Home Current Issue Previous Issues Published Ahead-of-Print Collections Podcasts For Authors Journal Info
Skip Navigation LinksHome > May/June 2001 - Volume 63 - Issue 3 > Partner Interactions Are Associated With Reduced Blood Press...
Psychosomatic Medicine:
Original Articles

Partner Interactions Are Associated With Reduced Blood Pressure in the Natural Environment: Ambulatory Monitoring Evidence From a Healthy, Multiethnic Adult Sample

Gump, Brooks B. PhD, MPH; Polk, Deborah E. PhD; Kamarck, Thomas W. PhD, and; Shiffman, Saul M. PhD

Free Access
Article Outline
Collapse Box

Author Information

From the Department of Psychology (B.B.G.), State University of New York at Oswego, Oswego, New York; and the Department of Psychology (D.E.P, T.W.K, S.M.S), University of Pittsburgh, Pittsburgh, Pennsylvania.

Address reprint requests to: Brooks B. Gump, PhD, Department of Psychology, SUNY Oswego, Oswego, NY 13126. Email: gump@oswego.edu

Received for publication June 8, 2000; revision received October 18, 2000.

Collapse Box


Objective: The objective of this study was to examine the effects of partner interactions on ambulatory blood pressure in a sample of 120 healthy adults who were monitored over a 6-day period.

Methods: After each blood pressure measurement, participants rated characteristics of ongoing social interactions, along with emotional activation, physical activity, talking, posture, and other covariates, with computer-assisted self-report assessments.

Results: Using multilevel modeling, we showed that blood pressure was significantly lower during social interactions with one’s partner relative to social interactions with any other person and relative to periods of not interacting. Interactions with partners also were characterized by significantly less talking and emotional activation and more intimacy and perceived emotional support; however, these differences did not mediate the partner effect on blood pressure. In addition, the relative benefits of interacting with a partner were not moderated by relationship quality, gender, or race.

Conclusions: The effects of social situations on ambulatory blood pressure may represent one pathway through which social relationships affect cardiovascular health.

Back to Top | Article Outline


DABS = Diary of Ambulatory Behavorial States;, DAS = Dyadic Adjustment Scale;, DBP = diastolic blood pressure;, SBP = systolic blood pressure.

Social contact has been shown to be an important correlate of cardiovascular health. For example, a growing number of prospective studies have demonstrated that socially isolated individuals, those with few close social ties, are at increased risk for cardiovascular and all-cause mortality (1–3). Marital ties may be especially health-promoting. For example, cardiovascular disease mortality for the married is nearly half that found for the unmarried or divorced (4), and the presence of marital and other close ties has been associated with improved prognosis in the setting of cardiovascular disease (5). The potential mechanisms accounting for the association between social relationships and cardiovascular risk have not been established, but some existing data suggest that these effects may be independent of socioeconomic status and established risk factors such as smoking, high blood pressure, and cholesterol (6, 7).

One relatively unexplored mechanism that may explain the salutary effects of social contact involves the potential effects of close social ties on autonomic functioning. Prolonged or exaggerated sympathetic nervous system activation has been implicated in a number of the pathophysiological processes that may set the stage for cardiovascular disease (8–11) and that may trigger cardiac events in predisposed individuals (12). Alterations in the social environment, in turn, may have important effects on sympathetic nervous system functioning with implications for cardiovascular health. For example, among socially dominant cynomolgus macaques, prolonged contact with strangers (periodic assignment to new social groups) has been shown to exacerbate atherosclerosis, an effect that is eliminated after pharmacologic sympathetic blockade (13). At the same time, macaques housed in single cages (an analog to social isolation in humans) have been shown to develop higher rates of atherosclerosis (14). These animals have higher heart rates during the evening hours than their group-housed counterparts despite lower levels of physical activity (15), suggesting possible alterations in autonomic control. Together these results support the possibility that the effects of social contact on cardiovascular risk may be explained in part by alterations in autonomic functioning. On the basis of these findings, we would expect beneficial effects to be most apparent when considering the effects of close social ties.

In humans, ambulatory blood pressure monitoring may be seen as a marker of ongoing sympathetic nervous system activity during daily life. A growing number of investigations have begun to examine psychosocial influences on ambulatory cardiovascular activity (16–18). Although episodes of social interactions have been recorded in a number of these studies (eg, Refs. 19 and 20), only two investigations in the extant literature have reported social interaction effects on cardiovascular activity in the context of specific role relationships. In a sample of normotensive and mildly to moderately hypertensive subjects who were monitored during their daily activities, Spitzer et al. (21) found that the presence of family members was associated with lower sitting SBP and DBP when compared with the presence of strangers or friends. They reported that the family-stranger difference in seated SBP for normotensives was 6 mm Hg, whereas seated SBP for hypertensives when they were with strangers was 21 mm Hg higher compared with readings taken when they were with friends. One of the limitations of this study was its use of an aggregated within-person analysis (see ref. 22), an analytic strategy that may introduce considerable error when used with daily monitoring data.

A more recent study, using a more sophisticated methodology (23), examined ambulatory blood pressure among New York City traffic agents during the working day, comparing cardiovascular activity during periods of talking with the public, talking to a supervisor, and talking with coworkers. Two control conditions were also examined in this study: resting baseline periods and control periods during which the agents were not interacting with others (“looking for violations”). Using multilevel modeling to adjust for the effects of posture, location, and the time spent within each social situation during the day, these investigators found that social interactions (especially talking with the public) were generally associated with higher blood pressure and heart rate compared with the control and baseline conditions, whereas talking with a coworker generally produced levels of blood pressure and heart rate that were intermediate compared with those in the other conditions. One of the interesting features of this study was its use of mediational models to examine the effects of affective changes (positive and negative mood) in accounting for the observed effects. Although social interaction type affected mood (talking with the public was associated with higher levels of negative affect) and mood showed effects on cardiovascular activity (lower levels of mood were associated with higher blood pressure), these influences did not account for or mediate the observed cardiovascular impact of social interaction.

In summary, both of the available studies have shown that social interaction may be associated with changes in cardiovascular activity during daily life, with such changes conditioned on the nature of the role relationships involved in the interaction. In both samples, interactions with strangers were associated with the largest increases in cardiovascular activity. Neither study demonstrated significant decreases in cardiovascular activity during social interactions relative to control conditions.

Unfortunately, limitations in both of these studies preclude our drawing broad conclusions about the effects of social interactions on ambulatory cardiovascular activity in a manner that might shed light on the autonomic effects of social contact during daily life. The Spitzer et al. (21) report was published before the widespread use of multilevel modeling in the ambulatory monitoring literature; because of the multiple influences that may affect cardiovascular activity in the natural environment, the use of such methods should permit more sensitive statistical control. The Brondolo et al. study (23), which used a sensitive statistical design, was nevertheless focused on workday experiences only, so it did not permit examination of family influences on cardiovascular activity.

In addition to the limitations described above, several issues remain to be addressed. Neither of the existing reports examined spousal influences on ambulatory cardiovascular activity. Given the salutary effect of marital relationships (4), one might anticipate unique effects of this type of social tie. Similarly, there are no existing data on the moderating effects of relationship quality or the quality of the interaction on ambulatory cardiovascular measures. Given the laboratory findings showing large blood pressure responses to social conflict within the marital dyad (24), one might expect that the quality of social interactions involving the spouse would vary in terms of their effects on acute cardiovascular responses during the day. Finally, neither of the existing reports have resolved the question about the mediating processes (eg, talking) accounting for social role effects.

With these concerns in mind, we designed the current study to examine the effects of social interaction on ambulatory cardiovascular activity in a community sample, with a special focus on interaction with a significant other and with attention to issues of relationship and interaction quality as well as appropriate focus on behavioral (ie, talking) and affective (ie, mood) mediators of these effects. To the extent that autonomic influences and ambulatory cardiovascular activity may play a part in contributing to cardiovascular risk, the current findings are expected to shed some light on some of the possible mechanisms accounting for the beneficial health effects of close social ties.

Back to Top | Article Outline



Participants (N = 120), ages 23 to 50 years (mean = 35) were recruited from the Pittsburgh metropolitan area by letter, poster, and newspaper advertisements. To permit robust race and gender comparisons, participants were recruited in roughly equal numbers from each of four gender and ethnic groups: African American women (N = 34), African American men (N = 26), white women (N = 30), and white men (N = 30). Only married individuals or those living with a significant other (3 months or more) were recruited to ensure a broad sampling of significant interpersonal interactions over the course of the monitoring period. Additional eligibility criteria included absence of chronic disease by self-report (and screening blood pressure assessments in the normal range: SBP <145 mm Hg and DBP <90 mm Hg), nonsmoking status, and no use of medication with autonomic or cardiovascular effects. Each participant who completed the study successfully was paid $250. Other publications based on this sample (17, 25, 26) did not include a focus on the effects of social interactions on cardiovascular activity.

Back to Top | Article Outline

The study design involved a training period and a “shakedown” day with feedback (providing practice and habituation to the measurement procedures). This was followed by a 6-day monitoring period for each participant. During the monitoring period, each participant wore an automated ambulatory blood pressure monitor and carried a handheld computer throughout each day. Blood pressure assessments were taken every 45 minutes; participants were instructed to complete an entry in the electronic diary (27) in response to each inflation of the cuff. Participants were instructed to begin electronic diary assessments shortly after awakening in the morning and to continue the monitoring procedures until they were ready for sleep in the evening. Participants were trained to put the electronic diary “to sleep” at night and to instrument the ambulatory monitor at the beginning of the day to facilitate complete data collection during the monitoring period.

The 6-day monitoring period was divided into two parts (a 2-day interval and a 4-day interval) with a 1-day break in between. After the 6 days of data collection, participants returned to the laboratory for payment and debriefing. All participants started the 6-day monitoring protocol on a Thursday or Friday morning, a schedule that ensured that the monitoring period included workdays as well as 1 weekend day. Although the blood pressure monitor was worn during the first 2 nights of the monitoring period, only the wake-time data during the 6 monitoring days are included in this report. Additional details about the methods used in this study can be found in the report of Kamarck et al. (25).

Back to Top | Article Outline
Psychosocial Measures
Diary of Ambulatory Behavioral States.

The electronic diary was used in this study as a platform for administering the Diary of Ambulatory Behavioral States (17, 25), a 58-item questionnaire designed for repeated real-time assessment of behavioral influences on cardiovascular activity in the natural environment. The close-ended questionnaire yields 11 multi-item subscales used to assess psychosocial characteristics (eg, mood and social interaction attributes) and a number of single-item subscales used to assess metabolic or postural characteristics (eg, posture, activity, and substance use) or other objective features of the environment (eg, location and number of people in the setting). For the purpose of this report, we used the DABS for assessing ongoing social interactions, for measuring relevant covariates, and for assessing potential mediators of any observed social interaction effects. Each set of variables is described below. DABS items not specifically relevant to this report are discussed in their entirety in other publications (17, 25).

Back to Top | Article Outline
Social interactions.

For the purposes of this study, a social interaction was defined as a “give-and-take exchange with others which may or may not involve conversation”(28). Four items from the DABS were used to assess social interaction characteristics for the present report. First, participants were asked 1) whether they were currently interacting (yes, no) or 2) were involved in an interaction during the past 10 minutes (yes, no). If participants answered yes to either item, they were administered a series of additional questions, including 3) “Whom did you interact with?,” a question to which they were provided with three response options (partner and others, partner alone, or others), and 4) “How many people did you interact with?,” to which they could respond 1 other, 2 others, 3 others, or 4+ others. To reduce confounding influences, only interactions with 1 other were examined for this report. Combining the information provided from these four questions allowed us to select three sets of observations for each participant: 1) observations involving no interactions, 2) observations involving interactions with partner only, and 3) observations involving interactions with one other individual. It should be noted that observations involving social interactions with more than one person were irrelevant for this report and were excluded from our analyses.

Back to Top | Article Outline

As indicated above, a number of additional determinants of cardiovascular activity were assessed using the DABS. Because a number of these were potentially correlated with the conditions of social interaction, the following items were included as covariates: 1)“Describe physical activity” (inactive, some movement, moderate, strenuous), 2) “Your posture?” (on your feet, sitting, lying down), 3) “Temperature comfort?” (comfortable, too cold, too hot). Each item was coded using categorical (dummy) variables to model possible nonlinear effects. Additional covariates included 4) “Consumed alcohol?” (no, yes), 5) “Consumed caffeine?” (no, yes), and 6) “Consumed a meal?” (no, yes).

Location is another potential confound in this study. Ambulatory blood pressure is thought to be higher at work than at home (29); at the same time, a majority of partner interactions typically occur in the home setting, making location a potential confound when considering partner effects. Location was measured with a single item (“Your location?” with responses of home, workplace, vehicle, outside, or other) and coded as a dichotomous variable in the present analyses (0 = home, 1 = not at home).

A seventh covariate used in this report was the DABS item “How many people in view?” (alone, 1, 2–3, 4+). Adjusting for this measure controlled for the potential extraneous effects of the presence of others on cardiovascular activation (eg, the effects of crowding), including those not involved in any interactions with the participant at the time of assessment.

Back to Top | Article Outline
Potential mediators.

Eight additional DABS measures were examined as potential mediators of any observed effect of social interactions on cardiovascular activity. DABS items assessing two dimensions of emotional activation (negative affect and arousal) were adopted from the Russell (30) mood circumplex and the Perceived Stress Scale (31). Negative affect items included questions about feeling “sad,” “frustrated/angry,” “nervous/stressed,” and having thoughts “that upset you.” The arousal scale was based on questions about feeling “alert” and “tired” (reversed). Participants answered these questions on four-point scales (NO!!, no??, yes??, YES!!).

If subjects endorsed a recent social interaction, they were administered a number of additional DABS ratings, which yielded the following five subscales: social conflict, positive interaction, intimacy, emotional support, and instrumental support. The social conflict subscale was developed on the basis of existing theories of social stress (32). With reference to a recent or current social interaction, participants were asked whether someone had 1) placed “unfair demands,” 2) had “interrupted or interfered,” 3) had been “judgmental or critical,” 4) had “ignored” the participant, or 5) had “argued” or been “in conflict” with the participant. Each of these questions was answered on a four-point scale (NO!!, no??, yes??, YES!!), and scores were averaged across items (33). The positive interaction scale was based on three items (with four-point scales) assessing whether the other person in a social interaction was “pleasant” (1 = pleasant, 4 = unpleasant), “passive” (1 = passive, 4 = dominant), and “warm” (1 = warm, 4 = cold). The intimacy scale was derived from the Rochester Interaction Record (28) and was based on two items (with four-point scales) assessing the level of intimacy for the interaction (1 = superficial, 4 = meaningful) and whether the outcome of the interaction was important (NO!!, no??, yes??, YES!!). The emotional support scale was based on four items assessing whether someone “expressed confidence in you,” “expressed care/concern for you,” “gave you positive feedback,” and “made you feel important” (scales coded as two-point scale: no, yes). Finally, the instrumental support scale was based on two items assessing whether “someone did you a favor” and whether “someone helped you with errand/task” (scales coded as two-point scale: no, yes).

An eighth variable used as a potential mediator involved whether the participant was talking at the time of cardiovascular assessment. Talking has been found to raise blood pressure levels in laboratory studies (34, 35); thus, it might be expected to contribute to differences in cardiovascular activity as a function of social context. Talking was assessed using a single DABS item (“talking during inflation?,” no, yes).

Back to Top | Article Outline
Relationship quality.

In addition to the diary and ambulatory blood pressure measures, participants were also asked to complete a series of global self-report questionnaires. One of these questionnaires, the Dyadic Adjustment Scale (36), was included as part of this report. The DAS is a widely used 32-item measure of marital or relationship quality with good internal consistency (Cronbach’s α= 0.96; Ref. 37). Scores range from 0 to 151, with higher scores representing lower levels of relationship distress.

Back to Top | Article Outline
Ambulatory blood pressure assessment.

The Accutracker DX (38) ambulatory monitor was used for cardiovascular assessment. This device is a noninvasive auscultatory monitor that is relatively quiet and comfortable for extended wear and that has been shown to accurately track cardiovascular changes during simulated ambulatory assessments involving physical exercise and mental stress (39).

Back to Top | Article Outline
Electronic diary assessment.

We designed a programmed protocol for administering the DABS using the Psion Organiser LZ (Psion, Ltd., London, UK), a handheld computer with a 4-line, 20-character LCD screen (80 characters total), 32 K of memory, a real-time clock/calendar, and an audio speaker (17). For research purposes, the computer organizer was adapted with a number of user-friendly features. For example, participants were required to use only the top row of the keypad for responding and scrolling, and the remainder of the unit was encased in plastic. The organizer was equipped with a static-free carrying case suitable for wearing over the shoulder or on the belt.

Each diary item was displayed on the screen under a brief caption describing the time frame of assessment (eg, “in the last 10 minutes”), and response options were presented in the form of two- to five-point scales. The programmed protocol prompted participants with only relevant probes. For example, participants were administered social interaction questions only if they indicated that they were involved in a recent or current social interaction.

Back to Top | Article Outline
Data Analysis
Data quality.

We obtained a relatively high rate of compliance with the data collection protocol for this study despite the demanding 6-day monitoring schedule. Of the 170 interested and eligible participants, a final sample of 120 participants with complete and useable data produced an average of 109 valid diary and blood pressure observations each (range, 68–155). In addition to sporadic noncompliance, the range of the number of available observations from each participant is accounted for by missed blood pressure readings, occasional equipment failures, and scheduling variations (different wake and sleep times). Additional details about compliance and subject attrition can be found in the report of Kamarck et al. (25).

Back to Top | Article Outline
Outlier and error detection.

The Accutracker DX displays error messages when technical problems may invalidate a blood pressure reading. We deleted all readings associated with weak Korotkoff sounds, microphone difficulties, or air leaks (test code 2, 3, and 4, respectively; Ref. 38). In addition, criteria described by Marler et al. (40) were used to define artifactual values: SBP values <70 mm Hg or >250 mm Hg were deleted, as were DBP values <45 mm Hg or >150 mm Hg. If SBP/DBP was less than [1.065 + (0.00125 × DBP)] or greater than 3, both SBP and DBP were deleted. Heart rate values <40 bpm or >200 bpm were also eliminated. Outlying readings were frequently associated with excessive arm movement during cuff inflation. On average, 7.4% of blood pressure readings for each participant in this study were deemed unusable for one of the reasons cited above.

Back to Top | Article Outline
Analytic models.

Data were analyzed using a multilevel modeling approach (SAS Proc Mixed, 41). This approach has several advantages over conventional regression methods, including its ability to model autocorrelation effects, correct for time-varying covariates, and handle unbalanced designs (22). Repeated measures of cardiovascular activity were regressed on the social interaction variables as well as several time-varying covariates that were centered about the sample mean. A spatial power function (41) was used to model autocorrelated errors for the within-subjects repeated measures. Additionally, the variance of the autocorrelated errors was divided into true and error components (ie, REPEATED/sub = subj type = sp(pow) (time) local). Both the intercept and the time-varying predictor variables were treated as having random coefficients (differing in value for each subject); however, the time-varying covariates were entered as fixed effects to reduce the complexity of the model. The variances and covariances of the intercept and time-varying predictor variables were estimated as a part of the model (G matrix = UN). Maximum likelihood methods were used to obtain a model solution.

Separate models were run for SBP, DBP, and heart rate. We created a categorical variable with three levels to characterize three within-subject conditions as described above: 1) observations involving no social interaction, 2) observations involving interactions with one’s partner, and 3) observations involving interactions with one other person (not one’s partner). For categorical or classification variables in the Proc Mixed analysis, SAS provides both an omnibus F value, testing the significance of the variable across all levels in the context of the model, as well as pairwise orthogonal contrasts, testing the significance of each level of the variable against a reference level, also in the context of the model. To obtain all possible pairwise comparisons, the categorical variable was coded two different ways in two separate models using different reference conditions in each case.

The association between social interactions and cardiovascular activity was assessed in separate models with and without the seven covariates described above. In addition, to better understand any significant findings, we examined mediational models using each of the eight hypothesized mediating variables as appropriate. A final set of models examined the role of relationship quality, gender, and race as moderators of the effects of social interactions on cardiovascular activity.

Back to Top | Article Outline


Number of Observations

In the data set as a whole, each participant produced an average of 109 observations over the course of the 6-day monitoring period, and reports of social interactions occurred during an average of 75 (69%) of these observations. We eliminated data from three subjects who reported no partner interactions; from the remainder of the sample, we eliminated all observations involving social interactions involving more than one other individual (29 on average across the sample). The remaining data set (N = 117) included an average of 77 valid observations per person, including 16 partner interactions (SD = 12.28), 30 interactions with one other (SD = 13.11), and 31 observations involving no interactions (SD = 14.49).

Back to Top | Article Outline
Covariate Selection

Preliminary random-effects regression models examined the influence of each of the seven time-varying covariates as determinants of SBP, DBP, and heart rate during the monitoring interval. Results suggested that all of the hypothesized covariates were significant in one or more models as predictors of one or more measures of cardiovascular activity, as depicted in Table 1. These results are similar to those reported previously (17), but they differ with respect to the number of covariates and the number of observations included in each model.

Table 1
Table 1
Image Tools
Back to Top | Article Outline
Relationship Quality: Effects on Social Situation

Relationship quality was examined as a correlate of social interaction across different social settings. We found that relationship quality was positively associated with the proportion of interactions with one other person that were with one’s partner (r = 0.28, p < .002) and negatively associated with the proportion of interactions with one other person that were with another person (r = −0.28, p < .002). 1 To illustrate these significant associations, Figure 1 shows the proportion of participants’ total observations occurring with one other person (eg, number of observations with the partner/total observations with one other person) as a function of relationship quality. For this figure, the proportion of observations is presented for participants in the highest and lowest quartiles of relationship quality.

Fig. 1
Fig. 1
Image Tools
Back to Top | Article Outline
Social Interactions
Cardiovascular activation.

Social interaction characteristics were significantly associated with SBP and DBP on a within-person basis (omnibus F (2,232) values = 9.02 and 31.64 and p values < .01, respectively). These effects remained in multilevel models that included our seven covariates (omnibus F (2,232) values = 7.82 and 5.38 and p values < 0.01 for SBP and DBP, respectively). In addition, a significant effect of social interactions on heart rate emerged in the covariate-adjusted models (F (2,232) = 5.03, p < .01). The patterns of the results (rank ordering of the conditions) remained the same with and without the covariate adjustments. As a result, all remaining analyses are reported with the covariate adjustments.

Pairwise comparisons revealed that periods of partner interaction were associated with significantly lower SBP relative to periods of interacting with one other person (p < .01) or when not interacting (p < .01; see Table 2). SBP during interactions with one other person (not one’s partner) did not differ significantly from SBP during periods of not interacting (p > .10).

Table 2
Table 2
Image Tools

DBP during periods of partner interaction and not interacting did not differ significantly (p > .10); however, both situations were associated with significantly lower DBP relative to periods of interacting with one other person (p values < .01; see Table 2).

Interactions with one other person (not one’s partner) were associated with significantly lower heart rates relative to periods of not interacting (p < .05), but neither differed significantly from periods of partner interaction (p values > .10, see Table 2).

Using the intercept and regression coefficients from Table 2, we calculated mean blood pressure for each condition (ie, interacting with one’s partner, interacting with another person, and not interacting). These means are reported in Figure 2 along with their appropriate standard errors.

Fig. 2
Fig. 2
Image Tools
Back to Top | Article Outline
Behaviors and mood.

Participants talked more during social interactions with one other person than during social interactions with a partner (t (115) = 2.47, p < .05), and talking was greater in both of these conditions relative to when not interacting (t (115) = 33.04, p < .0001 and t (115) = 27.06, p < .0001, respectively). Using the intercept and regression coefficient for social situations, the means for these conditions were calculated and are reported in Table 3. In addition, participants’ arousal was significantly higher during interactions with one other person relative to periods of interacting with a partner (t (116) = 8.56, p = .0003) or not interacting (t (116) = 10.45, p < .0001). Negative affect was significantly lower when not interacting relative to interactions with a partner (t (115) = 2.47, p < .05) and interactions with one other person (t (115) = 5.34, p < .01). Finally, interactions with a partner were perceived as having significantly more intimacy (t (116) = 8.21, p < .0001) and emotional support (t (116) = 6.27, p < .0001) relative to interactions with one other person.

Table 3
Table 3
Image Tools

Other perceptions of the social interaction, including social conflict, positive interaction, and instrumental support, did not differ as a function of the social situation (p values > .10).

Back to Top | Article Outline
Mediational analyses.

The results reported above indicate lower blood pressure during partner interactions relative to interactions with others or during periods of not interacting. In addition, partner interactions were associated with less talking, less arousal, more intimacy, and more emotional support relative to interactions with one other person. These behaviors and moods might mediate the effect of partner interactions on cardiovascular activity. In other words, the effects of partner interactions on cardiovascular activity might be explained by their impact on behavior and moods.

Talking is one potential mediator. Talking occurred during 58% of the social interactions measured here (on average). When participants’ blood pressure was regressed simultaneously on the three-level social interaction variable and the measure of talking (0, 1), the pattern of social interaction effects remained unchanged for SBP and heart rate, suggesting that neither of these effects was mediated by talking. In models predicting DBP, however, the inclusion of talking in the model did alter the pattern of social interaction effects. Although DBP during interactions with one other person (not one’s partner) was significantly higher than during periods of not interacting when talking was not controlled, this effect was no longer significant after the addition of talking (t (116) = 0.34, p = .74). This suggests that DBP was higher when interacting with one other person relative to not interacting specifically because of differences in talking (ie, talking was the mediator; see Ref. 41). DBP during partner interactions did not differ significantly from DBP during periods of not interacting when talking was not controlled (p > .50). However, after controlling for talking, DBP was significantly lower during partner interactions relative to periods of not interacting (t (116) = 2.05, p < .05). This suggests that the act of talking may serve to suppress or obscure an otherwise significant relationship between partner interaction and DBP reduction. Finally, DBP was significantly higher when interacting with one other person relative to interacting with one’s partner regardless of whether talking was entered as a covariate (t (116) = 2.38 without talking and t (116) = 2.29 with talking, p values < .05). Talking may increase DBP, then, and this effect may be especially strong among nonpartners or less familiar individuals; at the same time, when talking is held constant (ie, in the absence of talking), being around one’s partner seems to be associated with small reductions in DBP.

Given that arousal, intimacy, and emotional support were each associated with social situation and may be associated with cardiovascular activity (17), we examined whether any of these factors mediated the relationship between social situation and blood pressure response. Controlling for these variables in the analysis of social situation (using partner vs. nonpartner vs. not interacting for arousal and a dummy variable comparing partner vs. nonpartner for the measures of intimacy and emotional support), however, did not significantly diminish the association between social situation and blood pressure. Thus, the criteria for mediation were not met (42). In sum, despite an extensive assessment of behavioral and emotional states during social interactions, there was no empirical support for any of these variables mediating the observed reduction in blood pressure during interactions with a partner.

Back to Top | Article Outline
Moderators of social interaction effects.

The analysis of social interactions suggests a unique cardiovascular benefit resulting from partner interactions. We next addressed the question of whether these cardiovascular benefits of partner interactions might vary as a function of relationship quality, gender, and race. We entered main effects first, then potential two-way interactions (eg, social interaction by relationship quality), and then potential three-way interactions (eg, social interactions by relationship quality by gender). Of the 36 possible two-way and three-way interactions, we found only three significant three-way interactions. 2 Furthermore, these three-way interactions did not contain the variables for social situations involving the partner and were not readily interpretable. Therefore, these significant higher-order interactions were attributed to chance.

Back to Top | Article Outline


Social interaction with one’s partner is associated with reduced ambulatory blood pressure relative to the social interactions with other individuals. This effect was observed for both SBP and DBP. Although interactions with intimate partners are also marked by different affective states (more intimacy and support), these psychosocial qualities did not seem to mediate the effects on blood pressure. Furthermore, although one could imagine that in unsatisfying relationships partner interaction would have negative or stressful effects on mood and pressor effects on blood pressure, we found no evidence of this: Relationship quality did not moderate the effect of the partner’s presence.

Periods of interacting were significantly more likely to involve talking relative to periods of not interacting. Talking was associated only with significantly higher DBP. Therefore, controlling for talking did not affect the partner interaction effects on SBP. However, controlling for talking did result in DBP during partner interactions becoming significantly lower during periods of not interacting and interactions with one other person no longer differing from periods of not interacting. Finally, heart rate was significantly lower when interacting with one other person relative to not interacting, with heart rate during partner interactions falling between these two conditions and not differing significantly from either. This pattern of results may reflect greater attentional demands during social interactions with one other person relative to not interacting and corresponding effects of phasic attentional processes on heart rate (43). It should be noted that heart rate results appeared only in models that involved extensive covariate adjustments.

The mechanism for blood pressure reductions during partner interactions remains unclear. Potentially important covariates (eg, location, physical activity, and food or drink consumption) were controlled and therefore could not account for the observed effects of partner interaction. It is interesting that Brondolo et al. (23) also tested a number of potential mediators (eg, apprehensiveness and anxiety) and failed to find evidence of mediation. Therefore, it is possible that these effects are mediated by determinants outside awareness. For example, given that most interactions with a well-established partner are safe or predictable, a partner’s presence may act as a classically conditioned safety signal (cf. Ref. 44). Nonpartner interactions, however, because they may occur less frequently and involve greater uncertainty, may be more likely to be associated with a defense reaction (45) or heightened vigilance for threat (46), resulting in heightened sympathetic activation during daily life. There is increasing speculation that classically conditioned fear or defense responses may be activated subcortically (47), an effect that would be consistent with our inability to detect mediators with use of self-report measures.

Previous research has demonstrated the greatest cardiovascular and immunological reactivity to conflict among those reporting the highest levels of relationship quality (24, 48). Therefore, our evidence suggesting cardiovascular benefits of partner presence regardless of relationship quality seems contradictory. However, to the extent that participants in distressing relationships have significantly fewer social interactions with their partners, as demonstrated in the present study, they may avoid potentially adverse emotional and cardiovascular consequences of such interactions, thereby precluding the possibility of moderation. In other words, relationship quality may moderate the effect of partner interaction on blood pressure only when artificial constraints produce a sufficient number of partner interactions for those in distressing relationships (see laboratory-based studies, eg, Refs. 24 and 48).

Although the acute reduction of blood pressure during partner interactions was significant, this reduction was not large (on the magnitude of 1 to 1.5 mm Hg during each interaction). This effect is comparable to the effects of caffeine consumption and location during assessment, but not nearly as large as changes due to posture and physical activity. However, at a population level, such minor reductions in blood pressure, if sustained, could have substantial health effects. For example, it is estimated that 45.8 coronary heart disease events per 100,000 population would be prevented if DBP in the range of 70 to 79 mm Hg was reduced by an average of only 2 mm Hg in a previously untreated population (49). More research is necessary to determine whether these types of effects have any clinical significance in the prediction of cardiovascular disease.

In summary, the results of the current study suggest that social interactions with a partner or spouse have a beneficial effect on cardiovascular activity. Furthermore, those reporting less satisfying relationships also engaged in a smaller proportion of these potentially beneficial partner interactions. Because the present sample included only participants who were married or involved in a significant relationship, a discussion of possible mediators of the effects of marital status on ambulatory blood pressure is clearly speculative at this point. Nevertheless, an intriguing possibility that awaits additional research is that the effects of partner interactions on cardiovascular activity may partially account for the association between marital status and cardiovascular health. Speculations aside, these findings illustrate the potential utility of exploring the effects of social networks on health through examining autonomic influences of patterns of interaction in the natural environment.

This research was supported by National Institutes of Health Grants HL49410 and HL07560. Saul Shiffman is a founder of and principal in invivodata, Inc., which provides electronic diary technology for behavioral and medical research. We thank Harry Reis for his assistance in the development of the measures, and we acknowledge the following individuals for their assistance with the project: Leslie Smithline, Jeffrey L. Goodie, Hayley Thompson, Joey Yi-Kuan Jong, Joseph Schwartz, David Stoffer, Silviu Bacanu, Jianqing (James) Jin, Jean A. Paty, Maryann Gnys, Jon Kassel, Walter Perz, and Verne Pro.

Back to Top | Article Outline


Correlations of the DAS with the proportion of partner interactions out of the total number of observations was nearly identical; however, the correlation of the DAS with the proportion of other interactions out of the total number of observations was no longer significant. Cited Here...

All analyses were repeated using a clinical cutoff for relationship quality (cf, Ref. 36) and treating relationship quality as a dichotomous variable with participants considered low in relationship quality if their DAS scores fell below 100 (N = 31) and high in relationship quality if their DAS scores were 100 or greater (N = 89). In all cases, the results were nearly identical. Cited Here...

Back to Top | Article Outline


1. Kaplan GA, Cohn BA, Cohen RD, Guralnik J. The decline in ischemic heart disease mortality: prospective evidence from the Alameda County study. Am J Epidemiol 1988; 127: 1121–42.

2. Kaplan GA, Salonen JT, Cohen RD, Brand RJ, Syme SL, Puska P. Social connections and mortality from all causes and from cardiovascular disease: prospective evidence from eastern Finland. Am J Epidemiol 1988; 128: 370–80.

3. Orth-Gomer K, Johnson JV. Social network interaction and mortality: a six year follow-up study of a random sample of the Swedish population. J Chronic Dis 1987; 40: 949–57.

4. Kitagawa EM, Hauser PM. Differential mortality in the United States: a study in socio-economic epidemiology. Cambridge (MA): Harvard University Press; 1973.

5. Williams RB, Barefoot JC, Califf RM, Haney TL, Saunders WB, Pryor DB, Hlatky MA, Siegler IC, Mark DB. Prognostic importance of social and economic resources among medically treated patients with angiographically documented coronary artery disease. JAMA 1992; 267: 520–4.

6. House JS, Landis KR, Umberson D. Social relationships and health. Science 1988; 241: 540–4.

7. Berkman LF, Breslaw L. Social networks and mortality risk.In: Berkman LF, Breslaw L, editors. Health and ways of living: the Alameda County study. New York: Oxford University Press; 1983.p. 113–60.

8. Gordon D, Guyton JR, Karnovsky MJ. Intimal alterations in rat aorta induced by stressful stimuli. Lab Invest 1981; 45: 14–27.

9. Kaplan JR, Petterson K, Manuck SB, Olsson G. The role of sympathoadrenal medullary activation in the initiation and progression of atherosclerosis. Circulation 1991;(Suppl IV): 23–32.

10. Markovitz JH, Matthews KA. Platelets and coronary heart disease: potential psychophysiologic mechanisms. Psychosom Med 1991; 53: 643–68.

11. Rozanski A, Bairey N, Krantz DS, Friedman J, Resser KJ, Morell M, Hilton-Chalfen S, Hestrin L, Bietendorf J, Berman DS. Mental stress and the induction of silent myocardial ischemia in patients with coronary artery disease. N Engl J Med 1988; 318: 1005–12.

12. Kamarck TW, Jennings JR. Biobehavioral factors in sudden cardiac death. Psychol Bull 1991; 109: 42–75.

13. Manuck SB, Kaplan JR, Muldoon MF, Adams MR. The behavioral exacerbation of atherosclerosis and its inhibition by propranolol.In: McCabe P, Schneiderman N, Field T, Skylar J, editors. Stress, coping, and disease. Hillsdale (NJ): Lawrence Erlbaum Associates; 1991.p. 51–72.

14. Shivley CA, Clarkson TB, Kaplan JR. Social deprivation and coronary artery atherosclerosis in female cynomolgus monkeys. Atherosclerosis 1989; 77: 69–76.

15. Watson SL, Shivley CA, Kaplan JR, Line SW. Effects of chronic social separation on cardiovascular disease risk factors in female cynomolgus monkeys. Atherosclerosis 1998; 137: 259–66.

16. Gellman M, Spitzer S, Ironson G, Llabre M, Saab P, Pasin RD, Weidler DJ, Schneiderman N. Posture, place, and mood: effects on ambulatory blood pressure. Psychophysiology 1990; 27: 544–51.

17. Kamarck TW, Shiffman SM, Smithline L, Goodie JL, Paty JA, Gyns M, Jong JY. The effects of task strain, social conflict, and emotional activation on ambulatory cardiovascular activity: daily life consequences of “recurring stress” in a multiethnic adult sample. Health Psychol 1998; 17: 17–29.

18. Räikkönen K, Matthews KA, Flory JD, Owens JF, Gump BB. Effects of optimism, pessimism, and trait anxiety on ambulatory blood pressure and mood during everyday life. J Pers Soc Psychol 1999; 76: 104–13.

19. Guyll M, Contrada RJ. Trait hostility and ambulatory cardiovascular activity: responses to social interactions. Health Psychol 1998; 17: 30–9.

20. Stephens MAP, Crowther JH, Koss PG, Bolen K. Effects of daily activities and social behavior on blood-pressure elevation. J Behav Med 1988; 11: 553–63.

21. Spitzer SB, Llabre MM, Ironson GH, Gellman MD, Schneiderman N. The influence of social situations on ambulatory blood pressure. Psychosom Med 1992; 54: 79–86.

22. Schwartz JE, Stone AA. Strategies for analyzing ecological momentary assessment data. Psychosom Med 1998; 17: 6–16.

23. Brondolo E, Karlin W, Alexander K, Bobrow A, Schwartz J. Workday communication and ambulatory blood pressure: implications for the reactivity hypothesis. Psychophysiology 1999; 36: 86–94.

24. Ewart CK, Taylor CB, Kraemer HC, Agras WS. High blood pressure and marital discord: not being nasty matters more than being nice. Health Psychol 1991; 10: 155–63.

25. Kamarck TW, Shiffman SM, Smithline L, Goodie JL, Thompson HS, Ituarte PHG, Jong JY, Pro V, Paty JA, Kassel JD, Gnys M, et al. The Diary of Ambulatory Behavioral States: a new approach to the assessment of psychosocial influences on ambulatory cardiovascular activity.In: Krantz DS, Baum A, editors. Perspectives in behavioral medicine: technology and methods in behavioral medicine. Mahwah (NJ): Lawrence Erlbaum; 1998.p. 163–94.

26. Ituarte PHG, Kamarck TW, Thompson HS, Bacanu S. Psychosocial mediators of racial differences in nighttime blood pressure dipping among normotensive adults. Health Psychol 1999; 18: 393–402.

27. Shiffman S, Fischer LA, Paty JA, Gyns M, Kassel JD, Hickcox M, Perz W. Drinking and smoking: a field study of their association. Ann Behav Med 1994; 16: 203–9.

28. Reis HT, Wheeler L. Studying social interaction with the Rochester Interaction Record. Adv Exp Soc Psychol 1991; 24: 269–317.

29. Clark LA, Denby L, Pregibon D, Harshfield GA, Pickering TG, Blank S, Laragh JH. A quantitative analysis of the effects of activity and time of day on the diurnal variations of blood pressure. J Chron Dis 1987; 40: 671–81.

30. Russell J. A circumplex model of affect. J Pers Soc Psychol 1980; 39: 1161–78.

31. Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. J Health Soc Behav 1983; 24: 385–96.

32. Rook KS. The negative side of social interaction: impact on psychological well-being. J Pers Soc Psychol 1984; 46: 1097–108.

33. Smithline L, Kamarck T, Jin J. Ambulatory assessment of stress and social support episodes during daily living: a validation study. Ann Behav Med 1995; 17: S60.

34. Long JM, Lynch JJ, Machiran NM, Thomas SA, Malinow KL. The effect of status on blood pressure during verbal communication. J Behav Med 1982; 5: 165–72.

35. Enstrom I, Pennert K. Does it matter whether ambulatory blood pressure is recorded during a work day or a non-work day? J Hypertens 1996; 14: 565–9.

36. Spanier G. Measuring dyadic adjustment: new scales for assessing the quality of marriage and similar dyads. J Marriage Fam 1976; 38: 15–30.

37. Eddy JM, Heyman RE, Weiss RL. An empirical evaluation of the Dyadic Adjustment Scale: exploring the difference between marital “satisfaction” and “adjustment.” Behav Assess 1991; 13: 199–220.

38. Suntech Medical Instruments. Suntech Accutracker DX user’s guide, version 1.02. Raleigh (NC): Suntech Medical Instruments; 1993.

39. Goodie JL, Kamarck TW. Use of the Accutracker DX to assess behavioral influences on ambulatory cardiovascular activity: a preliminary investigation. Blood Pressure Monit 1996; 1: 135–40.

40. Marler MR, Jacob RG, Lehoczky JP, Shapiro AP. The statistical analysis of treatment effects in 24-hour ambulatory blood pressure recordings. Stat Med 1988; 7: 697–716.

41. SAS Institute. SAS/STAT software: changes and enhancements through release 6.11. Cary (NC): SAS Institute; 1996.

42. Baron RM, Kenny DA. The moderator-mediator distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986; 51: 1173–82.

43. Jennings JR. Memory, thought, and bodily response.In: Coles MGH, Donchin E, Porges SW, editors. Psychophysiology: systems, processes, and applications. New York: Guilford; 1986.

44. Weinberg JG, Levine S. Psychobiology of coping in animals: the effects of predictability.In: Levine S, Ursin GH, editors. Coping and health. New York: Plenum; 1980.

45. Julius S. The defense reaction: a common denominator of coronary risk and blood pressure in neurogenic hypertension? Clin Exp Hypertens 1995; 17: 375–86.

46. Gump BB, Matthews KA. Vigilance and subsequent cardiovascular reactivity in men: a preliminary study. Health Psychol 1998; 17: 93–6.

47. LeDoux J. Emotion: clues from the brain. Annu Rev Psychol 1995; 46: 209–35.

48. Kiecolt-Glaser JK, Malarkey WB, Chee M, Newton T, Cacioppo JT, Mao H, Glaser R. Negative behavior during marital conflict is associated with immunological down-regulation. Psychosom Med 1993; 55: 395–409.

49. Cook NR, Cohen J, Hebert PR, Taylor JO, Hennekens CH. Implications of small reductions in diastolic blood pressure for primary prevention. Arch Intern Med 1995; 155: 701–9.


blood pressure,; social interaction,; partner interactions,; cardiovascular health.

Copyright © 2001 by American Psychosomatic Society


Search for Similar Articles
You may search for similar articles that contain these same keywords or you may modify the keyword list to augment your search.