Skip main navigation
×

Greater Cardiovascular Responses to Laboratory Mental Stress Are Associated With Poor Subsequent Cardiovascular Risk Status

A Meta-Analysis of Prospective Evidence
Originally publishedhttps://doi.org/10.1161/HYPERTENSIONAHA.109.146621Hypertension. 2010;55:1026–1032

Abstract

An increasing number of studies has tested whether greater cardiovascular responses to acute mental stress predict future cardiovascular disease, but results have been variable. This review aimed quantitatively to evaluate the association between cardiovascular responses to laboratory mental stress and subsequent cardiovascular risk status in prospective cohort studies. We searched general bibliographic databases, PsycINFO, Web of Science, and PubMed, up to December 2009. Two reviewers independently extracted data on study characteristics, quality, and estimates of associations. There were 169 associations (36 articles) of stress reactivity and 30 associations (5 articles) of poststress recovery in relation to future cardiovascular risk status, including elevated blood pressure, hypertension, left ventricular mass, subclinical atherosclerosis, and clinical cardiac events. The overall meta-analyses showed that greater reactivity to and poor recovery from stress were associated longitudinally with poor cardiovascular status (r=0.091 [95% CI: 0.050 to 0.132], P<0.001, and r=0.096 [95% CI: 0.058 to 0.134], P<0.001, respectively). These findings were supported by more conservative analyses of aggregate effects and by subgroup analyses of the methodologically strong associations. Notably, incident hypertension and increased carotid intima-media thickness were more consistently predicted by greater stress reactivity and poor stress recovery, respectively, whereas both factors were associated with higher future systolic and diastolic blood pressures. In conclusion, the current meta-analysis suggests that greater responsivity to acute mental stress has an adverse effect on future cardiovascular risk status, supporting the use of methods of managing stress responsivity in the prevention and treatment of cardiovascular disease.

An estimated 17.5 million people died from cardiovascular disease (CVD) in 2005, representing 30% of deaths globally, implicating CVD as the leading cause of death worldwide.1 In addition to the well-established cardiovascular risk factors of family history, obesity, smoking, diabetes mellitus, and hypercholesterolemia, much effort has been devoted to the identification of other potential risk factors, including psychological stress. Two major research strategies have emerged for investigating the role of psychological stress in the development of CVD. The first is to carry out population-based assessments of associations between stress exposure and future CVD using standard observational epidemiological methods,2,3 but supply has limited information about the pathways involved. The second more mechanistic strategy is to measure cardiovascular responses to acute laboratory mental stress. Large, mental stress-induced cardiovascular responses, including both heightened reactivity and slow poststress recovery, are hypothesized to lead, more than time, to elevation of tonic blood pressure (BP) and the development of CVD.4–8

Heightened reactivity or delayed recovery are thought to be processes through which stress and other psychosocial factors, such as hostility, promote CVD risk.9,10 The advantage of psychophysiological stress testing is that sophisticated measures of cardiovascular functions can be carried out under standardized laboratory conditions. The method is limited to studies of short-term responses to acute stimuli that may lack ecological validity, so relationships with cardiovascular activity in everyday life are mixed.11 Nevertheless, these disadvantages are offset by being able to monitor or eliminate confounding factors and to manipulate stimuli experimentally, allowing the causal factors responsible for cardiovascular responses to be determined.9 The practical significance of acute stress responses has been questioned, so longitudinal follow-up studies have been conducted to establish whether people who show heightened acute stress responses are at greater risk for the development of CVD.

Over the past few decades, many studies of mental stress–induced cardiovascular response in relation to future CVD have been published, but the conclusions of reviews have been varied.5,12,13 This may be partly because clear distinctions have not been made between prospective studies and cross-sectional, retrospective, case-control or quasiprospective studies. Some reviews have also combined psychological challenges with physical stressors, such as the cold pressor, hand-grip, and treadmill exercises.14,15 Finally, there have been no recent reviews using meta-analytic techniques to quantify the extent to which cardiovascular response predicts cardiovascular status.5,13

The purpose of this systematic review was, therefore, to provide quantitative evidence for associations between cardiovascular response to laboratory mental stress and future cardiovascular risk status by analyzing all of the available prospective studies. We also aimed to address the question of whether stress reactivity and poststress recovery have different relationships with later CVD and whether associations differ in relation to factors such as sex, nature of stress tasks, and methodological quality (for fuller details, please see the online Data Supplement at http://hyper.ahajournals.org).

Methods

Data Sources and Searches

The present protocol is based on a widely recommended method for systematic reviews of observational studies16,17 (for details, please see the online Data Supplement).

Study Selection

On the basis of existing reviews of related literature,18 we defined an acute laboratory stressor as a task that lasted ≤1 hour and did not serve a function outside the laboratory setting. Criteria for inclusion were as follows: (1) publication as a peer-reviewed journal article in any language; (2) evaluation of an association of laboratory stress-induced cardiovascular response with future cardiovascular health status; (3) separate inclusion of samples if >1 construct of a laboratory stressor and >1 type of cardiovascular response were assessed in 1 article; (4) extraction of only effect sizes for psychological stress if physical stressors such as the cold pressor or hand grip, as well as psychological stress tasks, were used within 1 article and their effect sizes were separately presented; (5) separate inclusion of samples if the effects of laboratory stress-induced cardiovascular response on cardiovascular risk status were separately analyzed in men and women in 1 article; and (6) exclusion of the association with smaller sample size or poorer methodological quality if separate associations between stress responsivity and future cardiovascular status were reported in the same population sample in different publications.

Data Extraction and Quality Assessment

We extracted an r between the cardiovascular response predictor during stress reactivity and recovery and subsequent cardiovascular status. Cardiovascular predictors included systolic and diastolic BPs, heart rate (HR), cardiac output, cardiac index, myocardial ischemic responses, cardiac pre-ejection period,19 and HR variability20 (for details, please see the online Data Supplement).

We assessed the quality of all of the articles using a system for scoring methodological quality, on the basis of whether potential confounders were addressed, because these can contribute to biases associated with effect estimation. In light of previous evidence,21–23 we regarded an association as good quality if it revealed no statistical differences among cardiovascular response groups in age, sex, smoking, body mass index, biochemical risk factors for CVD (high density lipoprotein cholesterol, triglyceride, etc), the presence of medication affecting stress responses (antihypertensive drugs, etc), socioeconomic status, and baseline physiological activity, or if statistical adjustments were made for these factors, and if the association used consecutive or random recruitment of participants or representative sampling. We classified associations arbitrarily into high- or low-quality categories by whether they fulfilled ≥6 of these 8 criteria.

Data Synthesis and Analysis

We followed meta-analytic procedures that have been described previously elsewhere.18,24 Briefly, an effect size was the r between the cardiovascular response predictor and cardiovascular outcome or was calculated from the difference in subsequent cardiovascular status between the high and low cardiovascular response groups.25–28

Separate meta-analyses were carried out for associations of stress reactivity and stress recovery. If there was sufficient information available (≥4 associations), we aimed to carry out sensitivity analyses according to several characteristics. We simultaneously used the I2 statistic for homogeneity between studies.29 Finally, publication biases were estimated informally by a funnel plot and formally by using the Egger unweighted regression asymmetry test30 and the fail-safe number.31,32 All of the analyses were conducted by the Meta-Analysis Program33 (for details, please see the online Data Supplement).

Results

Summary of Associations

Figure 1 shows details of the flow diagram for this systematic review. The articles included in the meta-analysis are detailed in Table S1 (available in the online Data Supplement), whereas Table S2 summarizes the articles that were retrieved but excluded. The Table summarizes the characteristics of the 175 stress responsivity and cardiovascular risk status–related associations derived from the articles included in the review. These 175 associations were based on 31 cohorts and were published between 1986 and 2009. Most associations (72.6%) involved responses to cognitive tasks, such as mental arithmetic, mirror tracing, color-word interference (Stroop task), and so forth. The 3 major cardiovascular response predictors were systolic BP, diastolic BP, and HR (32.0%, 32.0%, and 17.1% of the associations, respectively), but some associations involved HR variability, pre-ejection period, cardiac output or cardiac index, or myocardial ischemic response. The most common cardiovascular outcomes were systolic and diastolic BP levels, which were evaluated in >30% of the associations, followed by carotid intima-media thickness and coronary calcification.

Figure 1. Flow diagram of systematic review (QUOROM statement flow diagram).

Table. Characteristics of the Associations Included in the Meta-Analyses

CharacteristicsEnrolled Associations
No.%
Total No. of stress responsivity and cardiovascular risk status–related associations175100
Sample size, average No.±SE329.9±49.9
Follow-up duration, ≥3 y9051.4
Laboratory stressors
    Cognitive task12772.6
    Stress interview137.4
    Public speaking116.3
    Emotion induction95.1
    Combined tasks158.6
Methodological quality score, 0 to 9, ±SD6.1±1.6
    Age details provided17499.4
    Sex details provided17197.7
    Smoking assessed or controlled8850.3
    Body mass index assessed or controlled14281.1
    Biochemical risk factors assessed or controlled6637.7
    Medication assessed or controlled13074.3
    Socioeconomic status assessed or controlled5732.6
    Baseline outcomes assessed or controlled175100.0
    Adequate sample recruitment5833.1
    Quality score ≥68950.9
Cardiovascular response predictors
    HR300.0
    Systolic BP5617.1
    Diastolic BP5632.0
    Mean arterial BP232.0
    Cardiac index or cardiac output71.1
    Myocardial ischemic response52.9
    Pre-ejection period84.6
    HR variability116.3
Cardiovascular outcomes
    Systolic BP5430.9
    Diastolic BP5330.3
    Mean arterial BP31.7
    Hypertension63.4
    Coronary calcification116.3
    Carotid intima-media thickness2212.6
    Carotid atherosclerotic plaques52.9
    Left ventricular mass63.4
    Cardiovascular disease events158.6

Stress Reactivity and Subsequent Cardiovascular Status

The overall meta-analyses found that greater stress reactivity was significantly associated with poor cardiovascular risk status longitudinally (r=0.091 [95% CI: 0.050 to 0.132]; P<0.001; Figure 2). This finding was confirmed by more conservative analyses of aggregate effects (r=0.129 [95% CI: 0.054 to 0.202]; P<0.001). Furthermore, the subgroup meta-analysis of methodologically strong associations (association quality score: ≥6) replicated this significant association between greater stress reactivity and poor future cardiovascular risk status (r=0.110 [95% CI: 0.050 to 0.170]; P<0.001). These analyses were not accompanied by significant publication bias using the Egger unweighted regression asymmetry test (P<0.10), whereas all showed very high heterogeneity (I2>0.90).

Figure 2. Stress reactivity and subsequent cardiovascular status: results of meta-analyses, subgrouping, and sensitivity analyses. CaC indicates coronary calcification; DBP, diastolic BP; HRV, HR variability; HT, hypertension; IMT, intima-media thickness; MIR, myocardial ischemic response; PEP, preejection period; SBP, systolic BP. CVD high-risk population included patients with cardiac diseases, atherosclerotic diseases, hypertension, and borderline hypertension. Bold text indicates P<0.05. *Significant publication bias by the Egger method (P<0.10).

Intriguingly, subanalyses showed that associations between the greater stress reactivity and future cardiovascular risk status seemed to be more pronounced in men (r=0.117), younger populations defined as age ≤18 years old (r=0.097), and those with longer follow-up periods (≥3 years; r=0.116). In the subanalysis by the types of stressor, only the cognitive task category remained significantly associated with later cardiovascular risk status (r=0.094). When separate analyses were carried out of different cardiovascular response predictors, associations with poor cardiovascular outcomes were demonstrated for systolic and diastolic BP reactivity (r=0.096 and 0.122, respectively). The most consistent cardiovascular outcomes related to greater stress reactivity were incident hypertension and higher systolic and diastolic BPs (r=0.101, 0.117, and 0.077, respectively).

Stress Recovery and Subsequent Cardiovascular Status

Poor stress recovery (defined as sustained cardiovascular activation above baseline levels during the posttask recovery period) was associated with impaired future cardiovascular risk status in the overall analyses (r=0.096 [95% CI: 0.058 to 0.134]; P<0.001; Figure 3). This association was replicated in both the subanalyses on the aggregate effects and the methodologically stronger associations (r=0.081 [95% CI: 0.009 to 0.151], P=0.027, and r=0.079 [95% CI: 0.041 to 0.118], P<0.001, respectively). These analyses were all accompanied by significant publication bias of the Egger unweighted regression asymmetry test (P<0.10). The fail-safe numbers in the overall effect, aggregate effect, and methodologically stronger associations were 183, 7, and 117, respectively. This implies that the association in the overall analysis was reasonably reliable.

Figure 3. Stress recovery and subsequent cardiovascular status: results of meta-analyses, subgrouping, and sensitivity analyses. DBP, diastolic BP; HRV, HR variability; HT, hypertension; IMT, intima-media thickness; SBP, systolic BP. Bold text indicates P<0.05. *Significant publication bias by the Egger method (P<0.10).

When we limited analyses to associations with follow-up periods ≥3 years, effects remained significant (r=0.082). We could not carry out other subanalyses on the basis of population characteristics, such as sex or age, because the number of associations was too small. When stressor types were tested separately, cognitive task associations exhibited significant associations between impaired stress recovery and future risk status (r=0.109). The analyses of different cardiovascular response predictors demonstrated that poor HR and systolic BP recovery from stress (r=0.114 and 0.130, respectively) were consistently associated with cardiovascular risk status. It is notable that, in analyses of separate cardiovascular outcomes, poor stress recovery was significantly associated with greater carotid intima-media thickness and higher systolic and diastolic BPs (r=0.140, 0.083, and 0.083, respectively).

Discussion

This is, to our knowledge, the first meta-analytic review to evaluate the association between cardiovascular responses to laboratory stress and later cardiovascular risk status. It indicates that greater reactivity to stress or slow recovery after mental stress predicts poor future cardiovascular status or progression of CVD risk. These effects were sustained in methodologically strong associations and in conservative analyses that tested associations with aggregated predictors within association populations.

Several mechanisms could relate cardiovascular response with progression of cardiovascular risk. Because the studies reviewed were observational rather than experimental, the direction of causality cannot be conclusively proven. It is possible that occult CVD led to heightened cardiovascular response so that previous disease status caused disturbances in responsivity rather than the reverse. In addition, underlying genetic, socioeconomic, or developmental processes may both stimulate alterations in cardiovascular response and accelerate CVD progression, without a direct connection between the two. Behaviors such as smoking, the use of medications, and factors like excess adiposity critically influence cardiovascular response to acute stress,21 while also exacerbating CVD status, so that they could theoretically be responsible for the effects observed. This latter explanation seems unlikely, because subanalyses showed that relationships between cardiovascular responses and subsequent risk status remained significant in studies that controlled for these behavioral factors.

If the causal chain is from heightened reactivity and/or impaired recovery to CVD risk, different pathways might again be involved. First, there could be a direct relationship between BP or hemodynamic responsivity and later CVD risk. Hyperreactive people may experience repeated episodes of elevated BP or disturbed hemodynamics as they go about their everyday lives, so that over the course of time their tonic BP will rise, eventually leading to incident hypertension.9 Our finding that effects were stronger in associations with longer follow-up periods is consistent with such a time-course hypothesis. However, the limited strength of effects indicates that not all individuals who show heightened reactivity or poor recovery experience accelerated progression of CVD risk. One important issue may be whether the individual is exposed to challenges in everyday life that elicit appropriate stress reactions. Light et al34 demonstrated that high stress reactivity predicted increases in tonic BP over a 10-year period only when associated both with a positive family history and with high levels of daily stress. Another possibility is that, over time, some people learn to cope more effectively with behavioral and emotional challenges, thereby reducing the frequency of hyperreactive episodes.

A second possibility is that the cardiovascular response is a marker of other physiological processes more directly involved in cardiovascular pathology. Inflammatory pathways and hemostatic responses are critical to atherogenesis and thrombus formation, and impaired poststress cardiovascular recovery is associated with prolonged hemostatic and inflammatory responses.35 Transient endothelial dysfunction is also elicited by acute stressors along with cardiovascular responses.36 Greater fibrinogen and interleukin 6 responses to acute stress are positively associated with increased ambulatory BP prospectively,37 whereas a recent study showed that arterial epinephrine and norepinephrine stress reactions predicted systolic BP over an 18-year period.38

We carried out subanalyses of categories of stressor and found that associations with future cardiovascular risk were shown only in the analyses of cognitive tasks. There may be a number of reasons for this. First, the number of associations involving cognitive tasks was much greater than for other stressor categories, so findings may be more robust. Second, the use of cognitive tasks may allow more effective manipulation of task characteristics, such as difficulty, controllability, and sustained effort, that are known to influence biological responses than is possible with other types of task. Third, the administration of cognitive tasks may be more readily standardized than is the case for tasks such as public speaking or interviews and may therefore elicit more reliable contingent responses.39

As noted in the Introduction, there is substantial literature relating cardiovascular response to the cold pressor with CVD risk and future hypertension. The results of these studies have been mixed.40–44 We excluded these studies on theoretical grounds, because the cardiovascular reactions to the cold pressor are driven in part by reflex hemodynamic changes rather than psychological stress effects. However, studies of the cold pressor and autonomic function are important, because they throw light on the regulatory processes implicated in CVD risk.45

The strength of the findings in this meta-analysis depends on the quality of the associations included. Many of the associations identified in the literature did not provide sufficient data for calculating hazard ratios or relative risk ratios. Hence, we used r as an effect size in this study. Effect sizes (r) in the region of 0.1 to 0.3 (or −0.1 to −0.3), 0.3 to 0.5 (or −0.3 to −0.5), and 0.5 to 1.0 (or −0.5 to −1.0) are generally considered to represent small, moderate, and large relationships, respectively, although this categorization is somewhat arbitrary and should not be observed too strictly.46 The interpretation of r depends on the context and purpose, but r can best be conceptualized in practical terms using a binomial effect size display.26 The binomial effect size display is most easily understood when the outcome is dichotomous, as it is for diagnosis or mortality. Using the binomial effect size display, we computed the probability of incident hypertension to increase by ≈23% (hazard ratio: 1.23), for individuals exhibiting greater compared with less stress reactivity in a binary division. The size of associations between stress reactivity and hypertension outcomes may appear small; however, it is worth pointing out that these effect sizes are not markedly different from those for many other biological indicators identified in prospective observational epidemiological research (eg, Reference 47).

Our review has several limitations. First, it was restricted to the evaluation of results in published articles. We found evidence of publication biases in the overall recovery analysis and in several subgroup analyses by the Egger unweighted regression asymmetry test. This implies a positive result bias, if authors are more likely to submit, or editors accept, positive than null (negative or inconclusive) results. However, in the analyses of stress recovery associations, the fail-safe number for the overall effect was 183, whereas it was 117 for the methodologically strong associations. This indicates that a large number of nonsignificant associations would have to be in the “file drawer” to negate the significance of the association. Second, the method of scaling cardiovascular response was inconsistent across associations, with some using continuous scores, whereas others involved binary, tertile, or quartile divisions. Third, few associations used consecutive or random recruitment of participants or representative populations, although these issues are important to observational studies.23 Fourth, we could not investigate the impact of some potentially important factors, such as family history or ongoing background stress, and their interaction with stress responsivity because of an insufficient number of associations addressing these issues. Finally, it should be recognized that laboratory studies of cardiovascular response model stress process acutely using artificial short-term stimuli. The magnitude of stress exposure is small in comparison with real life, but it is hypothesized that individuals who are reactive in the laboratory will also display greater cardiovascular response in everyday life.7,48

Perspectives

Evidence for the role of psychological distress and social/environmental adversity in hypertension and CVD comes from a variety of disciplines, including animal stress research, clinical studies of neural pathophysiology, and epidemiological studies of stress exposure and factors such as depression and hostility. Laboratory studies of physiological reactions to mental stress are becoming more sophisticated, with measurement of a broader range of neuroendocrine, inflammatory, and hemostatic variables, in addition to BP and HR, and assessment of general population samples rather than college students. Larger studies with longer tracking periods will permit inclusion of measures of stress exposure and coping during follow-up, allowing more robust analyses of the role of propensity to heightened stress responsivity. Few studies have yet been able to take account of genetic influences on cardiovascular responses or interactions with factors such as adiposity. Nevertheless, the current meta-analysis suggests that greater cardiovascular reactivity to and poor recovery after acute mental stress have adverse associations with future cardiovascular health. This research has implications both for the prevention of CVD risk progression and management of clinical conditions. A recent meta-analysis of randomized, controlled trials has documented the efficacy of psychological interventions in cardiac patients,49 whereas greater understanding of underlying physiological processes may permit the targeting of pharmacotherapy to reduce the impact of heightened stress responsivity.

We are grateful to colleagues in many research centers for providing the additional data required for meta-analysis.

Sources of Funding

This research was funded by the United Kingdom Medical Research Council (to Y.C.) and the British Heart Foundation (to A.S.).

Disclosures

None.

Footnotes

Correspondence to Yoichi Chida, Department of Medical Science, Happy Science Clinic, West Canyon II 3F, 1-12-20 Mizonoguchi, Takatsu-ku, Kawasaki City, Kanagawa 213-0023, Japan. E-mail

References

  • 1 World Health Organization. World Health Organization 2007 fact sheet on cardiovascular disease. Available at: http://www.who.int/mediacentre/factsheets/fs317/en/print.html. Accessed May 22, 2008.Google Scholar
  • 2 Kuper H, Marmot M, Hemingway H. Systematic review of prospective cohort studies of psychosocial factors in the etiology and prognosis of coronary heart disease. Semin Vasc Med. 2002; 2: 267–314.CrossrefMedlineGoogle Scholar
  • 3 Sparrenberger F, Cichelero FT, Adcoli AM, Fonseca FP, Weiss G, Berwanger O, Fuchs SC, Moreira LBe, Fuchs FD. Does psychosocial stress cause hypertension? A systematic review of observational studies. J Hum Hypertens. 2009; 23: 12–19.CrossrefMedlineGoogle Scholar
  • 4 Light KC. Cardiovascular responses to effortful active coping: implications for the role of stress in hypertension development. Psychophysiol. 1981; 18: 216–225.CrossrefMedlineGoogle Scholar
  • 5 Treiber FA, Kamarck T, Schneiderman N, Sheffield D, Kapuku G, Taylor T. Cardiovascular reactivity and development of preclinical and clinical disease states. Psychosom Med. 2003; 65: 46–62.CrossrefMedlineGoogle Scholar
  • 6 Linden W, Earle TL, Gerin W, Christenfeld N. Physiological stress reactivity and recovery: conceptual siblings separated at birth? J Psychosom Res. 1997; 42: 117–135.CrossrefMedlineGoogle Scholar
  • 7 Manuck SB. Cardiovascular reactivity in cardiovascular disease: “once more unto the breach.” Int J Behav Med. 1994; 1: 4–31.CrossrefMedlineGoogle Scholar
  • 8 Rutledge T, Linden W, Paul D. Cardiovascular recovery from acute laboratory stress: reliability and concurrent validity. Psychosom Med. 2000; 62: 648–654.CrossrefMedlineGoogle Scholar
  • 9 Steptoe A. Psychophysiological contributions to behavioral medicine and psychosomatics. In: Cacioppo JT, Tassinary LG, Bernston G, eds. Handbook of Psychophysiology. 3rd ed. New York, NY: Cambridge University Press; 2007: 723–751.Google Scholar
  • 10 Chida Y, Steptoe A. The association of anger and hostility with future coronary heart disease. J Am Coll Cardiol. 2009; 53: 936–946.CrossrefMedlineGoogle Scholar
  • 11 Kamarck TW, Lovallo WR. Cardiovascular reactivity to psychological challenge: conceptual and measurement considerations. Psychosom Med. 2003; 65: 9–21.CrossrefMedlineGoogle Scholar
  • 12 Schuler JHL, O'Brien WH. Cardiovascular recovery from stress and hypertension risk factors: a meta-analytic review. Psychophysiol. 1997; 34: 649–659.CrossrefMedlineGoogle Scholar
  • 13 Lovallo WR. Cardiovascular reactivity: mechanisms and pathways to cardiovascular disease. Int J Psychophysiol. 2005; 58: 119–132.CrossrefMedlineGoogle Scholar
  • 14 Lovallo W. The cold pressor test and autonomic function: a review and integration. Psychophysiol. 1975; 12: 268–282.CrossrefMedlineGoogle Scholar
  • 15 Arendt-Nielsen L, Lautenbacher S. Assessment of pain perception. In: Lautenbacher S, Fillingim RB, eds. Pathophysiology of Pain Perception. New York, NY: Kluwer Academic; 2004: 25–42.Google Scholar
  • 16 Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, Moher D, Becker BJ, Sipe TA, Thacker SB. Meta-analysis of observational studies in epidemiology. JAMA. 2004; 283: 2008–2012.Google Scholar
  • 17 Higgins JPT, Green S, eds. Cochrane Handbook for Systematic Reviews of Interventions, version 5.0.1 [updated September 2008]. The Cochrane Collaboration; 2008. Available at: http://www.cochrane-handbook.org.Google Scholar
  • 18 Chida Y, Hamer H. Chronic psychosocial factors and acute physiological responses to laboratory induced stress in the healthy populations: a quantitative review of 30 years of investigations. Psychol Bull. 2008; 134: 829–885.CrossrefMedlineGoogle Scholar
  • 19 Cacioppo JT, Berntson GG, Binkley PF, Quigley KS, Uchino BN, Fieldstone A. Autonomic cardiac control: II–noninvasive indices and basal response as revealed by autonomic blockades. Psychophysiol. 1994; 31: 586–598.CrossrefMedlineGoogle Scholar
  • 20 Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Heart rate variability: standards of measurement, physiological interpretation and clinical use. Circulation. 1996; 93: 1043–1065.CrossrefMedlineGoogle Scholar
  • 21 Kajantie E, Phillips DIW. The effects of sex and hormonal status on the physiological response to acute psychosocial stress. Psychoneuroendocrinology. 2006; 31: 151–178.CrossrefMedlineGoogle Scholar
  • 22 Laupacis A, Wells G, Richardson S, Tugwell P. Users’ guides to the medical literature: V–how to use an article about prognosis. JAMA. 1994; 272: 234–237.CrossrefMedlineGoogle Scholar
  • 23 von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med. 2007; 147: 573–577.CrossrefMedlineGoogle Scholar
  • 24 Lyubomirsky S, King L, Diener E. The benefits of frequent positive affect: does happiness lead to success? Psychol Bull. 2005; 131: 803–855.CrossrefMedlineGoogle Scholar
  • 25 Rosenthal R, Rubin DB. requivalent: a simple effect size indicator. Psychol Method. 2003; 8: 492–496.CrossrefMedlineGoogle Scholar
  • 26 Rosenthal R, DiMatteo MR. Meta-analysis: recent developments in quantitative methods for literature reviews. Annu Rev Psychol. 2001; 52: 59–82.CrossrefMedlineGoogle Scholar
  • 27 Follmann D, Elliott P, Suh I, Cutler J. Variance imputation for overviews of clinical trials with continuous response. J Clin Epidemiol. 1992; 45: 769–773.CrossrefMedlineGoogle Scholar
  • 28 Dersimonian R, Laird N. Metaanalysis in clinical-trials. Control Clin Trials. 1986; 7: 177–188.CrossrefMedlineGoogle Scholar
  • 29 Ioannidis J, Patsopoulos N, Evangelou E. Uncertainty in heterogeneity estimates in meta-analyses. BMJ. 2004; 335: 914–916.Google Scholar
  • 30 Egger M, Smith GD, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ. 1997; 315: 629–634.CrossrefMedlineGoogle Scholar
  • 31 Iyengar S, Greenhouse JB. Selection models and the file drawer problem. Stat Sci. 1988; 3: 109–135.CrossrefGoogle Scholar
  • 32 Scargle JD. Publication bias: the “file-drawer” problem in scientific inference. J Sci Explor. 2000; 14: 91–106.Google Scholar
  • 33 Masui K. Meta-analysis software user’s guide. In: Masui K, ed. Kokokara-Hajimeru Meta-Analysis. Tokyo, Japan: Shinko-Koueki Medical Press; 2003: 91–121.Google Scholar
  • 34 Light KC, Girdler SS, Sherwood A, Bragdon EE, Brownley KA, West SG, Hinderliter AL. High stress responsivity predicts later blood pressure only in combination with positive family history and high life stress. Hypertension. 1999; 33: 1458–1464.CrossrefMedlineGoogle Scholar
  • 35 Steptoe A, Marmot M. Psychosocial, hemostatic, and inflammatory correlates of delayed poststress blood pressure recovery. Psychosom Med. 2006; 68: 531–537.CrossrefMedlineGoogle Scholar
  • 36 Ghiadoni L, Donald AE, Cropley M, Mullen MJ, Oakley G, Taylor M, O'Connor G, Betteridge J, Klein N, Steptoe A, Deanfield JE. Mental stress induces transient endothelial dysfunction in humans. Circulation. 2000; 120: 2473–2478.Google Scholar
  • 37 Brydon L, Steptoe A. Stress-induced increases in interleukin-6 and fibrinogen predict ambulatory blood pressure at 3-year follow-up. J Hypertens. 2005; 23: 1001–1007.CrossrefMedlineGoogle Scholar
  • 38 Flaa A, Eide IK, Kjeldsen SE, Rostrup M. Sympathoadrenal stress reactivity is a predictor of future blood pressure: an 18-year follow-up study. Hypertension. 2008; 52: 336–341.LinkGoogle Scholar
  • 39 Dickerson SS, Kemeny ME. Acute stressors and cortisol responses: a theoretical integration and synthesis of laboratory research. Psychol Bull. 2004; 130: 355–391.CrossrefMedlineGoogle Scholar
  • 40 Matthews KA, Katholi CR, McCreath H, Whooley MA, Williams DR, Zhu S, Markovitz JH. Blood pressure reactivity to psychological stress predicts hypertension in the CARDIA study. Circulation. 2004; 110: 74–78.LinkGoogle Scholar
  • 41 Knox SS, Hausdorff J, Markovitz JH. Reactivity as a predictor of subsequent blood pressure: racial differences in the Coronary Artery Risk Development in Young Adults (CARDIA) Study. Hypertension. 2002; 40: 914–919.LinkGoogle Scholar
  • 42 Coresh J, Klag MJ, Mead LA, Liang KY, Whelton PK. Vascular reactivity in young adults and cardiovascular disease: a prospective study. Hypertension. 1992; 19: 218–223.Google Scholar
  • 43 Wood DL, Sheps SG, Elveback LR, Schirger A. Cold pressor test as a predictor of hypertension. Hypertension. 1984; 6: 301–306.LinkGoogle Scholar
  • 44 Carroll D, Davey Smith G, Willemsen G, Sheffield D, Sweetnam PM, Gallacher JE, Elwood PC. Blood pressure reactions to the cold pressor test and the prediction of ischaemic heart disease: data from the Caerphilly Study. J Epidemiol Community Health. 1998; 52: 528–529.CrossrefMedlineGoogle Scholar
  • 45 Mancia G, Grassi G, Ferrari AU. Reflex control of the circulation in experimental and human hypertension. In: Zanchetti A, Mancia G, eds. Handbook of Hypertension–Pathophysiology of Hypertension. Amsterdam, the Netherlands: Elsevier Science; 1997: 568–601.Google Scholar
  • 46 Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates; 1988.Google Scholar
  • 47 Danesh J, Wheeler JG, Hirschfield GM, Eda S, Eiriksdottir G, Rumley A, Lowe GDO, Pepys MB, Gudnason V. C-reactive protein and other circulating markers of inflammation in the prediction of coronary heart disease. N Engl J Med. 2004; 350: 1387–1397.CrossrefMedlineGoogle Scholar
  • 48 Steptoe A. Tools of psychosocial biology in health care research. In: Bowling A, Ebrahim S, eds. Handbook of Health Research Methods. Maidenhead, United Kingdom: Open University Press; 2005: 471–493.Google Scholar
  • 49 Linden W, Phillips MJ, Leclerc J. Psychological treatment of cardiac patients: a meta-analysis. Eur Heart J. 2007; 28: 2972–2984.CrossrefMedlineGoogle Scholar