various measures of pain and stress indices, these four subjective measures were included as covariates in all analyses of group differences in pain and stress. Multivariate analysis of covariances (ANCOVAs) were used to analyze group differences in pain intensity, pain unpleasantness, mean CPT pain (average pain rating during the CPT), and time to tolerance, as well as perceived social support during testing, difficulty, tension, concentration, and effort during the CPT. Separate ANCOVAs were also used to analyze delta scores for systolic blood pressure (SBP), diastolic blood pressure (DBP), heart rate (HR), negative affect, positive affect, and state anxiety. For cardiovascular measurements, delta scores were calculated by subtracting the average baseline measurement from the measurement taken immediately following the pain task. Delta scores for the subjective measurements were calculated by subtracting the baseline scores from the post-task scores. Where group differences emerged, Sidak-corrected post hoc comparisons were performed. Analysis of salivary cortisol levels over time was accomplished by calculating the formula for area under the curve with respect to ground (AUCg), an analysis that takes into account the change over time (20-, 25-, and 30-min postCPT) and the distance of these measures from ground, or baseline [59], and subsequently performing an ANCOVA. The AUCg was employed due to the lack of group differences in salivary cortisol concentrations at baseline. In order to determine if perceived access to social support impacted the stress- and pain-attenuating effects of social support in the laboratory, Pearson product-moment (Pearson’s r) correlations were performed in the entire sample and in each of the three groups separately. In order to reduce type I error, this analysis was limited to dependent variables which significantly differed as a function of support condition: pain intensity, pain unpleasantness, mean pain ratings, delta SBP, delta HR, and cortisol AUCg. Wherever significant main effects were found, omegasquared (ω2) calculations were performed to determine the strength of the effect sizes. Omega squared is less biased and more conservative than eta-squared and takes into account overall population variance [60]. Values of 0.01, 0.06, and 0.14 represent small, medium, and large effects, respectively [61].
ann. behav. med. Table 2 Intercorrelations among baseline subjective variables and pain and stress measures
Results Demographics and Subjective Measures No significant group differences emerged for oral contraceptive use or race (p>0.05 for all comparisons, Table 1). Furthermore, body mass index, age, trait anxiety, depression, sleep quality, perceived access to social support, and baseline state anxiety; affect; SBP; DBP; and HR did not differ between groups (p>0.05 for all comparisons, Table 1). There were, however, significant group differences in baseline perceived stress (F(2,72)=3.91, p=0.024), as the alone group rated higher than the neutral non-support group (p=0.021, Table 1). Correlational analyses among subjective measures and pain and stress variables revealed that trait anxiety was correlated with cortisol AUCg, r=0.23, p=0.05, while baseline positive affect was correlated with pain intensity, r= −0.28, p
