| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From St. John’s University (W.A.K., E.B.), Jamaica, New York; and State University of New York (J.S.), Stony Brook, New York.
Address reprint requests to: Elizabeth Brondolo, PhD, Department of Psychology, St. Johns University, 8000 Utopia Parkway, Jamaica, New York 11439. Email: brondole{at}stjohns.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONWorkplace Social SupportGender Differences in Social...Workplace Social Support and...METHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
METHODS: Participants included male (N = 36) and female (N = 34) New York City Traffic Enforcement Agents. Mixed-model regression analyses were used to assess the effects of self-reported measures of workplace social support on workday ambulatory blood pressure.
RESULTS: Workplace support was associated with workday ambulatory blood pressure levels, with the effects depending on the source of support and the gender of the participant. For women, immediate supervisor support was negatively associated with workday systolic blood pressure level, and a similar negative trend was found for workday diastolic blood pressure level. For men, coworker support was negatively associated with workday systolic blood pressure level. We found an interaction of social support with stress level (ie, baseline vs. high stress) such that immediate supervisor support was negatively associated with systolic blood pressure during high-stress conditions, but no effects were found under resting conditions.
CONCLUSIONS: These findings suggest that workplace social support is associated with workday ambulatory blood pressure, especially during stressful work periods.
Key Words: social support, • work stress, • blood pressure, • cardiovascular disease.
Abbreviations: ABP = ambulatory blood pressure;; BMI = body mass index;; BP = blood pressure;; CVD = cardiovascular disease;; CVR = cardiovascular reactivity;; DBP = diastolic blood pressure;; HR = heart rate;; SBP = systolic blood pressure;; SES = socioeconomic status.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONWorkplace Social SupportGender Differences in Social...Workplace Social Support and...METHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
Two theoretical models have been proposed to explain how social support may lead to lower cardiovascular activity. The main-effect model suggests that social support or social connection is a fundamental need and that social isolation will lead to elevated cardiovascular activity both under resting conditions and during periods of acute stress (2). In contrast, the buffer-effect model suggests that social support protects against increases in BP and HR under high stress conditions but has no effect in the absence of environmental or interpersonal stress (2).
As predicted by the main-effect hypothesis, correlational studies have consistently found that self-reported measures of social support are inversely related to resting BP (typically casual BP), although these effects have been small (r = 0.08; see Ref. 9). In contrast, laboratory studies that have experimentally manipulated the presence of social support have yielded results consistent with the buffering model. Specifically, most laboratory studies have failed to find effects for social support under baseline conditions, but social support has consistently reduced CVR during periods of acute interpersonal stress (8, 15, 16) . Moreover, a recent meta-analysis suggests that the size of this buffering effect may be substantial (r = 0.28; see Ref. 9).
|
Workplace Social Support |
|---|
|
TOPABSTRACTINTRODUCTIONWorkplace Social SupportGender Differences in Social...Workplace Social Support and...METHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
Previous research on workplace social support has been inconsistent. Self-reported measures of workplace social support have been negatively associated with the following: baseline BP in male factory workers (18), heart complaints in a predominantly male sample of Dutch workers (19), 24-hour ambulatory HR levels in men and women (20), and the incidence of coronary heart disease in female clerical workers (21). However, other studies have failed to find a direct relationship between workplace social support and BP (22–25) or other biological risk factors for CVD (25, 26) . Methodological problems, such as use of short, unreliable measures of workplace social support or unreliable measures of BP (eg, clinic BP as opposed to ABP) may partially account for these inconsistencies (9). Workplace social support may also yield benefits only under high-stress conditions. Inconsistent findings on the effects of workplace support may reflect variations among samples in the degree to which employees are exposed to job-related interpersonal stress.
Some investigators have also added workplace social support to the job strain model (27) to examine whether the addition of social support increases the predictive utility of this model. The job strain model conceptualizes stressful working conditions or "job strain" as conditions under which workers face high demands but have little control. Studying the combination of job strain and workplace social support serves as an indirect test of the buffer model. In two studies the combination of low workplace social support, high demands, and low control was associated with greater CVD risk than high job strain alone (28, 29) . In another study of male employees in New York City, workplace social support did not have a significant main effect on ABP, but the combination of job-related social support and job strain predicted ABP better than job strain alone (22). These results are consistent with the premise that workplace social support may help buffer the effects of stressful working conditions. However, these studies do not provide a direct test of the buffering hypothesis because differences in the effects of social support under high vs. low stress conditions were not explicitly tested.
|
Gender Differences in Social Support–Cardiovascular Relations |
|---|
|
TOPABSTRACTINTRODUCTIONWorkplace Social SupportGender Differences in Social...Workplace Social Support and...METHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
Inconsistencies in the findings of gender differences in the correlates of support may be a function of the types of support assessed. Some sources of social support may have different effects for men and women. Support from coworkers or friends seems to be more beneficial for men than women, whereas support from older relatives or supervisors may be more beneficial than peer support for women. Specifically, one study found that for men BP was inversely associated with social support from friends, whereas for older women BP was inversely related to social support from relatives (31). Similarly, Geller and Hobfoll (32) also found that coworker support and gender interacted in predicting work stress such that coworker support was negatively associated with perceived work stress for men but not women. The Framingham study found that supervisor support predicted lower rates of coronary heart disease in female clerical workers (21).
|
Workplace Social Support and Workday ABP in New York City Traffic Agents |
|---|
|
TOPABSTRACTINTRODUCTIONWorkplace Social SupportGender Differences in Social...Workplace Social Support and...METHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
Ambulatory monitoring strengthens the study because measurements can be obtained as agents participate in a variety of work activities, including interactions with the public. Studying workday ABP levels is particularly important because workday BP tends to be higher than home BP (28–30) and may be more closely related to cardiovascular outcomes (34, 35) .
Ambulatory monitoring permits us to test predictions from both the main-effect and buffering models. We examine the main-effect model by conducting between-person tests of the relationship of different sources of workplace support to baseline (resting) ABP levels as well as average workday levels. We explicitly test the predictions of the buffering hypothesis by examining the interaction of workday support and work stress level, specifically evaluating whether the within-person differences in BP and HR during high- vs. low-stress working conditions vary in relationship to perceived work-related social support. Although previous work has examined the main effect of workplace support on BP level or the additive effects of naturally occurring stressful job conditions (ie, job strain) and workplace support (28, 29) , to our knowledge this is the first naturalistic study that has explicitly examined the association of workplace support to BP under conditions of low and high stress.
The use of a sample comprising members of a single workforce permits the elimination of the effect of different job characteristics (eg, type of work, demands, control, physical characteristics of the work environment) and/or SES, which may indirectly affect perceptions of support and directly affect BP. In addition, more than 50% of the traffic agent workforce is African American, and the average starting income is below $25,000. The SES of this sample is appropriate to study because prior work suggests that workplace social support–CVD relations are stronger among people of lower SES (28, 29) .
Men and women are also equally represented in this employee group and have identical job responsibilities. This provides the opportunity to examine possible gender differences in the associations of workplace social support to workday ABP without the potential confound of gender differences in pay or job responsibilities. In this study we predicted that social support would be inversely associated with workday BP in both and men and women, but that coworker support would be more closely associated with BP among men and supervisor support would be closely related to BP among women.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONWorkplace Social SupportGender Differences in Social...Workplace Social Support and...METHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
The primary task of the agents participating in this study was to patrol the streets on foot and issue summonses. The agents included in this study worked 8-hour days, spending about 1 hour at the start and 1 hour at the end of their work shift in the office and the remainder of work time in the field. Agents served on six different work shifts, but prior analyses indicated that the time of their work shift did not affect ABP (36).
The structure of the workforce is such that each agent is assigned to a squad of approximately eight agents, and an immediate supervisor leads each squad. A unit supervisor commands several squads in the same work location. Agents have frequent contact with their immediate supervisor and much less contact with their unit supervisor. Agents generally patrol the streets alone, but they may have brief contact with their peers during the day. They have more extensive contact with coworkers and their immediate supervisors at the start and end of their tours. If problems or conflicts with motorists arise in the field, agents are able to contact their immediate supervisors by radio.
Equipment
Measures of SBP, DBP, and HR were collected throughout the workday using the Suntech Accutracker II (Suntech Medical Instruments, Raleigh, NC). The Accutracker consists of a computer-driven unit that is about the size of a portable cassette player and weighs under 1 pound, three ECG leads that attach to the chest, and a BP cuff that attaches to the nondominant arm. The device is preprogrammed to record BP and HR at preset intervals. The monitor is safe, reliable, and places minimal restrictions on the subject’s usual activities. This is critical because the agents are active throughout their workday. The Accutracker has been shown to be more accurate and more reliable in obtaining BP measurements from active subjects than many other available model (37), and the BP and HR measurements made by the Accutracker have been validated against intra-arterial readings and clinician measurements made with a mercury-column sphygmomanometer (38, 39) .
Procedure
Data collection occurred only during the workday. Research personnel met agents at their worksite at the beginning of each agent’s tour. In a quiet room, questionnaires were completed, and height and weight were measured. Agents were outfitted with an ABP monitor, and the accuracy of the ABP readings was checked with an aneroid sphygmomanometer. Before going out to the field, five baseline readings were obtained. Three seated resting readings were taken after agents had rested without talking for 10 minutes. Next, agents stood quietly for 2 minutes, after which two additional BP measurements were taken with the agent standing. After baseline readings were taken, the ABP monitors were programmed to measure BP and HR readings every 15 minutes throughout the workday. If an error occurred as a measurement was obtained, another reading was attempted 4 minutes later.
At the time of each field BP measurement, agents completed a diary page. The diary page assessed location (in the field or at the office), posture (sitting, standing, or walking), and activity (paperwork, traveling, looking for violations, writing summonses, talking to a motorist, talking to a coworker, talking to a supervisor, break, or eating).
Workplace Social Support
Social support was assessed with a modified version of the Job Content Questionnaire (40). The modifications were based on focus group discussions with more than 80 agents. Three separate scales measured perceived emotional support from coworkers (7 items), immediate supervisors (16 items), and unit supervisors (7 items). Sample items from these scales are as follows: "My coworkers care about me"; "My lieutenant [immediate supervisor] backs me up if there is a problem"; "My lieutenant is concerned about the welfare of those under him"; and "My captain [unit supervisor] listens and helps when I’ve had a conflict with the public." Agents rated items on a 4-point Likert-type scale in which 1 indicated "strongly disagree" and 4 indicated "strongly agree" (see Ref. 41 for a complete list of items).
Analyses revealed that all three of these scales had adequate internal consistency: coworker support, 
= 0.83; immediate supervisor support, = 0.77; and unit supervisor support, = 0.86. Data from another sample of agents (41) indicated moderate stability over a 4-month period (test re-test correlations were r = 0.64 for coworker support, r = 0.61 for immediate supervisor support, and r = 0.49 for unit supervisor support). In that same study (41), immediate supervisor support was positively associated with both productivity and job satisfaction, and unit supervisor support was inversely related to burnout, suggesting that these scales are differentially associated with important work-related variables.
In the present sample coworker support was slightly related to immediate supervisor support (r = 0.21, p < .08) but unrelated to unit supervisor support (r = 0.10, p > .41). Immediate supervisor support was significantly related to unit supervisor support (r = 0.51, p < .0001). The magnitudes of these correlations suggest that the three scales measure distinct phenomena, and their effects should be examined separately.
Analytic Plan
Data reduction.
A total of 1712 readings were attempted on the 70 subjects. Detection and deletion of artifactual readings followed advice from the manufacturer and the suggestions outlined by Clark et al. (42). SBP, DBP, and HR readings were deleted if the difference between SBP and DBP exceeded 90 mm Hg or was less than 20 mm Hg. SBP values were also deleted if they fell below 84 mm Hg or above 200 mm Hg; DBP values were deleted if they fell below 44 mm Hg or above 148 mm Hg. HR values were also deleted if they fell below 46 beats/min or above 131 beats/min. These ranges were chosen because points outside were uniformly accompanied by codes from the Accutracker indicating that an error (movement, inadequate Korotkoff sounds, etc.) had occurred during these periods. As suggested by Jamner et al. (43), if either the DBP or SBP reading was deemed invalid, then both BP readings were deleted. After the deletion of artifactual readings, the data set contained 78% of the BP readings and 76% of the HR readings attempted. In some cases the BP readings seemed valid and the Accutracker indicated that no error had occurred, but the HR values were out of the acceptable range. In these cases HR was deleted, but the BP readings were retained. Changes in respiration could have adversely affected calculation of the HR but not the BP readings. The invalid and valid ABP measurements did not differ in terms of the type of workday activity in which the agent was engaged (eg, issuing summonses, taking a break, etc.). There were no gender differences in the percentage of valid and invalid ABP measurements. The analyses presented here include the four subjects on medications likely to affect BP, but very similar results were obtained when the analyses were done excluding these subjects.
Overview of analysis plan.
Mixed-model regression analysis was used to test hypotheses. Mixed-model analyses allow a more rigorous examination of ecological momentary assessment data such as ABP data, in which repeated measures are collected from the same person (44). PROC MIXED, a relatively new statistical procedure from the SAS Institute, was used to perform these analyses (45). PROC MIXED is a generalization of the standard linear model that uses maximum likelihood to estimate and test the significance of between- and within-subject effects, controlling for variables associated with repeated observations from the same subject. These effects can be evaluated even when the between-subject and within-subject variances are nonorthogonal, which typically occurs when there are different numbers of observations for each subject or for each level of the independent variable (36, 46) . This was valuable since agents varied in the total number of acceptable BP and HR readings.
Analyses were performed using two error structures: compound symmetry (the standard error structure in repeated-measures analysis of variance) and a combination of compound symmetry and first-order autoregression. Compound symmetry assumes correlated residuals due to unexplained variation between participants. First-order autoregression is hypothesized to be relevant to most time-series data because cardiovascular measurements taken closer together in time are expected be more closely correlated than measurements taken at more distal points in time. In all cases except the one noted below, the combined error structure of autoregression and compound symmetry was used. The combined error structure is the more general error structure, and it has proved to fit ABP better in most of these analyses and in previous analyses done by our laboratory (36). In the analysis of predictors of DBP levels in women, the covariance parameter for the first-order autoregression error structure was close to zero, and PROC MIXED did not converge on a solution when the combined error structure was used, so compound symmetry was used.
Three separate sets of analyses were performed. In the first two sets of analyses we examined the relationship of workplace social support to resting baseline BP and HR level and to average workday BP and HR levels. In the next analysis we examined the relationship of support to BP obtained under conditions of both low and high stress, testing for an interaction of support and stress level. To clarify the interaction, we separately examined the relationship of support to BP obtained during conditions of high and low stress.
In each of these sets of analyses, the predictor variables included the three support measures and the interactions of each of these support scores with gender, given that gender differences in the effects of social support on BP or HR were expected. In the third set of analyses, stress level and its interaction with both support and gender served as additional predictor variables. Ambulatory BP and HR served as outcome variables. Between-person control variables included age, gender, and BMI. Within-person control variables included posture and location. Posture was dummy coded with two variables, one for standing (yes, no) and the other for walking (yes, no). Posture was regarded as "sitting" in all remaining observations for which posture was coded. Location was treated as a class variable with two levels: office and field. In the analysis testing whether the relation of social support to BP/HR is moderated by stress level (ie, high vs. low), the effect of the stress variable was treated as a random effect. The effects of the two within-person control variables were treated as fixed to reduce the complexity of the models.
To control for Type I error we routinely performed omnibus tests. In the first two sets of analyses we assessed whether the set of three social support variables and their interactions with gender significantly improved the models predicting either baseline or workday BP or HR, before examining the significance levels of the individual support measures and/or their interactions with gender. In the second set of analyses, we used omnibus analyses to assess whether the support measures and their interactions with gender and/or stress level were statistically significant before examining the significance levels of individual terms or interactions among the terms. Specifically, as suggested by Schwartz and Stone (44), we used the change in the -2LL statistic to test whether the addition of the support terms or their interactions with gender or stress significantly improved the fit of the model to the data. Significant interactions of social support by gender were followed up by separate tests in men and women of the relationship of social support to BP and HR. Significant interactions of social support by stress level were followed up by separate tests of the relationship of social support to BP and HR at low and high levels of stress.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONWorkplace Social SupportGender Differences in Social...Workplace Social Support and...METHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
Mean levels of BP and HR at baseline, throughout the full workday, and during high stress periods only are shown in Table 2 for men and women separately. BP was significantly greater in men than women, and these effects were significant in all but one condition (p values < .05). Women tended to have higher HR levels than men, but these effects were not significant.
|
The resting baseline was considered a low-stress condition. The three sitting resting readings and the two standing resting readings taken at the start of the workday were used as measures of baseline BP and HR. Sixty-six participants, each of whom had five valid BP or HR observations, were used in the analyses. Omnibus tests of the improvement in the models resulting from the addition of the three support scores and their interactions with gender were not significant (all p values > .25). When examined separately, none of the social support measures was independently related to baseline SBP, DBP, or HR.
Workplace Support and Workday BP and HR
To examine the association of support to BP and HR obtained under conditions varying in level of stress, analyses were performed with workday BP and HR serving as the dependent variables. BP and HR observations from all 70 participants were used to test the hypothesis that support was associated with BP and HR levels across the workday. On average, each agent had 17.5 valid BP measures taken during the workday (not including baseline measures). During the workday agents engaged in a variety of activities, including looking for violations; issuing summonses; communicating with coworkers, supervisors, and the public; taking breaks; and completing paperwork. These activities vary in perceived stress and associated negative affect (33, 36) .
Omnibus tests examining the effects of support scores alone did not significantly improve the model. However, as predicted, omnibus tests revealed that the support scores and their interactions with gender significantly improved models for workday SBP level (
2 = 14.1, df = 6, p < .05) and workday DBP level (2 = 13.5, df = 6, p < .05) but not workday HR level. There were significant interactions of immediate supervisor support and gender on both SBP level (ß = 14.27, SE = 5.87, df = 60, t = 2.43, p < .019) and DBP level (ß = 10.58, SE = 3.79, df = 60, t = 2.79, p < .008). Coworker support also interacted with gender on SBP workday level (ß = -9.34, SE = 4.21, df = 60, t = -2.22, p < .031) but not on DBP workday level. None of the interactions of unit supervisor support and gender were significant.
To evaluate the significant interactions, the relationship of immediate supervisor support to SBP and DBP and the relationship of coworker support to SBP were examined separately in men and women. For women, immediate supervisor support was negatively associated with SBP level (ß = -6.69, SE = 3.00, df = 31, t = -2.23, p < .034); a similar trend was found for DBP level (ß = -3.31, SE = 1.92, df = 31, t = -1.72, p < .096). For men, immediate supervisor support was not significantly related to workday BP. Figure 1 displays the interaction between immediate supervisor support and gender on workday SBP level. For men, but not for women, coworker support was negatively associated with SBP level (ß = -6.39, SE = 2.86, df = 32, t = -2.24, p < .033).
|
Tests of the Buffering Hypothesis
The buffering hypothesis suggests that support moderates the effects of stress on BP. This hypothesis predicts both that the association of support on BP will emerge under high-stress, but not low-stress, conditions and that the association of acute stress to BP will be smaller for individuals with high vs. low social support. To test the buffering hypothesis, we examined the interaction of social support and stress level.
Stress level is a within-person variable with two levels, low and high. The low-stress condition comprised the resting baseline, and the high-stress condition included episodes of issuing summonses or talking to motorists. Our prior research examining self-reported stress and BP indicated that the most consistently stressful work activities were issuing summonses and communicating with motorists. Focus group discussions revealed that members of the public often try to interact with agents when the agents are issuing summonses. In a prior report we noted that interactions with motorists were associated with increases in BP even when the nature of the communication did not involve overt conflict (36). Agents also report more negative emotions and less positive emotions when interacting with motorists as opposed to interacting with coworkers or supervisors (36). Workday readings that occurred as agents were interacting with motorists and/or issuing summonses were associated with greater BP than other workday activities. See Table 2 for average BP and HR levels obtained under high-stress and full-workday conditions. For this analysis we included only low- and high-stress episodes because the stress level associated with many of the other workday activities (including talking with coworkers, looking for violations, or completing paperwork) is more ambiguous. For example, the stress associated with the experience of completing paperwork can vary depending on circumstances specific to the office, season, or quality of supervision.
Fifty-seven agents were included in this analysis. The model to test the buffering hypothesis included the covariates and predictors used in the previous models. In addition, a within-person variable for stress level (baseline vs. high stress) was created. This stress level variable and interactions between workday support scores and stress level were also included as predictors. Finally, to test for the possibility that buffering effects would be moderated by gender, three-way interactions between support scores, stress level, and gender were included. The effect of stress level was treated as a random effect, and the effects of the other within-person control variables were treated as fixed effects.
Initial analyses were performed to check that activities identified as high stress were, in fact, on average associated with higher BP levels than those in the low-stress condition. A mixed-model analysis was conducted with age, gender, BMI, and posture serving as covariates and stress level (treated as a random effect) serving as the only predictor variable. Results indicated that stress level (baseline vs. high stress) was significantly related to SBP (ß = 11.52, SE = 1.61, df = 473, t = 7.15, p < .0001) and HR (ß = 6.97, SE = 1.74, df = 452, t = 4.00, p < .0001) but not DBP (ß = 2.05, SE = 1.60, df = 473, t = 1.28, p < .20). High-stress periods were associated with SBP levels almost 12 mm Hg higher than baseline and HRs almost 7 beats/min higher than baseline. Further analyses of the interaction of stress level and gender indicated that the effects of stress level on BP and HR were not moderated by gender.
An omnibus test revealed that the interactions of support scores and stress level significantly improved the model for SBP (2 = 12.9, df = 3, p < .005). The interaction of immediate supervisor support and stress level on SBP was significant (ß = -5.04, SE = 2.51, df = 468, p < .046). Interactions including the other measures of support were not significant. Other omnibus tests revealed that the two-way interactions of support and gender and the three-way interactions of support, gender, and stress level did not significantly improve the models for SBP, DBP, or HR.
A follow-up test of the interaction of immediate supervisor support and stress level on SBP revealed that immediate supervisor support was negatively related to SBP under high-stress conditions (ß = -8.54, SE = 3.39, df = 470, t = -2.52, p < .013), but the effect was not significant under low-stress conditions (ß = -2.47, SE = 2.20, df = 470, t = -1.12, p < .27). Although the test of the interaction between coworker support and stress level was not significant (p < .14), a similar pattern was found such that coworker support was negatively associated with SBP under high-stress conditions (ß = -5.68, SE = 2.82, df = 470, t = -2.02, p < .045) but not during low stress (ß = -2.49, SE = 1.89, df = 470, t = 1.31, p < 20).
A correlational analysis of the relationship of support scores to the number of readings obtained during high-stress episodes indicated that immediate supervisor support was weakly associated with the number of BP measurements obtained during high-stress periods (r = 0.24, p < .044), but support from coworkers or from unit supervisors was not significantly related to the number of BP observations taken during high-stress periods. The positive relationship of immediate supervisor support and the number of high-stress measurements suggests that agents who felt supported by their supervisors may have been more likely to take risks in terms of administering summonses and interacting with motorists.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONWorkplace Social SupportGender Differences in Social...Workplace Social Support and...METHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
The analyses investigated the main effect of social support by examining the association of support to BP and HR levels obtained at rest and throughout the workday. The buffer-effect model was investigated by testing the interaction of support and stress level on BP and HR. Results indicated that workplace social support was significantly related to both workday BP levels and SBP obtained under high-stress conditions, but as predicted, the specific effects depended on the gender of the participant and/or the source of workplace social support.
Our results are consistent with a body of laboratory research that suggests that the presence of a socially supportive person reduces CVR in socially threatening situations (for reviews, see Refs. 8 and 15). These field data extend laboratory findings that within-person situational variables (ie, high vs. low stress) moderate the effects of social support on BP. But unlike laboratory studies, which typically manipulate the presence or absence of a supportive individual, this study examined the effects of real-life global perceptions of workplace social support on workday BP during naturally occurring periods of high and low stress. The ambulatory data indicate that perceptions of workplace social support were negatively associated with both increases in BP during acute periods of stress and BP levels throughout the workday. Furthermore, the findings suggest that the perception that social support is available (as opposed to the physical presence of support) may be sufficient to yield cardiovascular benefits.
Overall our results are more consistent with the buffering-effect model of the social support–cardiovascular functioning relationship rather than the main-effect model. Measures of social support were not related to baseline measures of BP and HR, but significant effects were found for workday SBP levels, and trends were found for workday DBP. This naturalistic comparison of the main-effect model vs. the buffering-effect model provides particularly strong evidence that the stress-buffering theory better fits the data.
In this study significant effects of workplace social support on workday ambulatory BP or HR levels and reactivity emerge most clearly for measures of SBP. Our lack of findings for DBP in comparison to those found for SBP may be partly explained by the fact that agents responded to high-stress situations with much larger mean increases in SBP than DBP (12 vs. 2 mm Hg), suggesting that interpersonal stress had a larger impact on SBP than DBP. This finding is consistent with other stress-related research that has found that some stress-related personality characteristics, such as Type A behavior, are more closely associated with SBP than DBP (47). The differences found for SBP and DBP responses may also reflect a particular hemodynamic pattern. Social support and interpersonal stress may have had greater influence on cardiac output than vascular resistance, thus explaining the lack of effects for DBP. Future research can examine hemodynamic response patterns to better understand the nature of the relationship of social support to cardiovascular activation.
The results also suggest that different sources of social support in the workplace have different effects. For example, unit supervisor support was largely unrelated to BP and HR responses throughout the workday. Agents have less contact with their unit supervisors than with their immediate supervisors and coworkers. In addition, immediate supervisors are responsible for interacting with agents in the field and are more likely to serve as the first response in cases of difficulty. The closeness or regularity of contact may determine the strength of the support-BP relationship.
Some sources of workplace social support also seemed more important for one gender than the other in predicting workday levels of BP and HR. Immediate supervisor support was negatively associated with workday BP in women but not men. In contrast, coworker support was negatively associated with workday SBP level in men but not women. These findings are consistent with previous research that has found that coworker support is more closely associated with work and health-related outcomes for men, whereas supervisor support has been associated with health-related outcomes for women (21, 32) .
Men may benefit more than women from coworker support because close social relationships may sometimes be a source of stress or demand for women. Both men and women are more likely to seek social support from women, and playing the role of caregiver or support provider can be demanding and stressful (30). Thus, female agents may be more likely than male agents to be adversely affected by social problems and tensions among coworkers. This premise is also consistent with epidemiological research that has found more consistent benefits from social support in men than women (2, 30) .
No gender differences emerged in the relations of social support to acute increases in SBP during high-stress activities. This finding is consistent with most laboratory studies that have failed to find gender differences in the relations of social support to CVR during periods of acute stress (48, 49) . Although close social relationships may sometimes be a source of stress for women, social support from peers may still be beneficial during periods of high interpersonal stress. Likewise, men who may sometimes be unresponsive to support from their supervisors may benefit from immediate supervisor support during stressful work periods. Our finding of gender differences in social support for workday BP may reflect gender differences that emerged under moderate stress levels rather than the effects of social support during high stress.
Study Limitations
Although previous studies have often evaluated workplace social support in combination with the work demands/work control components of the job strain model (eg, Refs. 28 and 29), this study focuses solely on workplace social support. We did not measure work demands and work control because on an objective level these variables do not vary within our single occupation sample. However, we predicted that there would be considerable individual differences in perceptions of workplace social support, and in turn these perceptions were expected to be associated with cardiovascular responses throughout the workday. Nevertheless, it may be important to examine workplace social support in conjunction with other negative characteristics of the work environment (eg, lack of control, high demands, and low rewards).
Our overall interpretation of the results of this study is partly limited by the cross-sectional nature of the design. In theory workplace social support leads to reductions in workday cardiovascular activity; however, although unlikely, it is also possible that higher levels of workday cardiovascular activity lead to lower levels of perceived workplace social support. A third variable may also account for our effects. For example, high levels of depression may contribute to both lower levels of workplace social support and higher workday BP and HR levels. Data from our prospective, longitudinal study will permit a stronger test of the hypothesis that changes in social support produce changes in BP and HR
It is also unclear whether these results would generalize to other employee samples, particularly workers who encounter only small amounts of stress at work. However, many employee groups face high levels of interpersonal stress at work. For example, other law enforcement groups, customer service representatives, and others are exposed to high rates of interpersonal stress at work and may be responsive to workplace social support.
|
CONCLUSION |
|---|
|
TOPABSTRACTINTRODUCTIONWorkplace Social SupportGender Differences in Social...Workplace Social Support and...METHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONWorkplace Social SupportGender Differences in Social...Workplace Social Support and...METHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
Received for publication November 21, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONWorkplace Social SupportGender Differences in Social...Workplace Social Support and...METHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  C. J. Rodriguez, M. M. Burg, J. Meng, T. G. Pickering, Z. Jin, R. L. Sacco, B. Boden-Albala, S. Homma, and M. R. Di Tullio Effect of Social Support on Nocturnal Blood Pressure Dipping Psychosom Med, January 1, 2008; 70(1): 7 - 12. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Brondolo, D. J. Libby, E.-g. Denton, S. Thompson, D. L. Beatty, J. Schwartz, M. Sweeney, J. N. Tobin, A. Cassells, T. G. Pickering, et al. Racism and Ambulatory Blood Pressure in a Community Sample Psychosom Med, January 1, 2008; 70(1): 49 - 56. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. H. Wirtz, R. von Kanel, C. Mohiyeddini, L. Emini, K. Ruedisueli, S. Groessbauer, and U. Ehlert Low Social Support and Poor Emotional Regulation Are Associated with Increased Stress Hormone Reactivity to Mental Stress in Systemic Hypertension J. Clin. Endocrinol. Metab., October 1, 2006; 91(10): 3857 - 3865. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. S. Lett, J. A. Blumenthal, M. A. Babyak, T. J. Strauman, C. Robins, and A. Sherwood Social Support and Coronary Heart Disease: Epidemiologic Evidence and Implications for Treatment Psychosom Med, November 1, 2005; 67(6): 869 - 878. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S Radi, T Lang, V Lauwers-Cances, E Diene, G Chatellier, L Larabi, R De Gaudemaris, and for the IHPAF group Job constraints and arterial hypertension: different effects in men and women: the IHPAF II case control study Occup. Environ. Med., October 1, 2005; 62(10): 711 - 717. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
