Macleod and Davey Smith suggest that the association that we found between psychological characteristics, particularly obsessionality and the somatic symptoms of anxiety on the one hand and fatal IHD on the other [1] were due to confounding by socio-economic factors. They suggest that we should have shown estimates before and after adjustment for social class.

We did discuss the adjustment for social class in the paper and pointed out this had relatively little effect on the association. We now show the results unadjusted, adjusted for other potential confounding factors and after adjustment for these factors and social class to demonstrate this point.

	Phobic anxiety	Obsessionality	Somatic complaint	Total
Unadjusted
RR (95% CI) P value	1.06 (0.99–1.14) p=0.08	1.10 (1.04–1.16) p=0.001	1.14 (1.08–1.21) p=0.0001	1.26 (1.09–1.45) p=0.002
Adjusted for age, smoking, BMI, fibrinogen, factor VII, sys BP and cholesterol
RR (95% CI) P value	1.07 (0.99–1.15) p=0.08	1.08 (1.02–1.15) p=0.007	1.09 (1.02–1.15) p=0.007	1.28 (1.09–1.50) p=0.002
Adjusted for age, smoking, BMI, fibrinogen, factor VII, sys BP, cholesterol and social class
RR (95% CI) P value	1.07 (0.99–1.15) p=0.10	1.08 (1.02–1.15) p=0.009	1.08 (1.02–1.15) p=0.009	1.28 (1.09–1.50) p=0.003

RR = Relative Risk of fatal IHD for a one point increase in each sub scale of the Crown Crisp Experiential Index and for a 10 point increase in the total score from Cox Model for fatal IHD.

Although the scores on a number of the sub-scales - phobic anxiety, obsessionality / obsessional neurosis and functional somatic complaint (physical symptoms of anxiety) - were significantly higher in those of lower social class the magnitude of the association was relatively small. For example, as we stated in the paper, the median score on the obsessionality /obsessional neurosis sub scale was 6 for men in social class ! and 7 for men in social class V. This is not to deny that poverty is a pervasive cause of ill-health or that public policies to reduce inequalities in health and wealth are to be commended [2]

All epidemiological studies are open to the possibility that associations may be due to residual confounding and ours in no exception. However, we think that socioeconomic differences are unlikely to be the explanation of the associations found in this case. Ultimately though the debate can only be settled by intervention studies. On this point both we and Macleod and Davey Smith are in agreement.

Andrew Haines
London School of Hygiene and Tropical Medicine
Jackie Cooper , T.W.Meade
MRC Epidemiology and Medical Care Unit , St Bartholomew's and the Royal London School of Medicine and Dentistry

References (1) Haines A, Cooper, Meade TW Psychological characteristics and fatal ischaemic heart disease. Heart 2001;85:385-9

(2) Haines A. Heath I, Smith R. Joining together to combat poverty. BMJ 2000, 320: 1-2

Editor,

Haines and colleagues present a further example from observational epidemiology of an association between a psychosocial factor and cardiovascular health (1). They suggest that this association is likely to be causal - a suggestion apparently accepted by the popular press in their reporting of this research (2).

However, given the non-specificity of the association between a range of such factors - broadly those with a negative social connotation - and an equally diverse group of pathological outcomes (3,4,5), some scepticism seems appropriate. There are alternative explanations for these relationships. The authors do not discuss an important one of these. In their study, as with every published study that has reported such relationships, adverse exposure in relation to the psychosocial factor was associated with general social disadvantage. Adverse social position is associated with poorer health thus a confounded, non-causal association between any factor associated with adverse social position and poorer health is to be expected (6). Furthermore since it is accepted that all indices of social position, such as occupational class, are relatively crude, adjustment for them in multivariate models may have little influence (7,8). In their paper, Haines and colleagues do not show estimates before and after adjustment for social class. Instead they show fully adjusted estimates and since these are still conventionally significant conclude that effects are independent of the adjustment factors.

There is a political attractiveness to psychosocial, as opposed to material, explanations for social inequalities in health. Such explanations put responsibility at the level of the individual - in this case for being too fussy - and distract from the role of economic policy in creating material inequality (9).

In view of the importance of this issue we agree with the authors' final paragraph. Observational epidemiology has taken us as far as it can in relation to this question. Future research should concentrate on experimental studies. If psychosocial interventions are truly a promising strategy to improve cardiovascular health then this should be demonstrated in randomised controlled trials.

Yours sincerely,

JOHN MACLEOD Department of Primary Care and General Practice University of Birmingham, B15 2TT

GEORGE DAVEY SMITH Department of Social Medicine University of Bristol, BS8 2PR

References

1. Haines A, Cooper J, Meade TW. Psychological characteristics and fatal ischaemic heart disease. Heart 2001;85:385-389.

2. Meikle J. Fussing can double risk of heart attack. Guardian 2001, March 15.

3. Kauhanen J, Kaplan GA, Cohen RD, Julkunen J, Salonen JT. Alexthymia and risk of death in middle-aged men. Journal of Psychosomatic Research 1996;41:541-549.

4. Bosma H, Schrijvers C, Mackenbach JP. Socioeconomic inequalities in mortality and importance of perceived control: cohort study. BMJ 1999;319:1469-70.

5. Everson SA, Goldberg DE, Kaplan GA, Cohen RD, Pukkala E, Tuomilehto J, Salonen JT. Hopelessness and Risk of Mortality and Incidence of Myocardial Infarction and Cancer. Psychosomatic Medicine 1996;58:113- 121.

6. Davey Smith G, Dorling D. "I'm all right John": voting patterns and mortality in England and Wales, 1981-92. BMJ 1996;313:1573-77.

7. Davey Smith G, Harding S. Is control at work the key to socio- economic gradients in mortality? (letter) Lancet 1997;350:1369-70.

8. Phillips AN, Davey Smith G. How independent are independent effects? Relative risk estimation when correlated exposures are measured imprecisely. J Clin Epidemiol 1991;44:1223-31.

9. Lynch JW, Davey Smith G, Kaplan GA, House JS. Income inequality and mortality: importance to health of individual income, psychosocial environment, or material conditions. BMJ 2000; 320: 1200-1204.