Contribution of contemporaneous risk factors to social inequality in coronary heart disease and all causes mortality

Contribution of contemporaneous risk factors to social inequality in coronary heart disease and all causes mortality

Available online at www.sciencedirect.com R Preventive Medicine 36 (2003) 561–568 www.elsevier.com/locate/ypmed Contribution of contemporaneous ris...

174KB Sizes 0 Downloads 17 Views

Available online at www.sciencedirect.com R

Preventive Medicine 36 (2003) 561–568

www.elsevier.com/locate/ypmed

Contribution of contemporaneous risk factors to social inequality in coronary heart disease and all causes mortality Mark Woodward, B.Sc., M.Sc., Ph.D., C.Stat.,a,b,* Jane Oliphant, M.Sc.,a Gordon Lowe, M.D., FRCP, FFPHM,c and Hugh Tunstall-Pedoe, M.D., FRCP, FFPHMa a

Cardiovascular Epidemiology Unit, University of Dundee, Scotland Institute for International Health, University of Sydney, Australia c Department of Medicine, University of Glasgow, Scotland

b

Abstract Background. The relationship between low social status and premature mortality is well established, although the explanation for this link is unclear. This study explores the contribution to the social inequalities in coronary heart disease (CHD) and death of smoking status, cotinine, alcohol status, type A personality score, leisure activity, diabetes, systolic and diastolic blood pressure, body mass index, total and HDL cholesterol, triglycerides, fibrinogen, and vitamin C consumption. Methods. A random sample of 11,629 Scottish men and women, ages 40 –59 years, was recruited in 1984 –1987 and followed up for an average of 7.7 years for death and major coronary events. Social status was measured by housing tenure—renters being more socially deprived. Hazard ratios were computed from Cox models. Results. Adjusted for age, renters have 1.48 times the risk of CHD compared to owner– occupiers (95% CI: 1.21, 1.80) in men and 2.64 (1.89, 3.68) in women, and for all-cause mortality 1.55 (1.26, 1.90) and 2.12 (1.58, 2.84). The 14 risk factors explained 73% (men) and 77% (women) of the social differences in CHD. Equivalent figures for deaths were 51 and 64%. Conclusions. Fourteen contemporaneous risk factors, smoking being the most important, explain most of the social differential in CHD and death. © 2003 American Health Foundation and Elsevier Science (USA). All rights reserved. Keywords: Social class; Smoking; Coronary disease; Mortality

Introduction Lower socioeconomic status is strongly associated with increased coronary heart disease (CHD) prevalence [1,2] and incidence [1,3–7] and there is evidence of an increasing social differential, even as coronary rates overall are falling [3,8 –10] in the West. Attention has been devoted to possible explanations of social differences in coronary disease through the analysis of cardiovascular risk factors [1,11– 19]. For example, a 22-year cohort study in the United States [12] found that the three classical coronary risk factors (blood pressure, cholesterol, and smoking) were not * Corresponding author. Institute for International Health, University of Sydney, P.O. Box 576, Newtown, NSW 2042, Australia. Fax: ⫹2-93510008. E-mail address: [email protected] (M. Woodward).

sufficient to account for the observed relationship between social status and CHD. A similar study in Sweden [13] found a large effect of low occupation status after adjustment for the classical risk factors plus diabetes and history of myocardial infarction. The British Regional Heart Study [19] found that the attack rate of ischemic events for men in manual occupations exceeded that in nonmanual occupations by 44%, and that, although adjustments for smoking, blood pressure, obesity, and leisure time activity were able to lessen the social difference, it could not be fully explained by these risk factors. One American study [14] was able to account for socioeconomic differences in myocardial infarction, but only through adjustment for a large number of risk factors; furthermore this was a case– control study, and such studies are, generally, more susceptible to bias error than cohort studies. Some studies [12,13,17,18,20] have examined the impact of cardiovascular risk factors on

0091-7435/03/$ – see front matter © 2003 American Health Foundation and Elsevier Science (USA). All rights reserved. doi:10.1016/S0091-7435(02)00010-0

562

M. Woodward et al. / Preventive Medicine 36 (2003) 561–568

mortality from any cause, as well as from CHD alone. For example, the Whitehall Study [20] of over 11,000 U.K. male civil servants, found that adjustment for age, systolic blood pressure, cholesterol, smoking, and glucose intolerance made very little difference to the strong associations found between both employment grade and car ownership and all-cause mortality. As in the Whitehall Study, previous reports have broadly shown similar results for all-cause and CHD-specific mortality; that is, the cardiovascular risk factors explain only part of the social differential. If this is true, then there may be other factors, as yet unidentified or not quantified, which will explain the differential. Whether true or not, if those risk factors which contribute most to the excess risk among the most deprived can be identified, and ranked, then strategies for prevention may be better targeted. This article expands on previous work in the Scottish Heart Health Study (SHHS). We have shown that prevalent [2] and incident [21] CHD is related to housing tenure status (owner-occupiers or renters), this being a sensitive measure of social class in Scotland [2], where (at least when this study began) house renting was predominantly a feature of the socially disadvantaged. Housing tenure was the most strongly related to prevalent CHD of four personal measures of socioeconomic status: occupation, level of education, and years of education being the others [2]. We have also shown that renters have a considerably greater overall chance of premature death than do owner– occupiers in the cohort phase of the SHHS [21]. Here, we seek to explain the CHD incidence and allcause mortality differentials in housing tenure through proven cardiovascular risk factors. The analysis is applied separately to men and women, and therefore provides information, which is still relatively rare, on such social differentials among women, as well as showing what the two sexes had in common. The SHHS provides a particularly valuable data set in which to explore these issues, both because of the relatively high incidence of CHD in Scotland and the wide range of cardiovascular risk factors measured in the study.

Subjects and methods The SHHS involved the collection of baseline data from a random sample of 11,619 men and women of ages 40 –59 years in Scotland between 1984 and 1987. Subjects completed a questionnaire and attended a clinic for medical examination. The questionnaire solicited information on socioeconomic status, medical history, physical activity, smoking, alcohol consumption in the past week, and Bortner type A personality score [22]. A food frequency questionnaire was included [23]; this was subsequently validated in the Scottish population [24]. Height, weight, and blood pressure were measured, an electrocardiogram was recorded, and a sample of (nonfasting) venous blood was

Fig. 1. Path diagram showing the assumed relationship between housing tenure (as a measure of social class), coronary risk factors, and coronary heart disease. For explanations, see text.

taken. Plasma fibrinogen and serum triglycerides, total cholesterol, HDL cholesterol, and cotinine were determined. Cotinine is a metabolite of nicotine, and thus an objective measure of current tobacco smoke inhalation [25]. With their consent, follow-up of participants ran for an average of 7.7 years, to December 1993; 14 subjects were lost due to emigration. The Scottish NHS Register provided notification of all deaths with copies of death certificates, while the Scottish Common Services Agency identified emergency hospital admissions with a coronary diagnosis and cases of coronary artery surgery. CHD, as defined here, encompasses subjects who suffer definite or possible myocardial infarction (by WHO MONICA criteria, through case note review [26]), undergo coronary artery surgery, or who die from coronary disease, barring those subjects who have a nonfatal recurrence of a prerecruitment myocardial infarction. A more detailed explanation of the methodology of this study can be found elsewhere [21,27]. The causal model adopted for this article is illustrated by Fig. 1, which is a minor generalization of Fig. 4.2(c) of Woodward (1999) [28]. In this diagram, double-edged lines represent noncausal relationships and arrows show the direction of causal relationships which exist regardless of other factors in the diagram. Importantly, we assume that coronary risk factors act as potential confounding variables in the housing tenure–CHD relationship; we assume that housing tenure is not, itself, a consequence of any of the risk factors. From the range of variables collected in the SHHS, we chose 14 risk factors, all of which have been studied widely in cardiovascular epidemiology, to act as potential explanatory factors. These were cigarette smoking status (never/ex/current smokers), cotinine, alcohol consumption [nondrinker/light drinker (⬍22 units/week in men, ⬍15 units/week in women)/heavy drinker], Bortner type A personality score, leisure activity (inactive/average/active) at baseline, diabetes (yes/no), systolic blood pressure, diastolic blood pressure, body mass index (weight divided by the square of height), total cholesterol, HDL cholesterol, triglycerides, fibrinogen, and daily vitamin C consumption (estimated from the food frequency questionnaire and national standard portion sizes). Continuous variables were divided into fifths, to allow for nonlinear effects. Cox proportional hazards regression models were used to estimate hazard ratios for renters compared with owner– occupiers, adjusted for age and for the potential confounding factors in addition to age. Standard model checking

M. Woodward et al. / Preventive Medicine 36 (2003) 561–568

563

Table 1 Number (percentage) of deaths and coronary events by housing tenure status with hazard ratios (base ⫽ owner– occupier), by sex group Sex/outcome

Men CHD Death Women CHD Death

Housing tenure status

No. (%)

n

Hazard ratio (95% confidence interval) Age-adjusted

Multiple-adjusteda

Owner–occupiers Renters Owner–occupiers Renters

172 (6%) 229 (9%) 155 (5%) 226 (8%)

3067 2668 3067 2668

1 1.48 (1.21, 1.80) 1 1.55 (1.26, 1.90)

1 1.13 (0.85, 1.49) 1 1.27 (0.96, 1.69)

Owner–occupiers Renters Owner–occupiers Renters

48 (2%) 128 (4%) 65 (2%) 142 (5%)

3009 2841 3009 2841

1 2.64 (1.89, 3.68) 1 2.12 (1.58, 2.84)

1 1.37 (0.84, 2.23) 1 1.40 (0.91, 2.15)

a Adjusted for age, cigarette smoking status, cotinine, alcohol, type A behavior, leisure activity, diabetes, systolic and diastolic blood pressure, body mass index, total and HDL cholesterol, triglycerides, fibrinogen, and vitamin C.

procedures were followed [28], although not reported here. The percentage contribution of each individual factor, and of groups of factors, in addition to age, was measured as 100(HRage ⫺ HRage⫹confounders)/(HRage ⫺ 1)

(1)

where HR is the hazard ratio and the subscripts specify the explanatory variables used in the regression models. This method has been used previously in a similar context [17].

Results The housing tenure question was answered by 5735 of the 5754 men and 5850 of the 5875 women in the SHHS. Table 1 shows hazard ratios for relatively high compared to relatively low deprivation for both of the outcome variables, adjusted first for age and then additionally for the 14 risk factors. As reported previously [21], during the 7.7-year follow-up, men in rented accommodation were 1.48 times more likely to have a coronary event and 1.55 times more likely to die than male owner– occupiers, after accounting for age differentials. For women, the corresponding figures were 2.64 and 2.12. All four hazard ratios are statistically significant (P ⬍ 0.05). Multiple adjustment reduces the hazards ratios substantially, to between 1.13 and 1.40, and removes the statistical significance, albeit with wider confidence limits. Figs. 2 and 3 explore the contribution of each of the 14 risk factors to the reduction in hazard ratio. The risk factors are arranged in rank order within sex group for each outcome. Also shown are the percentage reductions in hazard due to the combination of the “classical” coronary risk factors: blood pressure (systolic and diastolic), smoking (cotinine and cigarette smoking status), and serum total cholesterol. This is compared with the total reduction due to all 14 risk factors and to the combination of risk factors which excludes the classical factors. Smoking is the clear leader in explaining the social differential: cotinine ranks

first twice, third, and fourth and cigarette smoking status ranks first, second twice, and third. Around 80% of the age-adjusted hazard ratio for CHD events, and 60% for all-cause mortality, is explained by the total of all 14 risk factors. Discussion Smoking is, predictably, an extremely important reason for the differences between deprivation groups in CHD and death from all causes: cigarette smoking and cotinine rank in the top four risk factors, for both endpoints in both sexes, for explaining social differentials in risk. However, it is interesting to note that whereas cigarette smoking status is a more important confounder for the effects of social differences than cotinine in men, the opposite is true in women. Since cigarette smoking status is a self-reported categorical measure of current or former smoking of cigarettes only, whereas cotinine represents an objective measure of current (baseline) consumption of any type of tobacco, the differences could be due to differential rates of quitting, deception, use of noncigarette tobacco or could be an artifact of the form of the variable and its distribution in the SHHS sample. Blood pressure has a moderate role in explaining social variation. Systolic blood pressure is more important than diastolic blood pressure in confounding social differences in both endpoints in women, whereas the two measures are equally predictive in men. HDL cholesterol is a more effective measure than total cholesterol in explaining the social differential in both men and women. This is consistent with a previous study [14], restricted to men, which found HDL cholesterol to be the most important variable in explaining the social class–CHD relationship. Indeed, in the current study, total cholesterol makes a small contribution toward widening the social gap when its effects are accounted for in CHD for women. A similar finding came, but in this case for men, from the first

564

M. Woodward et al. / Preventive Medicine 36 (2003) 561–568

Fig. 2. Percentage of the age-adjusted hazard ratio for a CHD event, for renters compared to owner– occupiers, explained by each risk factor alone, by the classical coronary risk factors (cigarette smoking status, cotinine, systolic and diastolic blood pressure, and total cholesterol) by other risk factors (alcohol consumption, Bortner type A personality score, leisure activity, diabetes, body mass index, HDL cholesterol, trigylcerides, fibrinogen, and vitamin C consumption), and by all 14 risk factors together. Numbers in brackets denote the rank of the age-adjusted hazard ratio for CHD of the variable concerned. The adjusted hazard ratios (corresponding to the percentages) show the hazard ratio for renters compared to owner– occupiers after adjustment for age and the risk factor(s) shown. The hazard ratio corresponding to zero percentage reduction is the age-adjusted hazard ratio for that sex group (see Table 1).

Whitehall study [1]. Another Scottish study [29] has reported lower plasma cholesterol in manual social class groups, for both men and women. Fibrinogen has an important association with social variation, as found in other studies [11,30,31]. It is the second most influential risk factor in male all-cause mortality, and the most important in male CHD. Its effect is not explained by smoking, since adjustment for cotinine leaves the fibrinogen and age-adjusted hazard ratio for housing tenure virtually unchanged (not shown here). In women, fibrinogen plays no role in explaining social differences in coronary disease, although it is important in explaining differences in mortality (ranked fourth). This could reflect the stronger association, in men than women, of fibrinogen with arterial disease, whether peripheral [32], carotid [33], or coronary [34], possibly due to differences in viscosity-increasing effects of fibrinogen between the sexes [32,33]. Fibrinogen appeared the strongest “nonclassical” risk factor in explain-

ing the social differential in CHD for men, and of mortality for men and women. Whether this reflects a causal role for fibrinogen, or fibrinogen is a marker of chronic inflammation (due to many possible environmental exposures) in lower social class groups, is not yet known [35,36]. Type A personality (by Bortner score) explains a substantial component of the social differential in CHD for women, and in mortality for men, but is less important otherwise. This sex difference is consistent with our earlier analyses from the SHHS [21]. Triglycerides play a larger explanatory role in social differences in CHD than in mortality for men, although the opposite is true in women. Vitamin C is effective in explaining the social class effect on coronary disease in men, but otherwise has no role in the current context. Diabetes has no role in explaining social variation in CHD or all-cause mortality, but this could be due to the low number of cases in this study. Alcohol status and body mass index, similarly, have little effect.

M. Woodward et al. / Preventive Medicine 36 (2003) 561–568

565

Fig. 3. Percentage of the age-adjusted hazard ratio for all-causes mortality, for renters compared to owner– occupiers, explained by each risk factor alone, by the classical coronary risk factors (cigarette smoking status, cotinine, systolic and diastolic blood pressure, and total cholesterol) by other risk factors (alcohol consumption, Bortner type A personality score, leisure activity, diabetes, body mass index, HDL cholesterol, trigylcerides, fibrinogen, and vitamin C consumption), and by all 14 risk factors together. Number in brackets denote the rank of the age-adjusted hazard ratio for all-causes mortality of the variable concerned. The adjusted hazard ratios (corresponding to the percentages) show the hazard ratio for renters compared to owner– occupiers after adjustment for age and the risk factor(s) shown. The hazard ratio corresponding to zero percentage reduction is the age-adjusted hazard ratio for that sex group (see Table 1).

The “classical” coronary risk factors (smoking, blood pressure, and total cholesterol) combined provide a better explanation of social differences in endpoints in women than in men. Indeed, nonclassical risk factors have more effect than classical risk factors for men. The comparison between these two groups of risk factors is very coarse because the nonclassical group is much larger, but this does illustrate the inability of the established CHD risk factors to explain social inequalities, as found by several other authors. The hazard ratios for CHD and death after adjustment for all the risk factors considered here are greater than unity, suggesting some residual positive effect of house renting (low social class). The four hazard ratios (two endpoints by two sex groups) all have 95% confidence intervals which straddle unity, so that the explanation of a chance residual association cannot be ruled out. On the other hand, the two female endpoints show an estimated excess hazard of about

40%, which is considerable. As Fig. 2 and 3 illustrate, several of the 14 variables have similar confounding effects, which is to be expected due to the tendency for risk profiles to overlap in human populations. Hence small differences in rankings are explainable by sampling variation. Sampling variation may be an explanation for the “rogue” finding that fibrinogen has no effect on the social differential in CHD, but has a large effect for total mortality, among women. All our risk factors were measured on a single occasion, so no account may be taken of regression dilution or withinsubject variation [37]. Variables which are measured most precisely are, all else being equal, more likely to have greater explanatory power. We can speculate that “usual,” rather than single values of risk factors would show quite different rankings in terms of relative ability to explain social differences. Unpublished data, for people within the same age range as the current study, from the Australasian cohorts included in the first wave of the Asia Pacific Cohort

566

M. Woodward et al. / Preventive Medicine 36 (2003) 561–568

Studies Collaboration [38] give regression dilution coefficients [39] of 0.9 for BMI, 0.8 for HDL cholesterol, 0.7 for total cholesterol, 0.6 for triglycerides, and 0.5 for diastolic and systolic blood pressure. Thus, as would be expected, BMI is the most reproducible of these six continuous risk factors, but the estimated linear relationship between the log hazard ratio and baseline systolic blood pressure is expected to underestimate the true slope, against “usual” systolic blood pressure, by a factor of 2. Given the crucial role of blood pressure in cardiovascular disease, regression dilution bias error is the most likely reason for its lack of importance in explaining the social differential in this study. A similar argument could be made for the relative lack of effect of total cholesterol compared to other lipids. Conceivably, all of the residual effect of housing tenure, after full adjustment, could be due to risk factor measurement error in its widest sense. As well as measuring risk factors only once, we only know current social status at baseline; we have no measure of social status in previous life. This is unlikely to be a real issue, since it is reasonable to suppose that most people will have maintained their relative status from childhood, within this study population. It has been suggested that early life experiences may affect future cardiovascular risk [40 – 42], and this, if true, would mean that social status could potentially drive risk factor levels, making the type of adjustments for confounding performed here inappropriate [28]. However, a recent review [43] has cast serious doubt on the validity of the major tenant of the “fetal origins” hypothesis, which is thus still open to debate. Comparing the ranking of the 14 variables as risk factors in their own right (as published earlier [21]) to their effect as confounders for housing tenure, for CHD the most striking difference is for total cholesterol. Although very important (most important for men) as a risk factor in its own right, it has little effect as a confounder for men and a negative effect for women. Smoking is more important as a confounder than as a risk factor, although it is important in both roles. By and large, relatively important risk factors are also relatively important confounders. For death from all causes, the only consistent differences in ranks between the sexes are for HDL cholesterol, which is more important as a confounder for social class, and for diabetes, which is more important as a risk factor. This article has concentrated upon housing tenure as a measure of social class, for comparison with our earlier report [2], and because this is the “best” measure in the current context [2,21,44], not least because it allocates study participants into two fairly equal-sized groups, leading to relative statistical efficiency compared to other measures [28]. Housing tenure was recognized as a measure of social status in the United Kingdom for many years, until government policies were introduced to promote purchases of properties by local authority tenants which has led to a considerable redistribution of owner-

occupation in the past 15 years. However, there are several other measures of social status [45], some of which are available from the SHHS, but which generally give weaker relationships of the same qualitative nature. To illustrate this, we have made comparative calculations using two of the most common ways of defining social status: occupational and educational classifications. The age-adjusted (with multiple-adjusted, as in Table 1, in parentheses) male hazard ratios for CHD are 1.16 (1.10) for manual vs non-manual workers and 1.44 (1.12) for those with less than 10 years of full-time education vs those with more, to be compared with 1.48 (1.13) for housing tenure ( Table 1). For women, corresponding results are 2.50 (1.55) and 1.52 (1.29), to be compared with 2.64 (1.37) for tenure. We did not ask questions about income in SHHS, because we did not anticipate accurate responses. In the Kuopio Ischemic Heart Disease Risk Factor Study, using the same methods as in this article, adjustment for a wide range of medical, biological, and psychosocial variables removed more than 60% of the effect, on incident myocardial infarction and allcause mortality, of a composite variable involving income, level of demands, and resources available in the workplace [46].

Conclusion Smoking explains around 40% of the social dimension of CHD and death in this study; smoking avoidance should go a long way toward removing the social differential in health among middle-age people. Tobacco control should, thus, be targeted at the most deprived. For men, as in the Caerphilly and Speedwell studies [11], fibrinogen is also important, but the other major coronary risk factors, blood pressure and total cholesterol, as well as obesity, have little role in explaining social differentials in CHD. Overall findings are very similar for women, except that, for them, fibrinogen is not an important confounder in the social class–CHD relationship.

Acknowledgments The SHHS has been funded by the Chief Scientist Office of the Department of Health of the Scottish Executive and by the British Heart Foundation. Jane Oliphant was funded by the Department of Health in London.

References [1] Rose G, Marmot MG. Social class and coronary heart disease. Br Heart J 1981;45:13–19.

M. Woodward et al. / Preventive Medicine 36 (2003) 561–568 [2] Woodward M, Shewry MC, Smith WCS, Tunstall-Pedoe H. Social status and coronary heart disease: results from the Scottish Heart Health Study. Prev Med 1992;21:136 – 48. [3] Marmot MG. Socioeconomic factors in cardiovascular disease. J Hypertens 1996;55:S201–5. [4] Office of Population Censuses and Surveys. Registrar General’s decennial supplement on occupational mortality 1979 – 80, 1982– 83. London: HMSO 1986. [5] Morrison C, Woodward M, Leslie W, Tunstall-Pedoe H. Effect of socioeconomic group on incidence of, management of, and survival after myocardial infarction and coronary death: analysis of community coronary event register. BMJ 1997;314:541– 6. [6] McLoone P, Boddy FA. Deprivation and mortality in Scotland, 1981 and 1991. BMJ 1994;309:1465–70. [7] Salomaa V, Niemela M, Miettinen H, et al. Relationships of socio-economic status to the incidence and prehospital, 28-day, and 1-year mortality rates of acute coronary events in the FINMONICA myocardial infarction register study. Circulation 2000; 101:1913–18. [8] Kawachi I, Marshall S, Pearce N. Social class inequalities in the decline of coronary heart disease among New Zealand men, 1975– 1977 to 1985–1987. Int J Epidemiol 1991;20:393– 8. [9] Lang T, Ducimetiere P. Premature cardiovascular mortality in France: divergent evolution between social categories from 1970 to 1990. Int J Epidemiol 1995;24:331–9. [10] Marmot MG, McDowall ME. Mortality decline and widening social inequalities. Lancet 1986;ii:274 – 6. [11] Baker IA, Sweetnam PM, Yarnell JWG, Bainton D, Elwood PC. Haemostatic and other risk factors for ischaemic heart disease and social class: evidence from the Caerphilly and Speedwell studies. Int J Epidemiol 1988;17:759 – 65. [12] Bucher HC, Ragland DR. Socio-economic indicators and mortality from coronary heart disease and cancer: a 22-year follow-up of middle-aged men. Am J Public Health 1995;85:1231– 6. [13] Engstrom G, Tyden P, Berglund G, Hansen O, Hedblad B, Janzon L. Incidence of myocardial infarction in women. A cohort study of risk factors and modifiers of effect. J Epidemiol Community Health 2000; 54:104 –7. [14] Hebert PR, Buring JE, O’Connor GT, Rosner B, Hennekens CH. Occupation and risk of nonfatal myocardial infarction. Arch Intern Med 1992;152:2253–7. [15] Holme I, Helgeland A, Hjerman I, Leren P, Lund-Larsen PG. Fouryear mortality by some socioeconomic indicators: The Oslo Study. J Epidemiol Community Health 1980;34:48 –52. [16] Luepker RV, Rosamond WD, Murphy R, Sprafka JM, Folsom AR, McGovern PG, Blackburn H. Socioeconomic status and coronary heart disease risk factor trends. The Minnesota Heart Survey. Circulation 1993;88:2172–9. [17] Lynch JW, Kaplan GA, Cohen RD, Tuomilehto J, Salonen JT. Do cardiovascular risk factors explain the relation between socioeconomic status, risk of all-cause mortality, cardiovascular mortality, and acute myocardial infarction? Am J Epidemiol 1996;144:934 – 42. [18] Pekkanen J, Tuomilehto J, Uutela A, Vartiainen E, Nissinen A. Social class, health behaviour, and mortality among men and women in eastern Finland. Br Med J 1995;311:589 –93. [19] Pocock SJ, Shaper AG, Cook DG, Phillips AN, Walker M. Social class differences in ischaemic heart disease in British men. Lancet 1987;2:197–201. [20] Davey Smith G, Shipley MJ, Rose G. Magnitude and causes of socioeconomic differentials in mortality: further evidence from the Whitehall Study. J Epidemiol Community Health 1990;44: 265–70. [21] Tunstall-Pedoe H, Woodward M, Tavendale R, A’Brook R, McCluskey M-K. Comparison of the prediction by 27 different factors of coronary heart disease and death in men and women of the Scottish heart health study: cohort study. BMJ 1997;315:722–9.

567

[22] Bortner RW. A short rating scale as a potential measure of pattern A behaviour. J Chron Dis 1969;22:87–91. [23] Yarnell JWG, Fehily AM, Milbank JE, Sweetnam PM, Walker CL. A short dietary questionnaire for use in an epidemiological survey. Comparison with weighed dietary records. Human Nutrition: Applied Nutrition 1983;37:103–12. [24] Bolton-Smith C, Smith WCS, Woodward M, Tunstall-Pedoe H. Nutrient intakes of different social class groups: Results from the Scottish Heart Health Study (SHHS). Br J Nutr 1991;65:321–35. [25] Woodward M, Tunstall-Pedoe H, Smith WCS, Tavendale R. Smoking characteristics and inhalation biochemistry in the Scottish population. J Clin Epidemiol 1991;44:1405–10. [26] Tunstall-Pedoe H, Kuulasmaa K, Amouyel P, Arveiler D, Rajakangas A-M, Pajak A. for WHO MONICA Project. Myocardial infarction and coronary deaths in the World Health Organization MONICA Project: registration procedures, event rates and case–fatality rates in 38 populations from 21 countries in four continents. Circulation 1994;90:583– 612. [27] Smith WCS, Crombie IK, Tavendale R, Irving JM, Kenicer MB, Tunstall-Pedoe H. The Scottish Heart Health Study: objectives and development of methods. Health Bull 1987;45:211–17. [28] Woodward M. Epidemiology: study design and data analysis. Boca Raton: Chapman & Hall/CRC, 1999. [29] Davey Smith G, Hart C, Watt G, Hole D, Hawthorne V. Individual social class, area-based deprivation, cardiovascular disease risk factors, and mortality: the Renfrew and Paisley study. J Epidemiol Community Health 1998;52:399 – 405. [30] Markowe HLJ, Marmot MG, Shipley MJ, et al. Fibrinogen: a possible link between social class and coronary heart disease. BMJ 1985;291: 1312–14. [31] Rosengren A, Wilhelmsen L, Welin L, Tsipogianni A, Teger-Nilsson A-C, Wedel H. Social influences and cardiovascular risk factors as determinants of plasma fibrinogen concentration in a general population sample of middle aged men. BMJ 1990;300:634 – 8. [32] Fowkes FGR, Pell JP, Donnan PT, Housley E, Lowe GDO, Riemersma RA, Prescott RJ. Sex differences in susceptibility to etiologic factors for peripheral atherosclerosis—importance of blood viscosity and plasma fibrinogen. Arteriosclerosis Thrombosis 1994;14:862–5. [33] Lee AJ, Mowbray PI, Lowe GDO, Rumley A, Fowkes FGR, Allan PL. Blood viscosity and elevated carotid intima-media thickness in men and women: the Edinburgh Artery Study. Circulation 1998;97: 1467–73. [34] Lee AJ, Lowe GDO, Woodward M, Tunstall-Pedoe H. Fibrinogen in relation to personal history of prevalent hypertension, diabetes, stroke, intermittent claudication, coronary heart disease and family history: the Scottish Heart Health Study. Br Heart J 1993;69:338 – 42. [35] Lowe GDO, Rumley A. Use of fibrinogen and fibrin D-Dimer in prediction of arterial thrombotic events. Thrombosis Haemostasis 1999;82:667–72. [36] Lowe GDO. Fibrinogen: an independent coronary risk factor. In: Betteridge DJ, editor. Lipids and vascular disease. London: Dunitz, 2000. [37] MacMahon S, Peto R, Cutler J, et al. Blood pressure, stroke and coronary heart disease. Part I, Prolonged differences in blood pressure: prospective observational studies corrected for regression dilution bias. Lancet 1990;335:765–74. [38] Asia Pacific Cohort Studies Collaboration. The effects of diabetes mellitus on the risks of major cardiovascular diseases and death in the Asia-Pacific region. Diabetes Care 2003;26:360 – 6. [39] Clarke R, Shipley M, Lewington S, et al. Underestimation of risk associations due to regression dilution in long-term follow-up of prospective studies. Am J Epidemiol 1999;150:341–53. [40] Barker DJP. Mothers, babies, and disease in later life. London: BMJ Publishing Group, 1994.

568

M. Woodward et al. / Preventive Medicine 36 (2003) 561–568

[41] Gunell DJ, Davey Smith G, Frankel S, et al. Childhood leg length and adult mortality: follow-up of the Carnegie (Boyd Orr) survey of diet and health in pre-war Britain. J Epidemiol Community Health 1998; 52:142–52. [42] Brunner E, Shipley MJ, Blane M, Davey Smith G, Marmot MG. When does cardiovascular risk start? Past and present socioeconomic circumstances and risk factors in adulthood. J Epidemiol Community Health 1999;53:757– 64. [43] Huxley R, Neil A, Collins R. Unravelling the fetal origins hypothesis: is there really an inverse association between birthweight and subsequent blood pressure? Lancet 2002;360:659 – 65.

[44] Woodward M. Small area statistics as markers for personal social status in the Scottish heart health study. J Epidemiol Community Health 1996;50:570 – 6. [45] Lynch J, Kaplan G. Socioeconomic Position. In: Berkman L, Kawachi I, editors. Social epidemiology. New York: Oxford University Press, 2000. [46] Lynch J, Krause N, Kaplan GA, Tuomilehto J, Salonen JT. Workplace conditions, socio-economic status, and the risk of mortality and acute myocardial infarction: the Kuopio Ischemic Heart Disease Risk Factor Study. Am J Public Health 1997;87: 617–22.