BIOLOGICAL INDICATORS IN WHITEHALL II AND ELSA

The importance of biological markers can be illustrated by showing how they shed light on two issues of causation. Working with economists, we have found that the link between socioeconomic position and health has two major explanations. Public health scientists tend to start from the position that social circumstances associated with socioeconomic position lead to ill health. Economists tend to start from the position that health, or some resilience factor, leads to socioeconomic position. To sort between these competing positions, it is helpful to have further specification of a causal model. We, for example, posit that low social position is associated with increased exposure to psychosocial factors, which in turn activate the autonomic nervous system and the hypothalamic-pituitary-adrenal axis to influence metabolism and disease risk.

We do not, in other words, view the body as a black box with social conditions going in at one end and disease coming out the other or vice versa. Rather, we seek to gain evidence to support or refute our causal model. One major candidate to explain the social gradient in disease is the group of factors characterizing the metabolic syndrome: low high-density lipoprotein (HDL) cholesterol, high plasma triglyceride, high waist to hip ratio, and disturbances of insulin and glucose metabolism. To look at this in Whitehall II, we had to arrange for approximately 8,000 people to come to a clinic in a fasting state, have blood drawn, drink a standard glucose load, and have blood drawn again precisely two hours later—a mammoth undertaking. We performed this at the Phase 3 (5-year examination) of the Whitehall II cohort, repeated it at Phase 5 (10-year examination), and again at Phase 7 (15 years). We found that the metabolic syndrome showed a clear social gradient in men and women—lower grade, higher prevalence—making it a promising candidate to be a biological intermediary between social position and coronary heart disease (Brunner et al., 1997).

A further step in the causal model was to conduct a nested case-control study, the cases being people with the metabolic syndrome and the controls a sample from the Whitehall II cohort without. People with the metabolic syndrome had evidence of increased urinary output of metabolites of both cortisol and catecholamines, indicating more stress-related neuroendocrine activity (Brunner et al., 2002). We also examined heart rate variability to see if the metabolic syndrome was related to this indicator of adverse autonomic (sympathetic-parasympathetic) balance. We found that heart rate variability was less in cases of metabolic syndrome. Involvement of autonomic function was shown further by the finding that low employment grade was associated with low heart rate variability. The link to psychosocial factors was made explicit by the finding that people with low control at work had lower heart rate variability (high risk) than those with high control (Hemingway et al., 2005).

These findings shed light also on the question of confounding. When we showed, in Whitehall II, that low job control appeared to be an important link between low socioeconomic position and risk of coronary heart disease, we had to confront the question of whether low job control was confounded by some other causal exposure that was linked to socioeconomic position (Marmot et al., 1997). We have now shown that chronic work stress, measured as low control/high demands/low supports (known as isostrain), is related to the metabolic syndrome (Chandola, Brunner, and Marmot, 2006). This independent association completes an important link in the causal chain from chronic work stress to congenital heart disease and illustrates the strength of complementing observational epidemiological studies of social factors with biomarker assessment.

We are only now beginning to make use of the biological markers in ELSA. One example illustrates their usefulness. We performed a comparison of socioeconomic differences in morbidity in England (ELSA) and the United States (HRS). We were struck by the finding of a higher rate of morbidity in the United States, even among affluent groups in the two countries. The possibility of higher levels of diagnosis and reporting in the United States had to be considered. Our economist colleagues took the initiative to use the biological measures in ELSA and, for comparison as HRS has no biomarkers, took data from the National Health and Nutrition Examination Survey. The U.S. samples had higher levels of glycosylated hemoglobin, of plasma fibrinogen, of C-reactive protein, and of blood pressure. These results suggest that there is genuinely more illness in the United States. If we had an isolated finding of, say, higher levels of glycosylated hemoglobin in the United States, this would have raised the possibility of laboratory differences across countries. The facts that each of our biomarkers were adverse in the United States compared with England and each of the reported morbidity measures were higher make it likely that the intercountry differences are valid (Banks, Marmot, Oldfield, and Smith, 2006).

In our view these findings illustrate the value of working across disciplines and across “levels” of measurement, by which we mean economic, social, psychological, biological, and medical disease outcomes. Each discipline would have been impoverished without the input of the others in decisions of what to measure, how to measure them, how to analyze data, and interpretation.

A limitation of large-scale epidemiological studies that investigate biological pathways is the difficulty of examining the dynamics of biological markers. For example, if the interesting information will come from examining not the resting level of a biomarker but how it responds to a stress in the system, we may miss some relationships by using one measurement of a biomarker. It took a great deal of effort to measure how glucose and insulin responded to a glucose load administered in the fasting state. It would be difficult to repeat this by examining in 8,000 people how other biomarkers responded to a psychological stressor. To conduct this sort of investigation, smaller scale intensive studies are appropriate, although they have the drawback that they do not have the statistical power to examine disease end points.

WHITEHALL PSYCHOBIOLOGY STUDY

Psychobiological studies give us the opportunity to understand the connections between psychosocial factors and health-related biological responses in more detail than is possible in large-scale epidemiological research. There are two basic strategies used in psychobiological research. The first is psychophysiological or mental stress testing, in which biological responses are measured in response to standardized psychological or social stimuli (Steptoe, 2005). A wide range of challenging stimuli is employed, including cognitive and problem-solving tasks, simulated public speaking, and interpersonal conflict. These tests are usually carried out individually in the laboratory or clinic, with continuous monitoring of cardiovascular activity, sampling of saliva for the measurement of stress hormones like cortisol, and periodic blood sampling. The value of this experimental approach is that biological responses to psychosocial stimuli can be evaluated under environmentally controlled conditions, reducing many of the sources of confounding that might otherwise be present. Sophisticated measures can be employed that are difficult to assess in large-scale studies; for example, we have found that psychological stress stimulates transient disturbances of vascular endothelial function (thought to be important in the development of coronary atherosclerosis), up-regulation of the genes controlling release of inflammatory cytokines, and activation of blood platelets (Brydon et al., 2005; Ghiadoni et al., 2000; Steptoe et al., 2003c). The design of these studies does not assume that responses to these tasks are important per se, but rather that they represent the way in which people respond biologically to everyday challenges.

A more direct way of assessing responses in everyday life is to use a second strategy, namely ambulatory or naturalistic monitoring. The purpose of these methods is to investigate biological function in real life at work, at home, and in other situations, rather than when people are tested in the screening clinic or laboratory. We can also study the covariation between biology, activities, emotions, and behavior. The design depends on the availability of unobtrusive measures of biological activity in everyday life, with techniques depending on the health problem under investigation. Research into cardiovascular diseases primarily uses ambulatory monitoring devices for the measurement of blood pressure, heart rate, and heart rate variability. Research on stress often involves saliva sampling for the assessment of cortisol, while studies of musculoskeletal problems and pain utilize lightweight monitors of electromyographic activity (muscle tension).

The Whitehall psychobiology study used both mental stress testing and naturalistic monitoring to explore the dynamic relationships between the social gradient, psychosocial stress factors, and biological responses relevant to coronary heart disease. We focused on the hypothesis developed in the full Whitehall II study that biological stress responsivity is greater in lower than higher socioeconomic status groups and partly mediates differences in disease risk (Steptoe and Marmot, 2002). We recruited 129 middle-aged men and 109 women from the Whitehall II cohort to take part in this study. They were sampled from higher (n = 90), intermediate (n = 81), and lower (n = 67) grades of employment so that we could compare responses across the social gradient, and all the participants were in full-time paid employment. We excluded individuals with known coronary disease or diabetes, a task that was made simple by the detailed data collected from each person for more than a decade as part of the main Whitehall II study. The response rate was 55 percent. Participants attended a psychophysiological stress testing session and carried out ambulatory monitoring of blood pressure, heart rate, and salivary cortisol over a normal working day. Recruitment of the sample was relatively easy because participants were familiar with the Whitehall study setup. However, some of the people we asked did not take part since they thought it would be difficult to commit the time to this more intensive investigation.

The psychobiology study has generated interesting results relating psychosocial factors such as work stress, depression, and loneliness with biological function, but the main emphasis has been on socioeconomic position. Box 2-1 summarizes findings from the laboratory psychophysiological stress testing concerning socioeconomic position (Steptoe and Marmot, 2005). An intriguing result was that, by and large, the social gradient was not strongly related to the magnitude of biological responses to standardized tasks. However, lower socioeconomic status was characterized by delayed recovery or prolonged stress activation, so that biological responses remained elevated after tasks had been completed, rather than returning back to baseline levels promptly. For example, in comparison with high-status individuals, we calculated that the odds of impaired poststress recovery in the lower status group were 3.85 (95% CI, 1.48-10.0) for diastolic blood pressure, and 5.91 (CI, 1.88-18.6) for heart rate variability (Steptoe et al., 2002). As noted in Box 2-1, other biological measures did not show differential reactivity or recovery, but rather elevated levels in lower socioeconomic status participants throughout the protocol.

BOX 2-1

Psychophysiological Stress Testing and Socioeconomic Position: Summary of Findings from the Whitehall Psychobiology Study. Systolic and diastolic blood pressure Heart rate

These findings have led us to think again about the role that biological stress responses might play in mediating social gradients in cardiovascular disease risk. Two simple models come to mind. The first is that there are differences in the magnitude and duration of biological responses between social groups that have developed through a lifetime of differential psychosocial challenge. The chronic allostatic load model formulated by McEwen (1998) suggests that the wear and tear imposed on biological regulatory systems as they are forced to respond repeatedly to life's demands ultimately provokes reduced adaptability and the failure of systems to operate within optimal ranges. The association between lower socioeconomic position and prolonged activation after tasks have ended is a manifestation of this pattern. The second possibility is that adverse health effects arise out of differential exposure to adversity. The biological systems of lower status groups receive more “hits” on a daily basis because of greater exposure to chronic stressors, failures of coping, or reduced psychosocial and material resources. The higher levels of activation both during stress and at rest observed in several biological parameters listed in Box 2-1 will contribute to increased disease risk.

Naturalistic monitoring studies complement the psychophysiological stress testing results while demonstrating how these processes play out in everyday life. We have found that ambulatory blood pressure was greater in the lower than higher occupational grade groups in the mornings of working days, even after controlling for physical activity levels, body weight, smoking, and other factors (Steptoe et al., 2003b). Early morning blood pressure levels have special significance for cardiovascular disease since there is a blood pressure surge soon after waking, which coincides with an increased incidence of cardiac events (Giles, 2006). We also found that the cortisol awakening response, the rise in cortisol over the first 20-40 minutes after waking in the morning, is greater in lower status groups (Kunz-Ebrecht, Kirschbaum, Marmot, and Steptoe, 2004). A heightened cortisol awakening response is also found in people experiencing work stress, financial strain, and depressive symptoms, and it appears to be an indicator of stress-related hypothalamic-pituitary-adrenocortical dysfunction (Steptoe, 2007).

There are two important limitations to the psychobiological studies that we have carried out so far. The first is that, although the study was quite large from the perspective of laboratory science, the sample size was still limited for addressing socioeconomic issues. Some of the findings summarized in Box 2-1 were graded across the three occupational status groups, while for others it was necessary to combine higher and intermediate groups and compare them with the lower occupational status group. Second, we have not shown associations with cardiovascular disease outcomes. Much larger samples need to be tracked over many years to document a relationship between disturbed stress responsivity and coronary heart disease end points, such as myocardial infarction or cardiac death. However, newer methods of imaging the coronary arteries have given us the opportunity to study the progression of subclinical coronary disease. A fresh study with a larger sample that began in 2006 is measuring psychobiological responses along with coronary artery calcification quantified using nuclear imaging. We will repeat this imaging of the coronary arteries after three years to provide direct evidence of the clinical significance of these processes.

INTERCHANGE BETWEEN EPIDEMIOLOGICAL AND PSYCHOBIOLOGICAL STUDIES

The nesting of psychobiological studies within the broader framework of the Whitehall II study has provided fertile ground for the exchange of methods and measures. Although the Whitehall II study is rooted in observational, population-based epidemiology and the psychobiology study has its origins in laboratory psychophysiology, each has gained from the association. For example, laboratory psychophysiological research has generally paid rather little attention to the sampling framework. Much work in this tradition continues to be carried out with select samples (notably college students), assuming that findings in such unique groups can be generalized to the rest of the population. The sampling framework provided by the Whitehall II cohort has allowed us to recruit participants into psychobiological testing who are much more representative of different social groups.

Findings from the Whitehall II study also give pointers to biological markers that we think would be interesting to study in more detail in laboratory work. For example, an association between elevated plasma fibrinogen and lower social class was identified in a pilot for the Whitehall II study, and this was subsequently confirmed in Whitehall II and other cohorts in the United States and Europe (Brunner et al., 1996 ; Markowe et al., 1985). Fibrinogen is involved both in thrombogenesis and in the stimulation of atherogenic cell proliferation, and elevated levels predict the development of coronary heart disease. We therefore assessed fibrinogen responses to acute stress in the psychobiology study and found that it increases acutely in response to stress, with higher levels in people from lower grades of employment (Steptoe et al., 2003a). In addition, we have seen that men who experience low levels of control at work show heightened fibrinogen stress reactivity, suggesting that it may be particularly relevant to the processes through which chronic work stress increases cardiovascular disease risk. We have also found that heightened fibrinogen stress reactivity predicts increases in ambulatory blood pressure over a three-year period (Brydon and Steptoe, 2005), a result that is consistent with emerging evidence for the role of inflammation in the development of hypertension.

Experience in the psychobiology study has also informed the measures in the Whitehall II and ELSA studies. An instance is cortisol assessment. Cortisol has long been recognized as an important stress hormone, but it has been difficult to measure in large-scale studies because of its marked diurnal variation. This means that it cannot be adequately assessed in single blood samples, and collecting urine samples over the night or day is practically inconvenient and potentially aversive. Sampling of saliva has revolutionized cortisol assessment, since repeated samples can be collected over the day, stored in domestic refrigerators, and returned to the investigators by post. We included saliva sampling over a working and weekend day in the psychobiology study, and it proved acceptable to participants as well as scientifically valuable (Kunz-Ebrecht, Kirschbaum, and Steptoe, 2004; Steptoe et al., 2003b). This experience has encouraged us to include salivary cortisol assessment in a recent phase of the Whitehall II study and in the ELSA cohort. We have successfully obtained saliva samples over the day from more than 5,000 participants in ELSA, with a response rate of 89 percent. In Whitehall II, we have data from about 4,600 people, representing 90 percent of those eligible for cortisol assessment. Each person was asked to collect six samples over the day, and 99.45 percent of these were collected. These response rates are very encouraging, given that participants had to take measures in their own time at home and at work. This indicates that even measurement of biological variables that require some effort from study participants may be feasible on a large scale with appropriate management of data collection.

The continuing interchange between epidemiology and psychobiology has also led us to analyze data in relation to emerging issues that were not near the top of the agenda when the studies were designed. A good example is the work we have been carrying out on the biological correlates of positive well-being. The notion that affective well-being or happiness has beneficial health effects over and above the mere absence of depression and distress has evolved over recent years with the development of the field of positive psychology (Pressman and Cohen, 2005). We realized that arguments concerning the role of positive states in health would be greatly strengthened by an understanding of the underlying biology. We utilized the data from the naturalistic monitoring phase of the psychobiology study to address this issue. As part of that study, we obtained ratings of happiness every 20 minutes over the working day, at the same time that blood pressure readings were taken by the automated monitor. We aggregated these over the working day, then tested associations with biological measures relevant to health.

We found that happiness was negatively associated with cortisol sampled over the day (Steptoe, Wardle, and Marmot, 2005). This relationship was independent of other factors known to influence cortisol, such as age, gender, smoking, body mass index, and grade of employment. It was also apparent on a weekend as well as a working day, and it was replicated in the same individuals three years later, suggesting a stable pattern (Steptoe and Wardle, 2005). Another interesting observation is that stress-induced fibrinogen responses were smaller among happier individuals. Effects all remained statistically significant when psychological distress (measured with the General Health Questionnaire) was included in the analysis. The results therefore suggest that affective well-being has favorable effects on biological responses relevant to health independently of the influence of negative psychological states.

LIMITATIONS TO THE USE OF BIOLOGICAL INDICATORS

The measurement of biomarkers is expensive, and collection of the data can be time-consuming and demanding both for the investigators and participants. The fact that a particular measure has been shown in animal research to relate to disease development, or that it predicts an outcome in clinical samples, is not a sufficient reason to embark on its assessment in a population cohort. Some biomarkers may be particularly significant at a specific point in the course of disease, while not being relevant at other stages. Other measures may be attractive from the scientific perspective but difficult to collect in practice. For example, there are currently no simple direct measures of sympathetic nervous system activity that are suitable for large-scale population studies. Measures such as heart rate or heart rate variability are stimulated by the sympathetic nervous system, but they are also influenced by other factors, while sweat gland activity indexes responses only in the cholinergic sudomotor component of the sympathetic nervous system. Urinary measures of catecholamine metabolites are possible, but these require 24-hour collections, as we have done in a subset of the Whitehall II cohort (Brunner et al., 2002). Such techniques are difficult to use on a large scale. The inclusion of biological indicators in a population study therefore requires thorough consideration of the literature to determine the relevance and potential value of the marker, consideration of the practicality of measurement, and careful thought about confounders.

Another fundamental issue concerns the dynamics of biological measures. Some indicators, such as heart rate, blood pressure, and cortisol, change rapidly over minutes or even seconds, while others, like C-reactive protein, are much more stable. This presents problems for both population and laboratory studies. In population studies, a very labile measure is difficult to capture with one or two assessments, as we have pointed out in the last section in relation to cortisol. Disappointing results have sometimes emerged from studies of psychosocial factors that have used one or two values to assess the biological marker. A good example comes from research on job stress, in which a relationship between job strain (high demands, low-control work) and blood pressure has emerged much more consistently from studies involving repeated measures over the day than from investigations that have used conventional clinical measures of blood pressure (Steenland et al., 2000).

From the perspective of psychobiological studies, important relationships between psychosocial experiences and biology may be missed if measures are taken at the wrong time. For instance, the cytokine interleukin 6 (IL-6) takes one to two hours to respond to psychological stress (Steptoe, Willemsen, Owen, Flower, and Mohamed-Ali, 2001). If blood samples are drawn before and immediately after a series of challenging stimuli lasting 15-20 minutes, then no IL-6 response will be observed. It is also becoming clear that stimuli of different intensities may be needed to elicit responses in different variables. An analysis of the experimental literature on cortisol responses to distress has shown that increases are much greater when social-evaluative tasks are used: situations like simulated public speaking or carrying out mental arithmetic in front of an audience (Dickerson and Kemeny, 2004). We have typically used much less challenging stimuli in our studies, since we have been interested in conditions that more closely reflect everyday life, and under these circumstances cortisol responses are very modest.

Another issue to be borne in mind is that the levels of biological markers are determined by multiple factors, of which current psychosocial experiences are only one set. Genetic factors, health status, demographic factors such as gender, age, and ethnicity, health behaviors, and clinical risk factor profiles may all be relevant. Even in the psychosocial domain, early life experience may exert an influence independently of current circumstances. For example, an analysis of the Dunedin birth cohort has recently shown that maltreatment in childhood is a predictor of adult inflammatory responses (including C-reactive protein and fibrinogen), independently of other early life risk factors, health behaviors like smoking, and adult stress and socioeconomic position (Danese, Pariante, Caspi, Taylor, and Poulton, 2007). In the light of these diverse influences, it is wise to expect that associations between biological markers and current social and economic experience will be moderate rather than strong.

PRACTICAL AND ETHICAL ISSUES

CONCLUSIONS: HARD AND SOFT DATA

There is a widespread belief that questionnaire data are “soft” and laboratory data and biomarkers are “hard.” This is simplistic. First, they are neither hard nor soft but convey different information. Second, questionnaire data and biomarker data each have their own requirements for quality control. In the case of biomarkers, these requirements range from the circumstances under which the specimens are collected, to their mode of collection, to their handling, to the important issue of laboratory standardization. For all types of data, there are issues of reliability and validity, since all measurements have potential for error, bias, and contamination. In many ways, social scientists are more sensitive to these issues than some laboratory scientists, who appear to believe that the numbers produced by their instruments reflect the absolute truth. Social determinants of health are indeed hard to grasp, and there are many pitfalls in measurement and interpretation. But the integrated use of social epidemiological measures, biomarkers, and psychobiological approaches provides a means of triangulation on major problems of health and disease risk in an aging population.

ACKNOWLEDGMENTS

Michael Marmot is supported by a Medical Research Council professorship, and Andrew Steptoe is a British Heart Foundation professor. The Whitehall II study is supported by grants from the Medical Research Council; the British Heart Foundation; the Health and Safety Executive; the National Heart, Lung, and Blood Institute; the National Institute on Aging; the Agency for Health Care Policy Research; and the John D. and Catherine T. MacArthur Foundation Research Networks on Successful Midlife Development and Socio-economic Status and Health. ELSA is supported by the National Institute on Aging and by several British government departments, including the Department for Education and Skills; the Department for Environmental, Food, and Rural Affairs; the Department of Trade and Industry; the Department for Work and Pensions; HM Treasury Inland Revenue; and the Office for National Statistics. The psychobiology studies are supported by the Medical Research Council and the British Heart Foundation. We thank all the participants in these studies and all the members of the study teams.

REFERENCES

1. Adler NE, Ostrove JM. Socioeconomic status and health: What we know and what we don't. Annals of the New York Academy of Sciences. 1999;896:3–15. [PubMed: 10681884]
2. Banks J, Marmot M, Oldfield Z, Smith JP. Disease and disadvantage in the United States and in England. Journal of the American Medical Association. 2006;295:2037–2045. [PubMed: 16670412]
3. Brunner EJ, Davey Smith G, Marmot M, Canner R, Beksinska M, O'Brien J. Childhood social circumstances and psychosocial and behavioural factors as determinants of plasma fibrinogen. Lancet. 1996;347:1008–1013. [PubMed: 8606563]
4. Brunner EJ, Marmot M, Nanchahal K, Shipley MJ, Stansfeld SA, Juneja M, Alberti KG. Social inequality in coronary risk: Central obesity and the metabolic syndrome. Evidence from the Whitehall II study. Diabetologia. 1997;40:1341–1349. [PubMed: 9389428]
5. Brunner EJ, Hemingway H, Walker BR, Page M, Clarke P, Juneja M, Shipley MJ, Kumari M, Andrew R, Seckl JR, Papadopoulos A, Checkley S, Rumley A, Lowe GD, Stansfeld SA, Marmot MG. Adrenocortical, autonomic, and inflammatory causes of the metabolic syndrome: Nested case-control study. Circulation. 2002;106:2659–2665. [PubMed: 12438290]
6. Brydon L, Edwards S, Jia H, Mohamed-Ali V, Zachary I, Martin JF, Steptoe A. Psychological stress activates interleukin-1 beta gene expression in human mononuclear cells. Brain, Behavior, and Immunity. 2005;19:540–546. [PubMed: 16214025]
7. Brydon L, Steptoe A. Stress-induced increases in interleukin-6 and fibrinogen predict ambulatory blood pressure at 3-year follow-up. Journal of Hypertension. 2005;23:1001–1007. [PubMed: 15834286]
8. Chandola T, Brunner E, Marmot M. Chronic stress at work and the metabolic syndrome: Prospective study. British Medical Journal. 2006;332:521–525. [PMC free article: PMC1388129] [PubMed: 16428252]
9. Danese A, Pariante CM, Caspi A, Taylor A, Poulton R. Childhood maltreatment predicts adult inflammation in a life-course study. Proceedings of the National Academy of Sciences, USA. 2007;104:1319–1324. [PMC free article: PMC1783123] [PubMed: 17229839]
10. Dickerson SS, Kemeny ME. Acute stressors and cortisol responses: A theoretical integration and synthesis of laboratory research. Psychological Bulletin. 2004;130:355–391. [PubMed: 15122924]
11. Ghiadoni L, Donald A, 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;102:2473–2478. [PubMed: 11076819]
12. Giles TD. Circadian rhythm of blood pressure and the relation to cardiovascular events. Journal of Hypertension Supplement. 2006;24:S11–16. [PubMed: 16601555]
13. Hemingway H, Shipley M, Brunner E, Britton A, Malik M, Marmot M. Does autonomic function link social position to coronary risk? The Whitehall II study. Circulation. 2005;111:3071–3077. [PubMed: 15939818]
14. Kunz-Ebrecht SR, Kirschbaum C, Marmot M, Steptoe A. Differences in cortisol awakening response on work days and weekends in women and men from the Whitehall II cohort. Psychoneuroendocrinology. 2004;29:516–528. [PubMed: 14749096]
15. Kunz-Ebrecht SR, Kirschbaum C, Steptoe A. Work stress, socioeconomic status, and neuroendocrine activation over the working day. Social Science and Medicine. 2004;58:1523–1530. [PubMed: 14759695]
16. Markowe HL, Marmot M, Shipley MJ, Bulpitt CJ, Meade TW, Stirling Y, Vickers MV, Semmence A. Fibrinogen: A possible link between social class and coronary heart disease. British Medical Journal. 1985;291:1312–1314. [PMC free article: PMC1417417] [PubMed: 3933646]
17. Marmot M, Banks J, Blundell R, Lessof C, Nazroo J, editors. Health, wealth, and lifestyles of the older population in England. London, England: Institute of Fiscal Studies; 2003.
18. Marmot M, Bosma H, Hemingway H, Brunner E, Stansfeld S. Contribution of job control and other risk factors to social variations in coronary heart disease incidence. Lancet. 1997;350:235–239. [PubMed: 9242799]
19. Marmot M, Brunner E. Cohort profile: The Whitehall II study. International Journal of Epidemiology. 2005;34:251–256. [PubMed: 15576467]
20. Marmot M, Davey Smith G, Stansfeld S, Patel C, North F, Head J, White I, Brunner E, Feeney A. Health inequalities among British civil servants: The Whitehall II study. Lancet. 1991;337:1387–1393. [PubMed: 1674771]
21. Marmot M, Shipley MJ, Rose G. Inequalities in health: Specific explanations of a general pattern? Lancet. 1984;I:1003–1006. [PubMed: 6143919]
22. McEwen BS. Protective and damaging effects of stress mediators. New England Journal of Medicine. 1998;338:171–179. [PubMed: 9428819]
23. Mensah GA, Mokdad AH, Ford ES, Greenlund KJ, Croft JB. State of disparities in cardiovascular health in the United States. Circulation. 2005;111:1233–1241. [PubMed: 15769763]
24. Pressman SD, Cohen S. Does positive affect influence health? Psychological Bulletin. 2005;131:925–971. [PubMed: 16351329]
25. Steenland K, Fine L, Belkic K, Landsbergis P, Schnall P, Baker D, Theorell T, Siegrist J, Peter R, Karasek R, Marmot M, Brisson C, Tuchsen F. Research findings linking workplace factors to CVD outcomes. Occupational Medicine. 2000;15:7–68. [PubMed: 10620786]
26. Steptoe A. Tools of psychosocial biology in health care research. In: Bowling A, Ebrahim S, editors. Handbook of health research methods. Maidenhead, Berkshire, England: Open University Press; 2005. pp. 471–493.
27. Steptoe A. Cortisol awakening response. In: Fink G, editor. Encyclopedia of stress. second. Vol. 1. Oxford, England: Elsevier; 2007. pp. 647–653.
28. Steptoe A, Feldman PM, Kunz S, Owen N, Willemsen G, Marmot M. Stress responsivity and socioeconomic status: A mechanism for increased cardiovascular disease risk? European Heart Journal. 2002;23:1757–1763. [PubMed: 12419295]
29. Steptoe A, Kunz-Ebrecht S, Owen N, Feldman PJ, Rumley A, Lowe GD, Marmot M. Influence of socioeconomic status and job control on plasma fibrinogen responses to acute mental stress. Psychosomatic Medicine. 2003;65:137–144. [PubMed: 12554825]
30. Steptoe A, Kunz-Ebrecht S, Owen N, Feldman PJ, Willemsen G, Kirschbaum C, Marmot M. Socioeconomic status and stress-related biological responses over the working day. Psychosomatic Medicine. 2003;65:461–470. [PubMed: 12764220]
31. Steptoe A, Magid K, Edwards S, Brydon L, Hong Y, Erusalimsky J. The influence of psychological stress and socioeconomic status on platelet activation in men. Atherosclerosis. 2003;168:57–63. [PubMed: 12732387]
32. Steptoe A, Marmot M. The role of psychobiological pathways in socioeconomic inequalities in cardiovascular disease risk. European Heart Journal. 2002;23:13–25. [PubMed: 11741358]
33. Steptoe A, Marmot M. Socioeconomic position and coronary heart disease: A psychobiological perspective. In: Waite LJ, editor. Aging, health, and public Policy. New York: Population Council; 2005.
34. Steptoe A, Wardle J. Positive affect and biological function in everyday life. Neurobiology of Aging. 2005;26(Suppl 1):108–112. [PubMed: 16213629]
35. Steptoe A, Wardle J, Marmot M. Positive affect and health-related neuroendocrine, cardiovascular, and inflammatory processes. Proceedings of the National Academy of Sciences, USA. 2005;102:6508–6512. [PMC free article: PMC1088362] [PubMed: 15840727]
36. Steptoe A, Willemsen G, Owen N, Flower L, Mohamed-Ali V. Acute mental stress elicits delayed increases in circulating inflammatory cytokine levels. Clinical Science. 2001;101:185–192. [PubMed: 11473494]