| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
REACTIVITY SPECIAL SECTION |
From the Department of Psychology, University of Pittsburgh (T.W.K.), Pittsburgh, Pennsylvania; and the Department of Psychiatry and Behavioral Sciences, University of Oklahoma Health Sciences Center (W.R.L.), Oklahoma City, Oklahoma.
Address correspondence to: Thomas W. Kamarck, Department of Psychology, University of Pittsburgh, 4403 Sennott Square, 210 S. Bouquet Street, Pittsburgh, PA 15260. Email: tkam{at}pitt.edu
ABSTRACT
OBJECTIVE AND METHODS: This article is a selective review of recent findings bearing on the conceptualization and measurement of cardiovascular reactivity to psychological challenge, with a focus on several issues relevant to the reliability, content validity, construct validity, and criterion validity of these measures.
RESULTS AND CONCLUSIONS: With respect to reliability, use of standardized task demands and aggregated scores are associated with enhanced short-term reliability, but the long-term reliability of cardiovascular reactivity has not been sufficiently documented. With respect to content validity, existing evidence suggests that "vascular" or "cardiac" tasks may evoke responses that reflect similar distributions of individual difference, whereas associations between responses to "physical" and "psychological" tasks are modest. The evidence is not clear at present with respect to the importance of including affective or interpersonal stimuli as part of trait reactivity assessments. With respect to construct validity, existing data show that cardiovascular reactivity to psychological challenge is largely independent of standard measures of autonomic function. With respect to criterion validity, recent studies point to a number of methodological limitations that may have restricted our ability to detect lab-to-life generalizability of reactivity measures in the past. Continued progress in understanding and measuring reactivity as an individual difference dimension is essential in helping us to evaluate emerging evidence examining the relationship between reactivity and disease risk.
Key Words: cardiovascular reactivity • measurement • biopsychology • cardiovascular disease.
Abbreviations: BP = blood pressure; CNS = central nervous system; CVR = cardiovascular reactivity; DBP = diastolic blood pressure; HR = heart rate; RSA = respiratory sinus arrhythmia; SBP = systolic blood pressure; SNS = sympathetic nervous system.
Individual differences in physiological responses to environmental events have been a longstanding concern in psychosomatic medicine (1) based, in part, on the assumption that such differences may be a marker of disease vulnerability. A more specific interest in cardiovascular reactivity (CVR) to psychological challenge has evolved as evidence has accumulated linking the Type A behavior pattern and hostility with exaggerated stress-related cardiovascular responding (2). The past decade has witnessed a growing literature examining the reactivity hypothesis, linking exaggerated cardiovascular reactivity with elevated risk for hypertension or coronary heart disease (3). Despite its increased prominence in the scientific literature, however, there is still little consensus on the meaning and measurement of CVR.
This review provides a status report on the conceptualization and measurement of CVR, with an effort to highlight a number of psychometric issues that are sometimes overlooked in this literature. Evidence examining CVR as a marker or risk factor for cardiovascular disease is beyond the scope of this review and is covered elsewhere in this series (4). We take the position that careful consideration of conceptual and measurement issues in CVR may be a precondition for evaluating the current literature on the reactivity hypothesis as well as for designing solid research in this area.
Following an introduction, four assessment principles will be used as a scaffolding for our discussion, and we will select several topics relevant to each of these principles for further examination: 1) the reliability of CVR, especially its stability across time; 2) the content validity of CVR, correspondence with the intended domain of measurement, and the degree to which various types of stimuli may trigger a person’s reactive tendencies; 3) the construct validity of reactivity measures, specifically their relationship with standard measures of autonomic function; and 4) the criterion validity of CVR measures, especially their generalizability from laboratory to real-life settings.
CVR is usually conceptualized as an individual difference or trait characteristic, and it is usually measured by examining changes in cardiovascular functioning elicited by aversive, challenging, or engaging laboratory tasks, such as mental arithmetic, cold pressor, or public speaking (5). As described by Lovallo and Gerin (6), there are a number of biopsychological processes that may contribute to individual differences in reactivity, involving inputs from the periphery (eg, differences in vascular morphology or receptor function) as well as the central nervous system (eg, differences in patterns of appraisal or centrally mediated autonomic activation).
In this review we are concerned primarily with CVR during psychological challenge: we define such challenges as situations that are motivationally relevant (ie, posing negative or positive consequences of importance to the individual) and that require adaptive (cognitive or motoric) responding. Unlike other types of adjustments, CVR to psychological challenge is assumed to be elicited by central command centers (frontal lobe, hypothalamic, and brainstem influences) priming motor output systems in preparation for fight or flight (eg, the "motor preparation hypothesis"; Refs. 7 and 8). Our emphasis on CVR to psychological challenge is based on the conceptualization of CVR as a potential mediator of the relationship between psychosocial factors and disease risk (9). This emphasis on psychological or central nervous system (CNS) influences on reactivity was emphasized in early research on this topic, showing that peripheral metabolic factors alone cannot account for variability in CVR across individuals (10). Although such an emphasis has conceptual and historical significance, the relative prognostic importance of central and peripheral determinants of reactivity has not been directly tested.
Although cardiovascular reactivity is usually conceptualized as a unitary construct, it is frequently measured by a variety of interrelated response parameters that are treated as statistically independent constructs. As an alternative, a number of investigators have proposed that reactivity should be conceptualized as a two-dimensional disposition, with individuals varying along dimensions of "cardiac" (eg, cardiac output) change and "vascular" (eg, peripheral resistance) change in response to standardized stimuli (11, 12). Converging evidence supports this two-factor solution for reactivity (13, 14). For example, in a recent study, changes in heart rate (HR), systolic blood pressure (SBP), diastolic blood pressure (DBP), stroke volume, and preejection period in response to four standardized laboratory tasks were examined. In comparing several statistical models, a two-factor solution was found to be the best fit for these multiple response parameters, and each of the two factors was reliable on retest (15). Related efforts have also been undertaken to characterize HR reactivity as a function of two dimensions (sympathetic and parasympathetic activity), in which varying contributions by each may underlie similar HR responses across individuals (16). Further work is needed to determine the usefulness of such models in identifying markers of disease risk.
RELIABILITY OF CARDIOVASCULAR REACTIVITY
A reliable measure is consistent or stable across time and settings. CVR is a measure of physiological change, and statisticians in the past have been pessimistic about our ability to measure change reliably (17). Reevaluating earlier assumptions, modern accounts are more sanguine about the measurement of change (18), and both arithmetic as well as residualized change scores have been shown to be reliable for the purpose of CVR assessment (19). Whereas individual differences in task-related autonomic or cardiovascular activation have traditionally been the focus of this research, the dynamics of cardiovascular recovery after task offset have garnered increasing interest in recent years (20). There is no consensus on the best method for measuring cardiovascular recovery after psychological challenge, however, and only a few studies have begun to assess the psychometric properties of various approaches (21, 22). Because it provides a larger literature from which to draw, we emphasize the measurement of task-related activation rather than recovery in the remainder of this report.
Although individual differences in CVR are assumed to reflect a stable disposition, their specific manifestations are expected to vary across settings and tasks. Trait differences in CVR, therefore, likely constitute only one of many determinants of cardiovascular responding in a given situation. As with any assessment, the testing setting for measuring such trait differences can be optimized by reducing effects of unwanted sources of variance (23). In line with this principle, we suggest that peripheral triggers of cardiovascular activation (eg, muscle movement) should be minimized when assessing CVR, under the assumption that these are of lesser interest. By contrast, mobilization of psychological effort should be optimized in the eliciting tasks.
When measuring individual differences in ability, test items that will maximize the distinction between respondents are selected; items that are more difficult, in general, are usually not the best suited for this purpose (23). Similarly, when assessing trait CVR, the focus should be on tasks that maximize variability rather than on tasks that maximize the magnitude of response per se. But laboratory tasks that produce dramatic responses may be useful for other purposes, for example, when exploring the effects of stressor exposure on myocardial ischemia or other triggers of acute events (24).
For a variety of individual tasks, differences in cardiovascular responding have been shown to track over time, although the magnitudes of these effects vary widely. For example, test-retest correlations for HR reactivity have ranged from 0.32 to 0.91 (5). Methodological factors seem to contribute to these inconsistent findings: Meta-analytic results suggest that BP reactivity measures are more reliable when tasks do not have speaking requirements (eg, mean test-retest of 0.27 vs. 0.52 for SBP reactivity in cognitive tasks involving speech and no speech, respectively; see Table 7 in Ref. 25). Calibrating task difficulty across subjects may also enhance reliability. Some investigators (26, 27) have used computer-based task protocols for this purpose; such protocols adjust task difficulty commensurate with performance, permitting participants to remain optimally challenged throughout each task.
Reliability of reactivity assessment is also enhanced by averaging across tasks and test sessions, reducing the idiosyncratic contribution of individual occasions of measurement and maximizing the effects of "trait" characteristics that may contribute to the overall score. The use of aggregation to improve reliability rests on standard psychometric principles (eg, Refs. 23 and 28–30). Short-term reliability, using data from 1-week and 1-month retests, has been shown to approach conventional psychometric standards using this approach. For example, retest reliability for SBP reactivity ranged from 0.71 to 0.81 across five studies when data were aggregated across tasks in a single session and from 0.83 to 0.90 when multiple task data were aggregated across two sessions (see Table 7 in Ref. 31).
Retest reliability of CVR may shrink as longer testing intervals are used, although there are few data available to address this question at present (25). Whereas reliability has been measured over intervals exceeding a year (eg, Refs. 32 and 33), existing reports are based on responses to individual tasks, and these do not take advantage of the aggregated assessments described above. The possibility that long-term reliability of reactivity measures may be attenuated by structural changes related to age or disease should also be considered.
In sum, there has been some progress in recent years in understanding the factors that may enhance reliable assessment of CVR, and a number of findings over the short term (weeks and months) are consistent with the conceptualization of reactivity as a stable trait. There seems to be some reduction in the stability of CVR over longer periods, but there is a need for additional research in this area, using methods that have been shown to enhance assessments over the short term.
CONTENT VALIDITY OF CARDIOVASCULAR REACTIVITY
As discussed above, aggregation of scores across tasks or sessions may serve to enhance the reliability of CVR measurement. The need to select representative test stimuli in the service of aggregation, however, poses significant challenges. One might want to be cautious about aggregating scores across tasks shown to be drawn from different conceptual domains (eg, tasks assessing different types of stressors or responses). The notion that tasks are conceptually different for the purpose of measuring CVR cannot be readily inferred, however, without empirical observation. For example, two tasks that are topographically quite varied, such as mental arithmetic and public speaking, may, in fact, evoke similar patterns of psychological effort with commensurate CNS-mediated changes in cardiovascular activation. It may not be meaningful to distinguish between "mental arithmetic reactivity" and "public speaking reactivity" under such circumstances.
Content validity refers to the extent to which test items adequately cover a representative sample of content. Across what types of task dimensions should we aggregate to ensure that the content domain of stable individual differences is appropriately assessed? This is an empirical question (which types of tasks or items load together?) as well as a theoretical one (what domain should be covered in the assessment of reactivity?). We will further consider this issue as it applies to four types of task dimensions in reactivity assessment: 1) "physical" vs. "psychological" stressors; 2) task differences in the patterns of physiological responding; 3) the role of negative emotional states; and 4) the role of interpersonal challenges.
Can Reactivity Be Assessed Equally Well With Physical As With Psychological Tasks?
We review the literature comparing responses to psychological or mental challenges to those elicited by "physical tasks," that is, aerobic exercise and cold pressor challenge. The response determinants that are unique to these two types of physical stimuli, as well as those that are shared with psychological tasks, will be considered.
Aerobic exercise.
A number of response determinants may be shared across both aerobic exercise and sedentary psychological challenges; for example, it is possible that a centrally mediated preparation for effort may be a feature of both types of situations (34), and differences in peripheral adrenergic receptor sensitivity may exert similar effects on responding across both domains as well. Although these two types of tasks share some common features, there are important differences. With aerobic exercise, unlike sedentary psychological demands, a major determinant of cardiovascular changes involves peripheral metabolic activity (35). Patterns of cardiovascular responding to these two types of challenges, therefore, diverge, with exercise triggering decreases, rather than increases, in DBP (36) and peripheral resistance (7).
Existing evidence suggests that, with respect to the cardiovascular changes they evoke, aerobic exercise and psychological challenges may reveal different distributions of individual difference. Two studies have compared cardiovascular responses during mental stress and aerobic exercise (13, 36), and both report similar findings: associations between responses to a mental and exercise stressor are relatively smaller (average r for HR reactivity = 0.03 (see Table 4 in Ref. 13), 0.19 (see Table 3 in Ref. 36) than those between responses to two psychological tasks (average r for HR reactivity = 0.45 in each study).
Given the independent set of response determinants that drive cardiovascular changes during exercise, it is not surprising that correlations with sedentary psychological tasks are relatively low. If the goal is to elicit large metabolic changes that may tax the workload of the heart, aerobic exercise testing has certainly proven to be a useful tool (37). If the goal is to characterize individual differences in cardiovascular functioning elicited primarily by psychological challenges, however, aerobic exercise may not be a good prototype of the intended content domain.
Anticipation of physical activity.
Unlike aerobic activity, anticipation of exercise seems to evoke descending cortical influences (38) in the absence of increased metabolic demand. Such circumstances may cause significant increases in HR and BP in advance of actual exertion (37), and these changes may be associated with individual differences. In a large sample of Finnish men (39), BP responses in anticipation of bicycle exercise were quite robust (mean SBP increases of 16 and 25.5 mm Hg among men in their 40s and 50s to 60s, respectively; Ref. 40), and in subsets of this sample, larger responses were found to be associated with left ventricular hypertrophy (40) as well as 4-year changes in BP and carotid disease (39, 41). Because these anticipatory responses reflect central commands rather than peripheral metabolic inputs, we might expect exercise anticipatory responses to correspond more closely with responses to mental challenge than with physical activity proper. The use of exercise anticipation involves a relatively standardized procedure, and it seems to evoke a response distribution with favorable characteristics. Further research is needed to examine the reliability of these responses and their association with other measures of stress-related CVR.
Cold pressor.
In addition to exercise, the cold pressor task is another "physical" stimulus used in reactivity research. Brief immersion of the hand or foot in ice water causes a rapid BP rise evoked by peripheral vascular mechanisms as well as by afferent signals involving pain and temperature (42). A cold pressor applied to the forehead is also associated with rapid vasoconstriction, this time accompanied by a bradycardic response characteristic of the diving reflex (43). Although cold pressor is thought of as a physical challenge, verbal reports of pain are positively associated with the BP change, and some have suggested thereby that psychological influences may be important determinants of BP responding (44). There is also evidence that instructions accompanying the cold pressor task may alter the responses evoked; when instructions emphasize the difficulty of the task and the effort required to tolerate discomfort, larger cardiovascular changes are produced by the task (45). This finding suggests that the anticipation of behavioral demands (and possibly the central command centers involved with sympathoadrenal activation under such circumstances) may play a role in cold pressor responses as well as in responses to cognitive tasks.
Although there is reason to suspect some common CNS determinants of responding across tasks, studies that have compared the cardiovascular effects of cognitive challenges with those associated with the cold pressor have shown mixed results: Some investigators demonstrate a significant correspondence between responses to the two types of tasks (46–48), whereas others have shown no such association (49, 50). When significant intertask associations do appear, responses to cognitive challenges tend to be more highly correlated with each other than with the cold pressor (eg, average intertask correlations for SBP reactivity of 0.65 vs. 0.38, respectively; see Table 4 in Ref. 48).
Should Tasks Producing Different Patterns of Responding Be Used to Assess Reactivity?
In the same way that individuals may differ with respect to their characteristic patterns of stress-related CVR, different types of tasks have been shown to evoke different patterns of acute change in cardiovascular performance. For example, exposure to shock avoidance has been shown to be generally associated with a cardiac pattern of response, whereas the mirror tracer task tends to evoke a vascular pattern of responding (13, 51). Two studies that have examined both individual and task effects on hemodynamic patterning have shown that one’s tendency to respond with cardiac or vascular responding persists across different types of tasks. Kasprowicz et al. (13), for example, found that subjects characterized as cardiac or vascular reactors to mental arithmetic could be similarly distinguished during a mirror tracing task, even though the latter task elicited, on average, larger peripheral vascular responses than did mental arithmetic across the sample as a whole.
Similar results were shown by Sherwood et al. (52), who examined a dyadic reaction time task with the subject as either a participant or an observer. They found that patterns of response differed in the two types of dyadic conditions, with the active (participant) task being associated with relatively larger changes in cardiac output and the passive (observer) condition eliciting larger changes in peripheral resistance. Once again, this study found that subjects maintained their characteristic response tendencies across tasks. Together these studies suggest that the use of tasks eliciting strong cardiac or vascular patterns of response does not present any particular advantage if the goal is to measure trait differences in CVR.
Should CVR Tasks Be Restricted to Those that Evoke Negative Emotion?
We have defined the domain of psychological challenge somewhat broadly, in a manner that does not require the use of assessment tasks that evoke strongly negative emotional states. This assumption is based, in part, on the evidence suggesting that factors altering psychological effort or engagement (eg, task difficulty and performance incentives) seem to be some of the most important determinants of the magnitude of task-related cardiovascular responding in the laboratory (eg, Refs. 53–55) and that altering these factors does not necessarily affect the emotional valence of a testing stimulus. We are unaware of any evidence suggesting that tasks that evoke strong negative emotions are more valid than less strongly evocative tasks as testing stimuli for assessing individual differences in CVR; indeed, responses to evocative tasks (eg, speaking about an anger-evoking topic) tend to be highly correlated with responses to less explicitly evocative performance tasks in the laboratory in terms of their ability to discriminate individual differences in cardiovascular responding (SBP reactivity for speech vs. reaction time r = 0.58 in one sample (see Table 2 in Ref. 47); r = 0.52 for blacks and 0.70 for whites (see Tables 2 and 3 in Ref. 48) in a replication sample).
Are Interpersonal Challenges Helpful?
Manipulations involving the social environment have been shown to affect the magnitude of cardiovascular responses to laboratory challenges. For example, experimenters who harass the subject or who pose an evaluative threat (55, 56) may enhance responding, whereas the presence of a friend or a supportive ally may reduce the cardiovascular response to an evaluative threat (57, 58). Social interaction, apart from a specific task, may also alter cardiovascular activity. Cardiovascular responding is enhanced by efforts to influence the opinions of another (59) or by social conflict (60). These findings, and the fact that daily stressors are often social ones (61), suggest that assessments of CVR should include social challenges along with cognitive or psychomotor tasks.
Social challenges in the laboratory pose special difficulties, however, because they are often more complex than cognitive tasks and involve a number of potential confounds, including interactions with the gender, race, or age of the participants (62). Moreover, the determinants of cardiovascular activation during social stressors may not necessarily differ from those during more exclusively cognitive tasks. Wright et al. (63) have argued that enhancements of cardiovascular activity during social evaluation may be due to the incentive properties associated with evaluation. Such properties may be modeled just as effectively using monetary payment without the use of social or interpersonal stimuli (whether the patterns of hemodynamic responding may differ under the influence of social or nonsocial incentives should be further investigated; see Refs. 64 and 65).1
It has been argued that social tasks in the laboratory may generalize to the natural environment more effectively than responses to cognitive stressors. Indeed social challenges seem to be superior to cognitive tasks in their ability to predict mean ambulatory BP levels (66). But there have been fewer efforts to predict stress-related responsiveness in the natural environment. The two available reports (67, 68) that have examined this question have provided no evidence that social tasks in the laboratory outperform nonsocial tasks in terms of their ability to predict responsiveness to naturally occurring social stressors. For example, Kamarck et al. (67) found that DBP responses to public speaking in the classroom were associated with DBP responses to cognitive tasks in the laboratory (r = 0.39, p < .01) but not with DBP responses to public speaking in the laboratory (r = 0.23, p > .10).
The added value of using interpersonal challenges to assess individual differences in CVR is therefore not proven, and additional studies are needed before we can conclude that social challenges are superior to cognitive ones at predicting reactivity in daily life. However, there are circumstances in which social challenges in the lab may be useful. For example, competition or harassment seem to be particularly effective in distinguishing cardiovascular responses of Type A or hostile individuals from their counterparts (69), so it may be helpful to use social tasks when evaluating these tendencies. In such an effort it is important to recognize that social cues may exert multiple influences on cardiovascular responding (58). More work is needed to characterize more clearly the dimensions of the interpersonal environment that may be uniquely relevant to the assessment of cardiovascular reactivity as well as the strategies that may optimize reliable assessment in this regard.
We have reviewed the data with respect to four types of task characteristics here, examining the extent to which each of these characteristics should be included in the domain of tasks evoking CVR to psychological challenge or, alternatively, the extent to which each may belong in a separate domain. Existing evidence suggests that tasks involving independent peripheral triggers of response, such as cold pressor and aerobic exercise, may yield distributions of individual difference that are only minimally similar to those associated with tasks involving exclusively psychological demands. Substantial differences in the nature of the responses evoked by so called "psychological" and "physical" stimuli in this literature should be acknowledged, and caution should be exercised in characterizing these various individual difference distributions as equivalent. On the other hand, existing data suggest that psychological tasks evoking different patterns of cardiovascular responding ("cardiac" vs. "vascular" tasks) may, nonetheless, elicit similar dimensions of individual difference and may be considered in the same domain of measurement.
With respect to the other two task characteristics we examined here, variations in felt emotion and the use of interpersonal stimuli, the data are mixed or are difficult to evaluate at present. We suggest that continued effort be devoted to developing standardized methods for evaluating individual differences in cardiovascular responding to social and "emotional" tasks for evoking CVR (70) and for evaluating the extent to which such stimuli, in fact, do evoke independent distributions of individual difference.
CONSTRUCT VALIDITY: CARDIOVASCULAR REACTIVITY AS A MEASURE OF AUTONOMIC OR SYMPATHOADRENAL FUNCTION
Whereas content validity is concerned with coverage of a content domain, construct validity involves the extent to which a test adequately reflects a presumed underlying theoretical construct. Interrelationships with other measures are sought to determine the extent to which the measurement in question fits within an existing body of theory or data (71). In a general sense, stress-related CVR is assumed to be a marker of autonomic or sympathoadrenal function (72). If CVR involves a dynamic interplay of central and peripheral inputs that arise during psychological challenge, however, this dimension of individual difference may be expected to show only small associations with standard measures of autonomic or sympathetic nervous system (SNS) activity, which do not involve the same types of CNS activation. In this section we examine the extent to which measures of CVR to psychological challenge contribute independent information when compared with traditional measures of autonomic or SNS influence. We will focus on 1) measures of parasympathetic influence on the heart, such as HR variability or respiratory sinus arrhythmia (RSA); 2) measures of baroreflex sensitivity, such as orthostatic challenges; and 3) measures of peripheral sympathetic or adrenomedullary activity, such as urinary catecholamines and adrenergic receptor sensitivity.
Parasympathetic Function
HR is known to be quite variable, even at rest; the extent of HR variability, especially in relation to respiration (with frequencies of about 0.15–0.4 Hz), has been shown to be mediated primarily by parasympathetic input to the heart (73). A small number of reports have examined the extent to which basal measures of parasympathetic function are correlated with CVR. The samples involved in these studies have been small, and the findings have been mixed to date.
Overall the findings suggest that this association may be shown more strongly in patient samples than in community samples. Two studies of patients with stable coronary artery disease have shown small but significant associations between the extent of cardiac vagal control, as assessed by RSA or frequency decomposition of HR variability, and blood pressure responses to laboratory tasks (Refs. 74 and 75; correlations of 0.30 and 0.34 with DBP reactivity, respectively). However, two comparable studies among healthy individuals have shown no significant relationships between RSA measures and hemodynamic reactivity (49, 76).
Alterations in the geometry of the heart and vessels in samples of patients with coronary heart disease may alter their responsiveness to neural inputs (77); this may explain the different correlates of HR variability in healthy samples as compared with post–myocardial infarction patient groups. It is possible that the small associations observed between CVR and HR variability in coronary heart disease samples may be attributable to "third factors," such as severity of disease. In any case, findings reviewed here do not suggest that measures of HR and BP reactivity are redundant with assessments of resting parasympathetic function.
Baroreceptor Sensitivity
The baroreceptor reflex is a negative feedback system acting to reduce BP variability, with the adaptive end of preserving constant transcapillary pressure for maintaining vital tissues (78). Baroreceptors are one important source of HR variability (73), and although this receptor system seems to be inhibited by acute mental stress (78), differences in the baroreceptors are a potentially important component of autonomic function that might be expected to place some limits on CVR.
The evidence to date, however, suggests that there may be little, if any, correspondence between baroreceptor sensitivity and stress-related CVR. In a large (N = 280) sample of healthy men, Fauvel et al. (79) showed that resting measures of baroreflex sensitivity (indexed by the slope of changes in HR after spontaneous changes in beat-to-beat measures of BP) were not related to measures of SBP changes during a presentation of the Stroop Color Word task. Unpublished results from another (N = 108) sample of healthy male college students also showed consistent findings, demonstrating no significant associations between pressor responses to upright tilt (responses thought to be mediated by baroreceptor function) and CVR to psychological challenge (80). In conclusion, existing evidence suggests that measures of CVR capture response determinants that are independent of those regulated by baroreceptor functioning.
Urinary Catecholamines
Because exaggerated cardiac reactivity has been shown to be associated with blood measures of sympathoadrenal output during mental stress (81), one might expect CVR to be related to cumulative markers of adrenal function, such as urinary catecholamines, as well. Surprisingly, there is very little published data addressing this question. Recently, McCaffery et al. (82) have shown that measures of cardiac output responding to a battery of laboratory tasks showed small correlations (r = 0.29) with urinary epinephrine measures in a sample of healthy young men. There were no significant associations between measures of vascular reactivity and urinary epinephrine or norepinephrine in this study. More information is needed to explore the extent to which CVR and cumulative catecholamine output can be seen as alternative measures of the same construct.
Adrenoreceptor Sensitivity
An additional measure of sympathoadrenal function that may be related to individual differences in CVR involves variations in the sensitivity or density of peripheral adrenergic receptors. Three studies have assessed the influence of ß2-receptor density using saturation binding techniques (83–85). All three have reported significant associations with HR reactivity to laboratory challenges, although in one case these associations were no longer significant after the initial 3 minutes of task responding (84) and in another case these effects were not reported separately (see below; Ref. 83).
In addition to receptor density, lymphocyte ß2-receptor sensitivity has also been examined as a correlate of reactivity. One study (86) showed that an indirect measure of receptor sensitivity (dose of isoproterenol required to raise HR by 25 beats/min) was associated with task-related HR responding on each of two tasks. A second study (83) showed that a combined measure of density and sensitivity (cyclic adenosine monophosphate response to isoproterenol in lymphocytes) accounted for 48% of the variance in HR response. A third study, using similar methods for evaluating receptor sensitivity (85), showed no significant effects. In short, evidence implicating the role of ß-adrenergic density and sensitivity in CVR is intriguing but preliminary at this point. Additional evidence is needed examining the peripheral determinants of individual differences in other parameters of responding, especially those involving predominantly vascular determinants and 
-adrenergic responding.
That receptor factors are relevant in understanding stress-related CVR is suggested, in part, by the evidence that ß-blocking pharmacological agents tend to eliminate HR changes and preejection period changes evoked by laboratory tasks (87, 88). Interestingly, ß-blockers seem to have minimal effects on BP reactivity (89), with some data suggesting that selective or unselective blockade may even increase vascular resistance (90) during mental stress. The absence of clear effects of ß blockade on BP responding may be due to the unopposed -adrenergic stimulation by epinephrine in the vasculature unmasked by ß blockade (88, 89) or by central counterregulatory controls (90). In any case, ß-adrenergic functioning should be expected to be more closely tied with differences in HR reactivity or changes in contractility (89) than with differences in peripheral vascular functioning. Less is known about the effects of -adrenergic blockade on cardiovascular reactivity, although there is some evidence that such a manipulation may alter forearm vascular resistance changes during psychological demands (91).
In sum, the evidence is modest, to date, that measures of CVR are redundant with standard measures of autonomic function. Small correlations are found between measures of HR variability and BP response, but these seem to be restricted to postmyocardial infarction patients, in whom autonomic functioning may be impaired. Baroreceptor sensitivity does not seem to predict responsiveness to mental stress. Modest associations are shown between HR or cardiac reactivity on one hand and urinary measures of catecholamines or ß-receptor function on the other. Measures of BP responding have not been well characterized on the basis of measures of adrenal or peripheral receptor functioning. Importantly, however, only a small number of published studies have addressed these topics to date.
In general these findings are consistent with our assumption that CVR to psychological challenge may reflect supramedullary determinants that are not captured by many autonomic function tests. For example, differences in the threshold or intensity of firing among hypothalamic nuclei during psychological challenge may be an important source of variability in CVR, and such differences may not be easily detectable by other standard measures of autonomic function. A couple of studies suggest that stress-related changes in preejection period, an index of sympathetic reactivity, may be associated with electrodermal lability (92, 93), a measure of sympathetic activation that has been shown to be associated with attention or arousal processes (94, 95). Only a few studies, however, have directly addressed the association between CVR to psychological challenge and corresponding activation in relevant CNS sites (96).
CRITERION VALIDITY: LAB-TO-LIFE GENERALIZABILITY AND ITS EVALUATION
The final section of this review examines what is known about the criterion validity of CVR, with a focus on its association with measures of response to stress during daily life. Criterion validity refers to the effectiveness of a test in predicting behavior, in this case response to challenges encountered outside the laboratory. Most of the models relating CVR to cardiovascular disease assume that reactivity shows some degree of generalizability to daily life. The strongest version of the reactivity hypothesis, for example, posits that exaggerated reactivity plays a causal role in disease risk. Such a model assumes that the effects of reactivity accumulate during the course of daily living, outside the laboratory. If individual differences in cardiovascular reactivity do not generalize beyond the lab, they could still serve as "risk markers" (97), but they become less plausible as a cause of disease.
Two strategies have been used to examine the lab-to-field generalizability of reactivity (98). One group of studies has compared laboratory responses with assessments of ambulatory BP, usually during un-selected periods of the day. A second group has compared responses in the laboratory to those occurring during discrete challenges in daily life (eg, public speaking, exams, dissertation defense). The results in both bodies of literature are inconsistent across tasks and parameters. Data summaries in this literature, for example, suggest that in the average study, only about 20% to 25% of the hypothesized comparisons are significant in the expected direction (see Table 1 in Ref. 98). Moreover, it is not uncommon to show significant associations between HR responses in the laboratory and BP responses in the natural environment, but to show no correspondence between HR responses across the two settings (eg, Ref. 99). Although such inconsistent associations might suggest that the reported findings are actually due to chance, they may well reflect methodological limitations instead.
In studies of laboratory tasks predicting ambulatory BP, for example, the mean and variability of daily BP are frequently used as the criterion response. Because there are many influences other than psychological demands affecting BP during daily life, such measures may not be comparable to assessments of laboratory reactivity. We should expect only those cardiovascular changes evoked by psychological demands in the ambulatory setting to be associated with laboratory measures of cardiovascular response, with no necessary lab-to-life correspondence for the ambulatory changes evoked by activity, posture, and other influences (97). Unfortunately, using the mean and variability of BP as a criterion standard does not allow us to examine the effects of daily life demands in isolation.
A multilevel modeling approach to ambulatory data can be designed to assess how ‘within-person’ fluctuations in ambulatory cardiovascular activity in response to the psychosocial demands of daily life may be moderated by ‘between-person‘ differences in laboratory-based CVR. Such an approach may be more effective than simply measuring ambulatory BP (ABP) means or variabilities to examine the correspondence between reactivity across laboratory and field settings. In a recent study, Kamarck et al. (100) assessed CVR in the laboratory and conducted 6 days of ambulatory monitoring in 335 older adults. Measures of current levels of ‘Task Demand’ and ‘Decisional Control’ were assessed from electronic diary reports completed after each ABP reading. The effects of activity, posture, and substance use on ABP were covaried in each case. Laboratory measures of SBP reactivity were unrelated to nonspecific ambulatory SBP mean or variability (r values = .01, .08, respectively, NS). However, periods of high ‘Task Demand’ or low ‘Decisional Control’ were associated with significantly larger SBP increases during the day among those who were more SBP reactive in the laboratory, when compared with those who were less reactive. High laboratory reactors showed elevated ABP readings, but only during demanding or uncontrollable periods of their daily lives, as indexed by electronic diary reports. These findings suggest that appropriate measurement of challenging situations, as well as control for extraneous determinants, may facilitate detection of laboratory-to-life correspondence of CVR using ABP (see also Ref. 101).
Comparable methodological limitations may also account for some of the modest results observed when laboratory responses are compared with responses to discrete life challenges. In most of the studies of this second type, laboratory and field reactivity are each measured in response to a single stimulus on a single occasion, or the response to several stimuli are examined in isolation. Such a strategy is prone to produce small results, because such measures are of limited reliability and may reflect only a limited content domain (see also Ref. 102). Because aggregation across multiple stimuli and occasions reduces the influence of unique situational variance, it may enhance the range of generalizability in a test score. For example, dispositional measures of personality are seldom associated with isolated actions, but they are reasonably valid when repeated observations and different relevant behaviors are used as criteria for prediction (103, 104).
In a recent study (67), students in a public speaking course were given a set of tasks in each of two laboratory sessions. They were then monitored during anticipation and delivery of public speeches in the classroom on two separate days. Correlations of HR, SBP, and DBP reactivity across settings were relatively low and insignificant when separate laboratory tasks and individual speech periods in the classroom were compared relatively low and insignificant (r values for HR, SBP, and DBP averaged 0.13–0.17). When scores within each of the settings were aggregated across task periods and testing sessions, however, significant correlations emerged for each of the three cardiovascular parameters (r values across settings = 0.26, 0.30, and 0.40 for SBP, HR, and DBP reactivity, respectively). Results suggest that some of the previous inconsistent findings in this literature may be explained by the use of single-task comparisons. One might anticipate that aggregation across a wider range of tasks should further enhance the magnitude of these results.
The literature on lab-to-life generalizability is characterized by inconsistent findings. One interpretation of these results is that laboratory measures may be irrelevant to daily life. As an alternative interpretation, we emphasize the methodological hurdles involved in examining this question, many of which may have not been adequately addressed in the previous literature. Preliminary evidence presented here suggests that more consistent findings may emerge when some of these methodological challenges are taken into account.
CONCLUSIONS AND SUGGESTIONS FOR FUTURE RESEARCH
We have documented recent progress in the conceptualization and measurement of CVR and have pointed to a number of areas in need of further investigation. Continued work in the measurement of reactivity is needed to facilitate consensus, insofar as method variability remains an important determinant of inconsistent findings in this literature. The following issues, in particular, merit further attention:
Preliminary evidence examining the hypothesized relationship between reactivity and disease risk presents us with a complex set of challenges. First, such evidence calls for thoughtful replication, which challenges us toward refinement of practical and replicable strategies for the assessment of CVR across a variety of different populations. Second, such evidence demands explanation, which challenges us to increase our understanding of the meaning of individual differences in CVR and their implications for disease risk. Our review suggests that we are making progress in grappling with both of these sets of challenges but that much work remains. With continued advances in the conceptualization and measurement of CVR, we predict substantial progress in the development of testable, biologically coherent models of psychosomatic processes and cardiovascular disease in the coming decade.
ACKNOWLEDGMENTS
We gratefully acknowledge the assistance of Richard Jennings and Peter Gianaros for their helpful comments on an earlier draft.
NOTES
1 We thank an anonymous reviewer for this suggestion. 
Received for publication June 5, 2001.
REFERENCES
This article has been cited by other articles:
 |
|
 |
|
  S. van Well, A. M. Kolk, and I. G. Klugkist Effects of Sex, Gender Role Identification, and Gender Relevance of Two Types of Stressors on Cardiovascular and Subjective Responses: Sex and Gender Match and Mismatch Effects Behav Modif, July 1, 2008; 32(4): 427 - 449. [Abstract] [PDF]
|
|
|
|
 |
|
 P. J. Gianaros, L. K. Sheu, K. A. Matthews, J. R. Jennings, S. B. Manuck, and A. R. Hariri Individual Differences in Stressor-Evoked Blood Pressure Reactivity Vary with Activation, Volume, and Functional Connectivity of the Amygdala J. Neurosci., January 23, 2008; 28(4): 990 - 999. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. F. Robles Stress, Social Support, and Delayed Skin Barrier Recovery Psychosom Med, October 1, 2007; 69(8): 807 - 815. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. V. Moseley and W. Linden Predicting Blood Pressure and Heart Rate Change With Cardiovascular Reactivity and Recovery: Results From 3-Year and 10-Year Follow Up Psychosom Med, November 1, 2006; 68(6): 833 - 843. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. M. Christian and C. M. Stoney Social Support Versus Social Evaluation: Unique Effects on Vascular and Myocardial Response Patterns Psychosom Med, November 1, 2006; 68(6): 914 - 921. [Abstract] [Full Text] |
