Annals of Behavioral Medicine

, Volume 42, Issue 1, pp 64–78

The Conjoined Effect of Naturalistic Perceived Available Support and Enacted Support on Cardiovascular Reactivity During a Laboratory Stressor

Authors

    • Department of PsychologyKarl-Franzens-University
  • Henrike Schlagert
    • Department of PsychologyJohannes Gutenberg-University Mainz
Original Article

DOI: 10.1007/s12160-011-9272-2

Cite this article as:
Schwerdtfeger, A.R. & Schlagert, H. ann. behav. med. (2011) 42: 64. doi:10.1007/s12160-011-9272-2

Abstract

Background

Studies have suggested that enacted social support has salutogenetic effects on cardiovascular activation during stress.

Purpose

This study aims to examine the conjoined effect of naturalistic perceived available support and enacted support on cardiovascular reactivity to a laboratory stressor.

Methods

Seventy-one participants assigned themselves to one of two conditions: enacted social support before the onset of a stressor or no support. Perceived available support was assessed via a questionnaire, and heart rate (HR), heart rate variability (HRV), mean arterial blood pressure (MAP), and baroreceptor reflex sensitivity (BRS) were analyzed in response to a speech task.

Results

Whereas perceived available support was unrelated to cardiovascular activity in the no-support condition, it was accompanied by attenuated HR and increased HRV in the enacted-support condition. Moreover, perceived available support was associated with greater HR, HRV, and BRS reactivity to the speech task and better MAP recovery.

Conclusion

Together, these findings support the assumption that different aspects of social support are related to different physiological processes.

Keywords

Blood pressureCardiovascular reactivitySocial supportStressNaturalistic available support

Social support and social network size have been proposed to constitute powerful psychosocial resources for health [1, 2]. For example, epidemiological studies have provided evidence to suggest that social support is associated with decreased morbidity and mortality. In particular, it has been found that high levels of social support were related to lower mortality rates from cardiovascular diseases [35]. Two pathways have been discussed that may mediate the relationship between social support and health: One path involves salutogenetic effects on health-promoting behavior, and the other path suggests that social support may impact different physiological systems within the organism. In particular, it has been found that social support has beneficial effects on the cardiovascular [610], the endocrinological [1113], and the immunological systems [1416].

Whereas epidemiological evidence linking social support to health outcomes has relied mainly on questionnaire measures of naturalistic available support or measures of social network size, laboratory studies focusing on physiological mechanisms have applied enacted social support manipulations involving the presence or absence of a third party (i.e., an accompanying friend or confederate of the experimenter) while participants performed a stressful task. In sum, although there is considerable evidence that enacted social support dampens cardiovascular stress reactivity [for a meta-analysis, see 17], there are also studies that have suggested rather elevated reactivity when a companion was present during task performance [e.g., 18]. These contradictory findings may be explained by the evaluative potential inherent in these situations. For example, when performing a challenging psychological task, the presence of a third party may increase challenge appraisals and lead to stronger effort mobilization, which, in turn, could result in elevated cardiovascular responses [18]. Thus, it seems important to minimize evaluation potential in studies that aim to examine the health-protective effects of social support in laboratory settings.

Importantly, although the experimental approach to social support is worthwhile because it allows for verifying a causal pathway from social support to beneficial physiological mechanisms, it is limited by comparably low ecological validity. For example, stressful situations in daily life often imply that individuals are faced with an aversive encounter while they are alone (because social support is not available in a particular situation), but receive social support before or after the occurrence of a stressor. Moreover, it should be noted that this paradigm is dissonant with operationalizations of social support as applied in epidemiological research linking support to disease [19, 20]. Thus, it would be necessary to relate such naturalistic measures of social support to cardiovascular reactivity in the laboratory. However, surprisingly few studies have examined this.

We were able to identify only four studies that analyzed associations between naturalistic social support measures and cardiovascular reactivity in the laboratory. Using a rather small sample size, Tardy, Thompson, and Allen [21] found that individuals who reported high levels of social support showed attenuated systolic blood pressure (SBP) and mean arterial pressure (MAP), and also elevated heart rate (HR) throughout the experiment. There was, however, no reliable relationship with cardiovascular reactivity to a speech task. Other authors have reported that the perceived quantity of social support was related to elevated cardiovascular reactivity during a mental challenge [22]. Hughes [19] also found evidence for elevated cardiovascular reactivity to a laboratory challenge for men with high levels of naturalistic social support. Finally, a study conducted by Roy, Steptoe, and Kirschbaum [23] found that the perception of a large social support network as compared to a small social support network was associated with elevated cardiovascular responses to a mental arithmetic task, and with a more pronounced blood pressure recovery after the stressor was terminated.

Taken together, these findings challenge the view of a general dampening effect of social support on cardiovascular stress reactivity as often reported in studies applying social support manipulations. Hence, it seems that naturalistic measures of social support cover a different aspect of social support. Indeed, it has recently been proposed that perceived (i.e., naturalistic) support should be separated from received (i.e., enacted) support, because both constructs seem to be linked with different psychological concomitants and health outcomes [20, 24]. In particular, Uchino [24] suggested that perceived support constitutes a stable interindividual difference factor that develops early in life and is related to feelings of control, self-efficacy, and the ability to “cope more effectively, flexibly, and proactively with life stressors” (Uchino [24], p. 241). By contrast, received support depends more on situational factors and its effectiveness may vary with the context (e.g., when the type of support matches the needs or challenges of the stressful situation). In line with this reasoning, Roy et al. [23] proposed that naturalistic measures of social support reflect a latent resource variable that is related to competence and self-efficacy. They speculated that naturalistic social support is indicative of the ability to more effectively cope with stress, involving a stronger mobilization of resources to a challenge (i.e., stronger reactivity), and a more rapid adaptation after stressor termination (i.e., better recovery). Indeed, such a response pattern has been referred to as physiological toughness [25, 26], which has been associated with lower allostatic load [27]. Hence, in line with previous research, naturalistic social support may exert its influence on health via more effective coping strategies during aversive encounters.

Interestingly, the same pattern of adaptive response has been associated with emotion regulation [2830]. In particular, it has been suggested that parasympathetic (i.e., vagal) innervation of the heart plays a key role in enabling the organism to rapidly adapt to changing environmental demands. A healthy organism is assumed to show an elevated parasympathetic tone, combined with the capacity to rapidly withdraw inhibitory parasympathetic efference to cope with environmental demands (i.e., meet several metabolic demands, including increased attention and information processing) [30]. Vagal innervation of the heart can be quantified by analyzing respiratory-related heart rate variability (HRV). Indeed, recent research suggests that elevated tonic HRV together with a pronounced reduction in HRV during stress is indicative of a better emotion regulation capability [28, 31, 32]. Even more, the Polyvagal Theory [33, 34] suggests that HRV may indicate the activity of a so-called social engagement system, linking vagal tone with different cranial nerves of the head that are involved in the regulation of various behaviors that play key roles in the exchange with the environment (i.e., facial expressions, looking, listening, vocalizing, self-soothing, and calming behaviors). It has been suggested that this network serves to adapt the organism to the dynamic nature of social communication. Hence, social engagement, deployment of emotion regulation strategies, and perceived security in attachment relationships should be accompanied by elevated tonic HRV and a more pronounced reduction in HRV during stress. Surprisingly, however, laboratory research linking social support to cardiovascular reactivity has largely neglected the role of HRV, although there is considerable evidence that affiliation and social interactions with intimate individuals are associated with higher HRV [35, 36].

Aim of the Study

Taken together, the relationship between social support and cardiovascular reactivity is rather complex and seems to depend on the operationalization of social support. Although both lines of research (i.e., studies examining associations with naturalistic social support and studies manipulating social support) appear to have produced inconsistent findings with respect to cardiovascular reactivity, they seem to converge with respect to the suggested salutogenetic effect of social support. Importantly, to the authors’ knowledge, only a few studies have examined the conjoined impact of both types of social support on cardiovascular responses during an acute laboratory challenge. The study of conjoined effects seems particularly important because a history of previous positive experiences of social support could have an impact on the effect of enacted social support in a particular situation [20, 24]. For example, it has been suggested that past experiences with naturalistic support could signify implicit support and reduce threat appraisals in light of aversive encounters when a third party is present during a laboratory challenge [37]. In accordance with this assumption, Fontana et al. [37] found that women who were accompanied by a friend in the laboratory and who were highly satisfied with the support they generally received from others exhibited smaller increases in HR to psychologically challenging tasks than women who were not satisfied with the support they generally received. A similar finding was reported by Kors, Linden, and Gerin [38]. These authors found that perceived closeness to a friend and length of the friendship was associated with attenuated SBP reactivity to a mental stressor when the friend was present.

Whereas salutogenetic effects seem to prevail when naturalistic social support and enacted support coincide, the empirical basis is more challenging when focusing on naturalistic social support in the absence of enacted support. On the one hand, there appears to be evidence that naturalistic social support could signify implicit support in the laboratory, leading to higher self-efficacy and more effective coping when encountering aversive situations alone [23, 24]. On the other hand, it has been hypothesized that individuals with a positive history of social support may habitually show a preference for consulting members of their social network to cope with stressors [19]. However, when social support is unavailable in these situations, naturalistically supported individuals may experience a coping deficit, leading to higher threat appraisals, and ultimately, elevated stress reactivity. Whichever interpretation is more valid (coping benefit or coping deficit), both viewpoints would be compatible with the finding of elevated cardiovascular reactivity in naturalistically supported individuals, thus challenging the interpretability of the data. Therefore, it may be worthwhile to consider cardiovascular recovery as well. Delayed cardiovascular recovery has been proposed to constitute a more profound risk factor for cardiovascular diseases than peak reactivity to a challenge [39, 40]. In particular, it has been hypothesized that a response pattern of elevated cardiovascular reactivity and comparably fast recovery may reflect an adaptive response to a stressor, which is related to comparably low allostatic load. Conversely, elevated peak reactivity together with delayed recovery may be indicative of higher allostatic load [27].

Hence, the aim of this study was to examine the joint effect of enacted support and perceived available support on cardiovascular reactivity and recovery. Moreover, stressful situations in daily life often imply that individuals are faced with an aversive encounter while they are alone, but receive social support before or after the occurrence of a stressor. Thus, the manipulation of enacted support was restricted to the baseline period of the task. This procedure was also aimed at preventing the well-observed finding that socially evaluative cues inherent in many social support situations may counteract the salutogenetic effects of social support [4143], leading to elevated, rather than attenuated, cardiovascular reactivity.

We expected that naturalistic social support would be related to attenuated levels of baseline cardiovascular activity, especially when enacted support was available [37]. Moreover, in light of previous studies on the relationship between naturalistic available support and cardiovascular reactivity [22, 23], we predicted that naturalistic available support would be related to elevated cardiovascular reactivity during a psychological challenge, but to attenuated cardiovascular activation during recovery. We recorded various cardiovascular variables to gain a broader view of the salutogenetic effects of social support. In particular, HR, HRV, MAP, and baroreceptor reflex sensitivity (BRS) were analyzed.

Method

Participants

Participants were 71 volunteers (55 women). They had a mean age of 22.54 years (SD = 3.57) and a mean body mass index (BMI) of 21.60 kg/m2 (SD = 2.35). In the sample, 11% were smokers and 77% reported regular physical exercise. No participant reported cardiovascular diseases or the intake of cardiovascular medication. Participants were not allowed to consume caffeine or cigarettes 2 h prior to the experiment. Participants were recruited through oral communication at the university campus. They were informed that an experiment on social interaction and cardiovascular activation would be conducted and that they could take part individually or together with an intimate person. They were paid 10 Euros for participating (alternatively, they could receive course credit when applicable).

Study Design and Stressor

Participants assigned themselves to one of two conditions: enacted-support condition (n = 33, 25 women) and no-support condition (n = 34, 28 women). We decided not to randomly assign participants to the conditions to assure ecological validity. Randomly assigning participants to either the enacted-support condition or no-support condition would imply that participants in the enacted-support condition would need to bring an accompanying person with them to the lab, irrespective of whether the participant or his/her companion would be willing or able to do so. This could bias the salutogenetic effects of social support to a considerable degree. Other studies have manipulated social support to be present or absent by instructing a collaborator of the experimenter to support or not to support the participants [e.g., 6, 45]. Again, this is an artificial situation that is not directly comparable to social support as it unfolds in everyday life. Hence, our aim was to keep social interaction during the enacted-support condition as natural as possible. During the enacted-support condition, participants were instructed to attend the first phase of an experiment together with another person (partner, close friend, family member, student fellow). Specifically, in the enacted-support condition, participants could communicate with the other person during a resting period of 10 min prior to the experiment during which the experimenter was absent. Approximately 76% of the sample was accompanied by a close friend, 12% by their partner, and 12% by a fellow student.

During the no-support condition, participants were placed alone in the laboratory, but there was also a short period of interaction (about 5 min) with the female experimenter. Following a 5-min resting period during which participants were left alone, the experimenter re-entered the room, checked the physiological equipment, and posed standardized questions to the participant. In particular, she asked participants whether they were seated comfortably and then asked them to disclose their study interests and previous experiences with psychophysiological experiments. The experimenter was instructed not to act emotionally and not to smile while interacting with the participant, thus aiming for a rather neutral atmosphere. It should be noted that a perceived positive relationship with the experimenter has been found to significantly diminish cardiovascular stress reactivity [45]. After a conversation period of approximately 5 min, she left the room. We decided to restrict interaction in the no-support condition to 5 min in order to keep the situation as neutral as possible. Longer interactions could lead to acquaintance, or otherwise a completely neutral interaction of 10 min could be perceived as rather weird and unnatural. Special care was taken to ensure that the experimenter was unknown to the participant. In particular, when the experimenter was familiar with one particular participant (in 23% of the cases), a same-sex confederate took over the interaction part of the experiment. In order to reduce the influence of the experimenter throughout the proceeding experiment, we used standardized instructions via a computer screen and applied computerized questionnaires in both conditions.

A self-relevant stress task of 3 min duration was applied thereafter. Participants were instructed to prepare and deliver a speech elaborating on their personal strengths and assets and their personal weaknesses in front of a camera. The task was embedded between a preparation period in which participants were instructed to prepare the speech for 3 min and a recovery period (3 min). Importantly, all participants prepared and delivered the speech alone.

Measures

Perceived Available Support

Perceived available support was assessed using the Berlin Social Support Scales [46]. The Berlin Social Support Scales were developed in the area of coping with cancer, but can also be applied in a variety of different contexts. The scale “perceived available support” comprises eight items assessing emotional (e.g., “There is always someone there for me when I need comforting”) and instrumental social support (e.g., “There are people who offer me help when I need it”) in naturalistic settings. Items are rated on a 4-point Likert scale ranging from 1 (not true) to 4 (absolutely true). The mean score of this scale was M = 3.68 (SD = 0.36), indicating rather high levels of perceived available support. The reliability was good (Cronbach’s alpha = 0.87).

Quality of the Interaction

The quality of the interaction with either the experimenter (no-support condition) or the accompanying person (enacted-support condition) was rated via four items (pleasantness of the interaction, perceived support during the interaction, satisfaction with the interaction, and a general rating of the quality of the interaction). Items were rated on 7-point bipolar Likert-type scales between the poles −3 (unpleasant, not supported, not satisfied, and not good) and +3 (pleasant, supported, satisfied, and good). The mean score was M = 2.44, SD = 0.57, indicating rather high levels of rated quality. The reliability of this scale was good (Cronbach’s alpha = 0.83).

Cardiovascular Reactivity

Cardiovascular variables were recorded throughout the phases of the task. We focused on HR, HRV, BRS, and MAP. HR and HRV were recorded via an electrocardiogram, applying a chest lead. Ag/AgCl-electrodes were filled with Hellige gel and attached to the torso with adhesive collars. The signal was amplified using a Coulbourn biosignal amplifier (S75-05) and high-pass filtered with 1 Hz to yield a time constant of 0.16 s. It was sampled with 1,000 Hz to allow accurate detection of the R-wave. Blood pressure was recorded by means of a tonometric device that allows continuous monitoring of beat-to-beat changes in blood pressure (Colin CBM-7000). Recording is based on arterial tonometry, which makes use of 30 cutaneously applied pressure sensors (piezoresistive transducers) that compress the top 15% of the underlying radial artery against the bone. From the resulting contact stress, the intraluminal blood pressure can be approximated [47]. The sensor is placed with a wrist brace over the radial artery. Oscillographic cuff measurement is casually needed for calibration of the arterial tonometer. A normal-sized adult cuff was applied at the same arm site as the tonometric sensor. Advanced multiplexing electronics combined with automatic sensor positioning relative to the artery ensure optimal placement of the sensor. Two specific values are crucial for obtaining reliable data: hold-down pressure and signal strength. Whereas hold-down pressure defines the pressure applied to the artery in millimeters of mercury and should not exceed 140 mmHg or drop below 40 mmHg, signal strength refers to the strength of the tonometric signal in relation to the oscillographic signal. Maximum signal strength is 100%, and values should not drop below 60%. We calibrated the signal at the beginning and at the end of the experiment. Mean hold-down pressure varied between 65.09 and 61.08. Mean signal strength was 84% at the beginning and 91% at the end of the experiment. Thus, blood pressure could be reliably recorded, and readings seemed trustworthy. The Colin CBM-7000 has been proven to satisfy the standards of the Association for the Advancement of Medical Instrumentation for mean systolic and diastolic blood pressure measurements as well as the FDA (US Food and Drug Administration) standard for intensive care units [48]. This device has also been validated in cardiovascular stress research where it has shown low artifact ratings and high accuracy [49].

Control Variables

Because participants were not randomly assigned to conditions, other questionnaires were given to examine comparability across conditions. We focused on depression and coping because these variables have been consistently related to cardiovascular reactivity in previous studies. In particular, depressive symptoms have been related to either elevated or blunted cardiovascular reactivity in the laboratory [5053] and dispositional coping has also been associated with cardiovascular reactivity during laboratory stressors. In particular, high cognitive avoidant coping and low vigilance coping (i.e., repressive coping) have been associated with increased cardiovascular reactivity to self-relevant stressors [5456].

Depressive Symptoms

We assessed depressive symptoms by means of the trait version of the State-Trait Depression Scales ([57]; German version [58]). The scale was developed to assess depressive mood in healthy individuals and excludes somatic symptoms, which are more prevalent in clinical depression. It comprises 10 items assessing cognitive–affective symptoms, which can be subdivided into dysthymia (e.g., “I feel blue”) and euthymia (e.g., “I feel strong”). Participants are instructed to rate how they generally feel. Response options are given on a 4-point frequency scale between the poles 1 (never) and 4 (very often), resulting in a possible range of scores between 10 and 40. The mean total score for this sample was 16.52 (SD = 3.35). The reliability of this scale proved to be acceptable (Cronbach’s alpha = 0.80).

Coping

We assessed dispositional coping with the German version of the Mainz Coping Inventory [59]. The Mainz Coping Inventory was developed to assess preferred coping strategies on a dispositional level. It measures vigilant and cognitive avoidant coping strategies directly in four ego-threatening (public speaking, exam, job interview, mistake on the job) and four physically threatening situations (dentist, inexperienced driver, group of people, turbulent flight) of varying controllability. For each situation, five vigilant and five avoidant response options are given in a true–false format. Because we applied a self-relevant stressor, only the ego-threat subscales were filled out. Answers were summed separately for vigilance and for cognitive avoidance items across all ego-threatening situations to yield measures of dispositional coping. The reliability of both scales was acceptable (for cognitive avoidant coping: Cronbach’s alpha = 0.70; for vigilant coping: Cronbach’s alpha = 0.71). The Mainz Coping Inventory has been well validated. Vigilant coping is related to planning, restraint coping, and anxiety, whereas avoidant coping is related to positive problem orientation, positive affectivity, self-esteem, and repressive coping [59, 60].

Procedure

Upon arriving at the laboratory, participants signed informed consent forms and were told that if they wished to discontinue at any time, they could leave without giving a reason. They then filled out the questionnaires on depression, coping, social support, and demographic and lifestyle variables. Subsequently, the electrodes were attached, the calibration of the blood pressure device was conducted, and participants were requested to relax for the next 10 min. In the enacted-support condition, they were instructed that they could communicate with their affiliate during this time frame, but should not move too much because of the sensitive technical equipment. The experimenter then left the room. In the no-support condition, the experimenter returned to the laboratory after approximately 5 min and asked standardized questions about study interests and previous experiences with experiments. Special care was taken that the experimenter behaved in a neutral fashion, not confirming participants’ answers and not smiling. After the baseline period, participants were asked to rate the quality of the social interaction. The experiment then proceeded with the instructions of the speech task. Participants were asked to prepare and deliver a speech. They were asked to imagine that they were invited to a job application interview, elaborating on their strengths and weaknesses in front of a camera. Further, they were instructed that their speech would be evaluated by experts for appearance and authenticity. After the preparation period (3 min), the camera was adjusted to enhance social-evaluative cues. After that, speech delivery ensued for 3 min. After the speech, the recovery period (3 min) followed. Finally, the blood pressure calibration was conducted and thereafter the electrodes were detached and participants were debriefed about the purpose of the study. Participants were instructed not to communicate with other to-be participants about the content of this experiment.

Data Reduction and Analysis

The electrocardiogram was analyzed by means of a semiautomatic peak detection software (written with LABVIEW 6.0i). The electrocardiogram of each participant was visually inspected on a 30-s basis and low-pass filtered with 30 Hz to overcome gross movement artifacts. Interbeat intervals (IBIs) were calculated in milliseconds for each 30-s segment. Extraordinarily strong successive IBI variations were corrected if necessary by a moving average procedure if they differed by more than a multiplier of 1.5 or 0.7 from the previous IBI. HR for each segment was calculated by dividing 60,000 by the respective interbeat intervals. Additionally, HRV was calculated as the square root of the mean of the sum of the squares of differences between successive IBIs (RMSSD) to obtain a time-domain measure of parasympathetic control of the heart [6163]. Due to the skewed distribution of this variable, it was log-transformed prior to analyses (lnRMSSD).

The tonometric blood pressure device was attached through the serial line to an AT-type computer for further offline analyses. MAP was recorded in millimeters of mercury throughout the experiment and averaged across each period (baseline, preparation, speech, and recovery). In addition, BRS was calculated offline using the IBI series of the electrocardiogram and the SBP readings as recorded with the Colin CBM-7000. BRS was calculated with the sequence method [64]. That is, the time series of SBP and IBI were scanned with a self-written software (via LABVIEW 6.0i), capable of identifying sequences in which SBP and IBIs concurrently increased (up sequence) or decreased (down sequence) over three or more beats (zero lag). The minimum change in SBP was required to be 1 mmHg and the minimum change in IBI was required to be 2 ms. The linear regression of SBP and IBI was then calculated for each sequence separately and averaged for each experimental period prior to analysis. The slope (sensitivity) of the BRS was expressed as change in IBI (in milliseconds) per millimeters of mercury change in SBP.

For analyzing cardiovascular activation during the course of the experiment, we applied linear mixed-effects (LME) modeling for each cardiovascular variable separately. In a first step, we aimed to predict cardiovascular activation by sex (0 = men, 1 = women), age (grand-mean centered), BMI (grand-mean centered), task period (quadratic function), condition (enacted-support condition, no-support condition), perceived available support (mean centered), and the interaction of condition and perceived available support, thereby allowing different intercepts and slopes for participants. In a next step, we aimed to analyze cardiovascular reactivity and its relation with both naturalistic social support and enacted social support in more detail. Again, LME modeling was applied, but this time we entered task period as a factorial variable (0 = baseline/reference, 1 = preparation, 2 = speaking, 3 = recovery). This allowed us to examine reactivity with respect to the baseline measure for each period separately. Importantly, we analyzed the interactions of condition and period, perceived available support and period, and condition × perceived available support × period as fixed effects predictors, and participants as random effects. Moreover, we allowed heteroscedasticity with respect to period for each participant (random intercepts and random slopes for each period). This model proved superior when compared to more simple models according to log-likelihood tests.

LME models have been recommended in psychophysiological research [65] because they offer several advantages over traditional ANOVA-based algorithms (e.g., more efficient handling of missing data, more powerful tests that allow for modeling the error variance, including higher flexibility). We applied the open source language and environment for statistical computing R (version 2.10.1) [66], using the lmer program of the lme4 package (version 0.99375-32; for an overview of lme4 [67]). We report regression coefficients (b; absolute effect size in milliseconds), standard errors (SE), and their ratio (t statistic). As the formulas for the degrees of freedom for inferences based on t or F distributions do not apply in mixed-effects models, the calculation of p values gets problematic and can lead to anticonservative decisions. However, for comparably large data sets, it has been recommended to interpret t values exceeding 2 SE as significant [68]. Four participants were excluded from the analyses of cardiovascular reactivity because they reported caffeine intake before the experiment, leaving a total of 67 individuals.

Results

In a first set of analyses, we aimed to examine whether the participants in the two conditions were comparable on the psychological, demographic, or lifestyle measures, thus assuring comparability across conditions. There were no differences with respect to age, t(69) = 0.061, ns; sex, χ2 = 0.40, ns; BMI, t(69) = −0.66, ns; perceived available support, t(69) = 0.34, ns; depressive symptoms, t(69) = −0.77, ns; cognitive avoidant coping, t(69) = 1.53, ns; vigilant coping, t(69) = 0.45, ns; smoking, χ2 = 1.37, ns; and physical exercise, χ2 = 0.25, ns.

In line with predictions, however, there was a difference between conditions with respect to the quality of the social interaction. It was found that participants in the enacted-support condition rated the interaction as being of higher quality (M = 2.56, SD = 0.54) as compared to the participants in the no-support condition (M = 2.33, SD = 0.58), t(69) = 1.77, p = 0.04, one-tailed. The effect size (Cohen’s d) was moderate (d = 0.42), though.

Preliminary Analyses of the Cardiovascular Variables

Figure 1 depicts the various cardiovascular measures used in the course of the experiment; measures for each condition are presented separately. To examine whether there were any differences in cardiovascular activation with respect to the experimental condition and naturalistic social support, we calculated LME models for each cardiovascular variable separately, thereby treating task period as a quadratic variable. We found a significant quadratic effect of task period for each variable (b = −116.79, t = −11.41 for HR; b = 0.95, t = 6.59 for lnRMSSD; b = −101.67, t = −10.37 for MAP; b = 33.16, t = 5.16 for BRS), documenting significant changes in response to the stressor. We will focus on these reactivity measures later in more detail. Moreover, we found a significant main effect of condition for HR (b = 4.92, t = 2.05), documenting lower HR in the enacted-support condition than in the no-support condition (see also Fig. 1). For lnRMSSD, there was a significant interaction of perceived available support and condition (b = −0.24, t = −2.12), and this interaction was marginally significant for HR (b = 14.06, t = 1.97). We examined these interactions further by calculating simple-slope analyses. Therefore, we rescaled perceived available support at the standard deviation, thus allowing us to analyze individuals high (1 SD above the mean) and low (1 SD below the mean) in perceived available support, thereby controlling for all other covariates. Specifically, two continuous variables were calculated that were scaled to zero at either 1 SD above (i.e., high perceived available support) or 1 SD below the mean (i.e., low perceived available support). Then, two additional analyses were run in which the newly computed high and low perceived available support variables were separately entered into the equation replacing the original perceived available support variable. Importantly, this kind of analysis makes use of the whole sample size, thus retaining the same statistical power as the previous models.
https://static-content.springer.com/image/art%3A10.1007%2Fs12160-011-9272-2/MediaObjects/12160_2011_9272_Fig1_HTML.gif
Fig. 1

Cardiovascular activation throughout different task periods for the enacted-support condition (dashed lines) and the no-support condition (solid lines). Whiskers indicate ± 1 standard error. Individuals in the enacted-support condition showed significantly lower Heart Rate than participants in the no-support condition (left side, top). No other group effects were significant. lnRMSSD Heart Rate Variability, MAP Mean Arterial Blood Pressure, BRS Baroreceptor Reflex Sensitivity

The findings of these analyses are depicted in Fig. 2. Individuals low in perceived available support showed virtually no difference in HR between conditions (b = −0.06, t = −0.02), whereas individuals high in perceived available support exhibited lower HR in the enacted-support condition than in the no-support condition (b = 9.90, t = 2.95). A similar—but inverse—pattern was found for lnRMSSD. Again, participants with comparably low levels of perceived available support showed no substantial difference between conditions (b = 0.015, t = 0.26), but those with comparably high levels of perceived available support were found to exhibit significantly higher lnRMSSD in the enacted-support condition as compared to the no-support condition (b = −0.16, t = −2.92). Taken together, these findings indicate that naturalistic social support is associated with cardiac deactivation in the course of the task when enacted support is available prior to a stressor, and it is unrelated with cardiac activity when individuals do not receive support prior to a stressful task.
https://static-content.springer.com/image/art%3A10.1007%2Fs12160-011-9272-2/MediaObjects/12160_2011_9272_Fig2_HTML.gif
Fig. 2

Interactions of condition and perceived available support on heart rate (HR; top graph) and heart rate variability (lnRMSSD; bottom graph). Participants who reported high levels of perceived available support (1 SD above the mean) as compared to participants with low levels of perceived available support (1 SD below the mean) exhibited lower cardiac activation when they were accompanied by another person (social condition). There were no significant differences for the nonsocial condition

Reactivity of HR and HRV

The results of the analyses of HR and HRV reactivity can be found in Tables 1 and 2. For HR, there were significant increases from baseline to both preparation (b = 3.99, t = 2.81) and speech (b = 16.12, t = 7.62), indicating that the stressor was effective. Moreover, HR dropped below baseline levels during recovery (b = −4.65, t = −4.85). Perceived available support was significantly negatively associated with HR (b = −13.44, t = −2.44), documenting lower baseline HR with increasing levels of naturalistic social support. There was also a significant interaction of perceived available support and condition (b = 14.15, t = 2.13). Again, we examined this effect further by calculating simple-slope analyses as described earlier. These analyses revealed that individuals with comparably low levels of perceived available support did not differ significantly on baseline HR across conditions (b = −0.41, t = −0.24). However, individuals with comparably high levels of perceived available support showed attenuated baseline HR in the enacted-support condition as compared to the no-support condition (b = 4.60, t = 2.95). Hence, these results essentially replicated our previous findings (see Fig. 2), implying that the combined effect of perceived available support and enacted social support resulted in cardiac deactivation. Furthermore, there were significant interactions of perceived available support and preparation (b = 10.35, t = 2.19), and perceived available support and speech (b = 20.43, t = 2.90). These findings indicated that individuals who reported high levels of perceived available support showed considerably higher HR increases during preparation (b = 7.72, t = 5.58) and speech (b = 24.36, t = 11.82) as compared to individuals who reported lower levels of perceived available support (b = 2.14, t = 1.44 for preparation and b = 13.31, t = 6.03 for the speech). The result for the speech period is depicted in Fig. 3 (top graph). No other effects were significant. In particular, there were no significant three-way interactions of period, condition, and perceived available support.
Table 1

Linear mixed-effects model for predicting heart rate (HR)

Variable

HR

ba

SE

t

Intercept

76.61

2.63

29.13*

Sex (0 = men, 1 = women)

1.44

2.81

0.51

Age

−0.23

0.31

−0.76

BMI

−0.61

0.50

−1.22

PAS

−13.44

5.50

−2.44*

Condition (0 = social, 1 = nonsocial)

4.19

2.24

1.87

Condition × PAS

14.15

6.65

2.13*

Preparation (vs. BL)

3.99

1.42

2.81*

Speech (vs. BL)

16.12

2.12

7.62*

Recovery (vs. BL)

−4.65

0.96

−4.85*

Condition × preparation

1.88

1.98

0.95

Condition × speech

5.43

2.95

1.84

Condition × recovery

1.03

1.33

0.78

PAS × preparation

10.35

4.73

2.19*

PAS × speech

20.43

7.04

2.90*

PAS × recovery

0.09

3.19

0.03

Condition × PAS × preparation

−4.96

5.86

−0.85

Condition × PAS × speech

−9.67

8.73

−1.11

Condition × PAS × recovery

1.70

3.95

0.43

BMI body mass index, PAS perceived available support, BL baseline

*p < 0.05

aUnstandardized partial regression coefficients

Table 2

Linear mixed-effects model for predicting heart rate variability (lnRMSSD)

Variable

lnRMSSD

ba

SE

t

Intercept

1.65

0.05

36.19*

Sex (0 = men, 1 = women)

−0.02

0.05

−0.50

Age

−0.01

0.01

−1.49

BMI

0.01

0.01

1.41

PAS

0.33

0.10

3.43*

Condition (0 = social, 1 = nonsocial)

−0.08

0.04

−2.03*

Condition × PAS

−0.29

0.12

−2.45*

Preparation (vs. BL)

−0.07

0.02

−3.24*

Speech (vs. BL)

−0.13

0.03

−4.01*

Recovery (vs. BL)

0.02

0.02

1.34

Condition × preparation

0.04

0.03

1.37

Condition × speech

−0.03

0.05

−0.59

Condition × recovery

0.02

0.02

0.76

PAS × preparation

−0.12

0.08

−1.57

PAS × speech

−0.29

0.11

−2.69*

PAS × recovery

−0.05

0.06

−0.98

Condition × PAS × preparation

0.05

0.09

0.51

Condition × PAS × speech

0.13

0.13

0.97

Condition × PAS × recovery

0.05

0.07

0.70

BMI body mass index, PAS perceived available support, BL baseline

*p < 0.05

aUnstandardized partial regression coefficients

https://static-content.springer.com/image/art%3A10.1007%2Fs12160-011-9272-2/MediaObjects/12160_2011_9272_Fig3_HTML.gif
Fig. 3

Interactions of task period and perceived available support on heart rate (HR; top graph) and mean arterial pressure (MAP; bottom graph). Individuals with high levels of perceived available support (1 SD above the mean) exhibited elevated cardiac reactivity to the speech task and more pronounced blood pressure recovery after the stressor than individuals with low levels of perceived available support (1 SD below the mean)

The results were comparable for HRV. In particular, lnRMSSD significantly decreased in response to preparation (b = −0.07, t = −3.24) and speech (b = −0.13, t = −4.01), indicating vagal withdrawal to stress. Perceived available support was positively associated with baseline lnRMSSD (b = 0.33, t = 3.43), suggesting increasing vagal tone in naturalistic socially supported individuals. There was also a significant main effect of condition (b = −0.08, t = −2.03), indicating that vagal tone was higher during baseline in the enacted-support condition as compared to the no-support condition. Moreover, a significant condition × perceived available support interaction restricted this effect (b = −0.29, t = −2.45). Again, simple-slope analyses were calculated to examine this interaction further. The findings replicated our preliminary analyses (see Fig. 2). For participants low in perceived available support there was no significant difference between conditions (b = 0.01, t = 0.37) but individuals high in perceived available support showed a significantly higher lnRMSSD in the enacted-support condition as compared to the no-support condition (b = −0.09, t = −3.30), suggesting that they exhibited a higher vagal tone during the enacted-support condition baseline as compared to individuals during the no-support condition baseline, whereas there was no such difference in individuals with low levels of perceived available support. Finally, there was a significant interaction of perceived available support and the speech phase (b = −0.29, t = −2.69). Rescaling perceived available support at the standard deviation revealed that individuals with high scores on perceived available support showed a comparably stronger reduction in vagal tone during the speech task (b = −0.22, t = −7.07) than individuals with low scores on perceived available support (b = −0.06, t = −1.87). Hence, this finding mirrored the results for HR and suggests that perceived available support is related to elevated cardiac reactivity during active laboratory challenges.

Reactivity of MAP and BRS

The results for MAP and BRS are depicted in Table 3. For both variables, there were significant effects for the speech phase (b = 18.44, t = 10.13 for MAP; b = −3.39, t = −2.55 for BRS), indicating an increase in MAP and a decrease in BRS in response to the stressor. There was also a significant increase in MAP during preparation (b = 8.35, t = 4.25). Moreover, a two-way interaction of perceived available support and recovery for MAP (b = −23.01, t = −2.62) indicated that participants with high levels of perceived available support exhibited lower MAP during recovery as compared to participants with low levels of perceived available support. Again, we rescaled perceived available support at the standard deviation and reanalyzed the model. It was found that participants scoring high in perceived available support showed a marginally significant decrease during recovery relative to the baseline period (b = −5.01, t = −1.94), whereas those scoring low on perceived available support exhibited a significant increase (b = 6.23, t = 2.24). This finding is depicted in Fig. 3 (bottom graph). For BRS, there was evidence of a stronger reduction during the speech stressor in individuals with high levels of perceived available support as indicated by a significant perceived available support × speech interaction (b = −14.73, t = −3.34). Simple-slope analyses revealed a more pronounced physiological stress response in individuals with high levels of perceived available support (b = −7.67, t = −5.94) as compared to individuals with low levels of perceived available support (b = −0.81, t = −0.59). No other effects were significant.
Table 3

Linear mixed-effects models for predicting mean arterial blood pressure (MAP; left side) and baroreceptor reflex sensitivity (BRS; right side)

Variable

MAP

BRS

ba

SE

t

ba

SE

t

Intercept

79.17

4.07

19.47*

18.66

2.17

8.59*

Sex (0 = men, 1 = women)

0.50

4.31

0.12

2.39

1.98

1.21

Age

0.93

0.47

1.97

−0.16

0.22

−0.73

BMI

1.24

0.77

1.61

−0.01

0.36

−0.04

PAS

−7.58

8.60

−0.88

10.74

5.40

1.99

Condition (0 = social, 1 = nonsocial)

−1.48

3.51

−0.42

−1.25

2.23

−0.56

Condition × PAS

16.04

10.41

1.54

−5.94

6.61

−0.90

Preparation (vs. BL)

8.35

1.97

4.25*

−2.47

1.47

−1.67

Speech (vs. BL)

18.44

1.82

10.13*

−3.39

1.33

−2.55*

Recovery (vs. BL)

2.32

2.65

0.88

1.85

1.30

1.43

Condition × preparation

−0.56

2.74

−0.21

0.92

2.06

0.45

Condition × speech

−3.53

2.54

−1.39

−1.70

1.85

−0.92

Condition × recovery

−3.42

3.71

−0.92

−1.09

1.81

−0.60

PAS × preparation

−3.39

6.54

−0.52

−4.88

4.91

−0.99

PAS × speech

−7.76

6.06

−1.28

−14.73

4.41

−3.34*

PAS × recovery

−23.01

8.82

−2.61*

1.94

4.31

0.45

Condition × PAS × preparation

−5.45

8.11

−0.67

2.35

6.08

0.39

Condition × PAS × speech

−1.97

7.51

−0.26

10.10

5.47

1.85

Condition × PAS × recovery

14.28

10.97

1.30

−2.74

5.36

−0.51

BMI body mass index, PAS perceived available support, BL baseline

*p < 0.05

aUnstandardized partial regression coefficients

Discussion

The aim of this study was to examine the conjoined effect of a social support manipulation and naturalistic social support on various cardiovascular measures during an acute aversive encounter. We were able to observe interactive effects between the social support measures. In particular, there was a significant interaction of perceived available support and enacted support on both HR and HRV. This finding documented that perceived available support was accompanied by attenuated cardiac activation during the experiment in the enacted-support condition, but not in the no-support condition. That is, individuals high in perceived available support exhibited lower HR and elevated HRV when they were accompanied by a close friend or partner during the baseline period as compared to when they engaged in a neutral conversation. This finding is entirely consistent with previous studies [37, 38], suggesting that individuals with high levels of naturalistic support may be more receptive, and thus, more likely to benefit from enacted support. For example, Fontana et al. [37] found that participants who were accompanied by their best friend during a laboratory task and who were more satisfied with the support that they generally received in real life exhibited lower HR as compared to participants who were not satisfied with their social support. Our findings may suggest that because of a history of previous experiences of social support, the mere presence of an intimate person at the beginning of the experiment could signify implicit support and reduce threat appraisals in light of a potentially insecure laboratory situation. It should be noted that we were not able to pursue this research question in more detail because we did not directly assess prior history of social interactions. However, it is reasonable to assume that individuals who report rather high levels of naturalistic support (i.e., those who score high on perceived available support) also had more favorable social interactions in the past. To examine this hypothesis further, we would recommend assessing prior history of social interactions in more detail in future studies. Nonetheless, our results are in accordance with the assumption that the presence of intimate others in the laboratory is accompanied by safety signals, which could trigger the social engagement system, especially in individuals with high levels of naturalistic social support [33], leading to elevated HRV and lower HR.

Taken together, our results suggest that enacted social support is especially effective in individuals with high levels of naturalistic social support, whereas individuals who are generally not supported in daily life may not benefit from this situation. It should be mentioned, though, that further research on the interaction of perceived and received support seems warranted. As Uchino [24] suggests, it may also be possible that individuals who generally perceive high levels of support may not benefit from enacted support because it could threaten their co-developed sense of esteem or control. Similarly, O’Donovan and Hughes [69] found that HR was substantially elevated among individuals with high levels of perceived network support when functional support in the laboratory was available. The authors argued that individuals with high levels of naturalistic available support could be “more sensitive to impending stress when they are primed by the notification that help may be required” (O’Donovan and Hughes, [69], p. 1153) during stressor performance. Our data do not allow a more thorough analysis of this effect because enacted social support was restricted to the baseline period of the task. Hence, it would be interesting to manipulate social support during different phases of a psychological challenge (anticipation, task performance) and examine the interplay of perceived control and social support on cardiovascular variables.

It should be emphasized that enacted social support had no impact on cardiovascular reactivity during the later periods of the task. There was no significant relationship between condition and cardiovascular reactivity for preparation, speech, and recovery. This result seems to contradict previous studies that found attenuated reactivity for conditions in which social support was manipulated to be present [e.g., meta-analytic findings; 17]. It should be noted, however, that in most of these studies, the support provider was present during task performance, whereas in our study, social support was provided only at the beginning of the experiment during the baseline period. Our preliminary analyses of the data revealed that participants in the enacted-support condition showed significantly lower HR throughout the course of the experiment as compared to participants in the no-support condition, thus suggesting that the stress-dampening effect of social support prevailed over the course of the task. In our view, this finding is compatible with previous studies on attenuated reactivity during social support because both lines of evidence suggest that enacted social support dampens cardiovascular activation when it is actually provided. Moreover, in addition to previous studies, our results suggest that social support effects may endure for a considerable time, thus leading to sustained cardiac hypoactivation. Further research is certainly needed to examine the time line of the effects of social support manipulations.

Another finding was particularly interesting. We observed that participants with high levels of perceived available support showed larger HR increases, larger HRV decreases, and larger BRS decreases during the speech task, documenting a more pronounced cardiovascular reactivity to a social-evaluative stressor. Of note, this response pattern was independent of enacted social support. It has been suggested that naturalistically supported individuals may exhibit increased cardiovascular activation when encountering stressful events on their own because it may be that they usually consult members of their social networks to help them cope with these situations [19]. Hence, the elevated cardiac reactivity in naturalistically supported individuals may reflect a coping deficit, thus possibly putting these individuals at risk for cardiovascular diseases, as proposed by the reactivity hypothesis [e.g., 70, 71]. Although this view may reflect a plausible interpretation of the data, caution is warranted because other explanations of this effect exist. Importantly, most of the evidence linking cardiovascular reactivity to disease stems from studies focusing on blood pressure, whereas for cardiac reactivity (e.g., HR, HRV), the evidence is less clear [72]. Because we could not find a significant positive association between perceived available support and MAP during the speech (in fact, the relationship was rather negative!), we suggest that our findings do not necessarily mean increased risk for cardiovascular diseases in individuals with comparably high levels of naturalistic support.

On the contrary, individuals with comparably high levels of naturalistic social support showed better MAP recovery after the speech task was terminated despite showing elevated cardiac peak reactivity. Of note, delayed blood pressure recovery has been discussed to constitute a more profound risk factor for cardiovascular diseases than peak reactivity to a challenge [39, 40]. This assumption is entirely consistent with the theory of allostatic load [27], which suggests that a failure to return to baseline levels after termination of a stressor increases allostatic load, which, in turn, puts individuals at risk for cardiovascular diseases. Conversely, it has been suggested that peak reactivity to a challenge together with rapid recovery could indicate physiological toughness, which is related to more effective coping and better health. For example, Schwerdtfeger et al. [26] found that self-efficacy in teachers was positively related to HR and negatively related to HRV and cortisol secretion, probably reflecting positive performance, emotional stability, and, ultimately, better health. Furthermore, there is evidence to suggest that HR reactivity to a laboratory challenge is inversely related to carotid intima-media thickness, indicating lower risk for atherosclerosis in high-HR-reactive individuals [e.g., 73, 74]. Even more, recent research has suggested that elevated cardiovascular stress reactivity is related to better self-reported health [75, 76] and could increase immunocompetence [77]. Thus, several lines of evidence converge to suggest that elevated cardiac reactivity together with better blood pressure recovery as found for naturalistically supported individuals may not indicate increased disease risk, but rather an adaptive response leading ultimately to better health.

Taken together, the findings of this study support the assumption that naturalistic social support is separable from received support and may be interpreted in terms of a resource variable, reflecting adaptive coping and challenge instead of threat appraisals in light of self-relevant stressors [24]. The findings are entirely consistent with those of a former study [23] that found that naturalistic social support was related to elevated cardiovascular reactivity, but also to faster recovery after the stressor was terminated, implying that naturalistic social support may be related to favorable health outcomes.

Strengths and Limitations

This study has several strengths that need to be outlined. First, this study is among the few that have examined the conjoint effects of naturalistic support and enacted support during a laboratory challenge. Consistent with previous research [37], we found that enacted social support was accompanied by attenuated cardiac activation in participants with high levels of naturalistic social support, and that naturalistic support per se was associated with elevated reactivity, but also with better recovery after stress [23]. Importantly, participants were comparable with respect to several relevant personality and lifestyle measures across the social support manipulation conditions, thus assuring internal validity. Further, we applied the most recent methodological data analysis methods, namely mixed-effects models, which are advantageous over traditional ANOVA approaches. Finally, in recording a wider range of cardiovascular variables, we were able to show that social support was associated with various cardiovascular measures during baseline, task performance, and recovery.

On the other hand, there are also some caveats that need to be discussed. First, participants were not randomly assigned to the support and neutral conditions, but rather assigned themselves to one condition to assure ecological validity in this laboratory setting. We are aware that this approach precludes causal interpretations. However, various demographic, lifestyle, and personality measures suggested comparability between conditions, although there may have been significant differences with respect to other nonassessed dimensions (e.g., extraversion, neuroticism). In order to verify a causal pathway, a randomized assignment should be used in future studies, although this could challenge ecological validity. Second, although the ratings of the interaction quality suggested that the manipulation of enacted social support was quite successful, it should be noted that the effect size was only moderate. It turned out that the quality of interactions in both conditions was rated as rather high. We would have expected a more pronounced difference between conditions because the experimenter was instructed to behave neutrally in the no-support condition, following a standardized protocol. Of note, relationship positivity in experimental settings has been associated with attenuated blood pressure reactivity [45]. Thus, it might be speculated that our findings are lower-bound estimates of the effects of social support on cardiovascular reactivity that might increase when the no-support condition is perceived as less positive. It may have also been the case that participants in the no-support condition wanted to please the experimenter, which may have resulted in more positive evaluations.

Relatedly, during the no-support condition, there was a short period of interaction with the experimenter, which may have helped participants to better adapt to the experimental context. This may have countered the intended effect, resulting in a dampening of cardiovascular reactivity in this group. However, our study protocol aimed to minimize interactions with the experimenter in the course of the task by using computerized instructions. Nonetheless, in order to prevent familiarization with the experimenter, future studies could employ a trained confederate to maximize the variance between conditions.

Third, it should be noted that the nature of the accompanying persons in the enacted-support condition was not homogeneous. A minority of participants was accompanied by fellow students who may have differed in their ability to support the participants. This may have resulted in a blurring of effects. Hence, we would recommend restricting the nature of the accompanying persons in future studies to enhance statistical power.

Fourth, it turned out that perceived available support was generally high in this sample, with a comparably small standard deviation. That is, the present study is not able to answer the question of whether truly low levels of perceived available support (i.e., social isolation) are also related to reactivity of the cardiovascular system. However, previous research has suggested that loneliness constitutes a robust risk factor for cardiovascular diseases [7882], which is compatible with our finding of less effective coping in individuals with low levels of perceived available support. It should also be noted that a restriction of range with respect to perceived available support may have underestimated the effects reported here. Nonetheless, for future studies, it seems important to better contrast social support manipulations and to recruit more participants with truly low levels of naturalistic social support in order to examine the generalizability of our findings.

Finally, the sample size was rather moderate, thus possibly compromising the robustness of the effects. However, we want to emphasize that we applied mixed-effects modeling, which has better power to detect effects because it involves a model for the error variance [66]. Hence, in our view, the findings seem trustworthy, although larger sample sizes would certainly be desirable. Similarly, mainly women participated in this research, thus precluding generalizations to men. We tried to recruit more males by explicitly asking men to participate in this study. However, they were not as likely to participate as females. This is not an uncommon phenomenon in psychological studies though. Hence, we could not examine sex differences with respect to social support in more detail because of the comparably low power of the design. It should be noted that previous research has often found sex differences with respect to cardiovascular consequences of social support [44, 83], thus necessitating the study of gender effects in future studies.

Conclusions

Notwithstanding these limitations, the present study suggests that social support has reliable effects on the cardiovascular system. Whereas enacted social support seems to have dampening effects on the activation of the cardiovascular system, particularly in individuals with high levels of naturalistic social support, these individuals also seem to show elevated reactivity to a challenge and better recovery, possibly reflecting more effective coping with stressors. Hence, naturalistic social support seems to constitute a psychosocial resource variable that is associated with better health.

Conflict of interest statement

The authors have no conflict of interest to disclose.

Copyright information

© The Society of Behavioral Medicine 2011