161
aldosterone is responsive to angiotensin and hybrid steroids are not in excess, the mutation may be different, perhaps causing excessive responsiveness to angiotensin (eg, double dose of the promoter region of CYPIIB2). Clinically, we are suggesting that primary aldosteronism is a commoner cause of hypertension than previously suspected, and better screening tests may lead to earlier diagnosis with the normokalaemic variety already becoming common. In patients with efficient antihypertensive mechanisms, a normotensive form may be identifiable by abnormal aldosterone/renin ratio. Such normotensive forms are likely to be found more frequently in relatives of patients with primary aldosteronism. Identification of the genetic abnormalities will permit detection, by genetic screening, of those predisposed to primary aldosteronism, and measures such as a low-salt diet might delay or prevent the development of hypertension and hypokalaemia. In collaboration with groups in Australia and elsewhere we are currently examining the DNA from our patients with primary aldosteronism, seeking mutations in GYP 11 Bl, CYPllB2, and other candidate genes. We are also seeking evidence of disturbed aldosterone biosynthesis in the relatives of patients with primary aldosteronism. REFERENCES 1. Lifton RP,
Dluhy RG, Powers M, et al. A chimaeric 11 &bgr;-hydroxylase/ aldosterone synthase gene causes glucocorticoid-remediable aldosteronism and human hypertension. Nature 1992; 355: 262-65. 2. Sutherland DJA, Ruse JL, Laidlaw JC. Hypertension, increased aldosterone secretion and low plasma renin activity relieved by dexamethasone. Can Med Assoc J 1966; 95: 1109-19. 3. Matsuo K, Kawai K, Tsuchiyama H, Veki V. Glucocorticoidsuppressible hyperaldosteronism: ultrastructural observation of a case.
Acta Pathol Jpn 1985; 35: 1511-19. 4. Miura K, Yoshinaga K, Goto K, et al. A case of glucocorticoid-responsive
hyperaldosteronism.J Clin Endocrinol 1968; 28:
1807-15.
5. Gordon RD, Stowasser M, Tunny TJ, Klemm SA, Finn WL, Krek AL. Clinical and pathlogical diversity of primary aldosteronism including a new familial variety. Clin Exp Pharmacol Physiol 1991; 18: 283-86. 6. Woodland E, Tunny TJ, Hamlet SM, Gordon RD. Hypertension corrected and aldosterone responsiveness to renin-angiotensin restored by long-term dexamethasone in glucocorticoid-suppressible hyperaldosteronism. Clin Exp Pharmacol Physiol 1985; 12: 245-48. 7. Gomez-Sanchez CE, Gill JR, Ganguly A, Gordon RD. Glucocorticoidsuppressible aldosteronism: a disorder of the adrenal transitional zone. J Clin Endocrinol Metab 1988; 67: 444-48. 8. Stowasser M, Gordon RD, Tunny TJ, Klemm SA, Finn WL, Krek AL. Primary aldosteronism: implications of a new familial variety. J Hypertension 1991; 9 (suppl 6): S264-65. 9. Stowasser M, Gordon RD, Tunny TJ, Klemm SA, Finn WL, Krek AL. Familial hyperaldosteronism type II: five families with a new variety of primary aldosteronism. Clin Exp Pharmacol Physiol 1992; 19: 319-22. 10. Hoefnagels WHL, Kloppenborg PWC. Hazards of long-term dexamethasone treatment in primary aldosteronism. N Engl J Med 1982; 306: 427. 11. Gordon RD, Hamlet SM, Tunny TJ, Klemm SA. Aldosteroneproducing adenomas responsive to angiotensin pose problems in diagnosis. Clin Exp Pharmacol Physiol 1987; 17: 175-79. 12. Gordon RD, Gomez-Sanchez CE, Hamlet SM, Tunny TJ, Klemm SA. Angiotensin-responsive aldosterone-producing adenoma masquerades as idiopathic hyperaldosteronism (IHA, adrenal hyperplasia) or low renin essential hypetension. J Hypertension 1987; 5 (suppl 5): S103-06. 13. Tunny TJ, Gordon RD, Klemm SA, Cohn D. Histological and biochemical distinctiveness of atypical aldosterone-producing adenomas responsive to upright posture and angiotensin. Clin Endocrinol 1991; 34: 363-69. 14. Axelrod L, Vickery AL (Jr). Case Records of the Massachusetts General Hospital: a 52-year-old man with hypertension, hypokalemia, and an adrenal mass. N Engl J Med 1992; 326: 1617-23. 15. Hiramatsu K, Yamada T, Yukimura Y, et al. A screening test to identify aldosterone-producing adenoma by measuring plasma renin activity: results in hypertensive patients. Arch Intern Med 1981; 141: 1589-93. 16. Hamlet SM, Tunny TJ, Woodland E, Gordon RD. Is aldosterone/renin ratio useful to screen a hypertensive population for primary aldosteronism? Clin Exp Pharmacol Physiol 1985; 12: 249-52. 17. McKenna TJ, Sequeira SJ, Heffernan A, Chambers J, Cunningham S. Diagnosis under random conditions of all disorders of the renin-
angiotensin-aldosterone axis, including primary hyperaldosteronism. J Clin Endocrinol Metab 1991; 73: 952-57.
VIEWPOINT Should
medication be stopped for exercise testing?
antianginal
dispute the contribution of exercise electrocardiography or radionuclide ventriculography to the management of ischaemic heart disease. Reports of their value were based largely on tests done after withdrawal of antianginal drug therapy for several hours or days.12 This practice stems from the observation that anti-anginal drugs, in particular beta-blockers, can mask otherwise detectable abnormalities—ie, ST-segment depression during exercise electrocardiography or a drop in left ventricular ejection fraction during exercise radionuclide ventriculography.3,4 Few would
Two
points need
be taken into account in the of the value of exercise testing. First, atheromatous coronary disease can be diagnosed only by coronary angiography. Exercise tests simply indicate the functional effect of the obstruction and thus the probability of the presence and extent of the lesion. Secondly, although such testing may also yield prognostic information, the diagnostic value of the test may differ from its prognostic value. In patients in whom there is doubt about the clinical diagnosis, unstable angina or myocardial infarction are
assessment
to
unlikely to occur on withdrawal of therapy. In practice, instability is usually encountered only in patients receiving several drugs for chronic angina, in whom the likelihood of coronary disease is already high. This observation argues against drug withdrawal for testing in such patients, since doing so may not only provoke instability but also be of little value in management. In someone with a typical history, an abnormal test after drug withdrawal is unsurprising. However, a test in the same patient that shows persistent ischaemia despite medication may induce a sense of urgency.
For example, to determine whether early invasive investigation and treatment are indicated after myocardial infarction, it seems more sensible to undertake prognostic testing with the patient on rather than off any medication for secondary prevention. When a patient with typical angina is rendered symptom-free by drugs, testing with the patient on Department of Cardiology, St Bartholomew’s Hospital, London EC1A 7BE, UK (R. Lim, MRCP, D. S. Dymond, FRCP). Correspondence to Dr Richard Lim.
ADDRESS:
162
such medication may give some idea of prognosis while the treatment is continued and whether urgent referral for angiography is needed. When angiography indicates that medical treatment is as justifiable as revascularisation, the result of exercise testing with the patient on medication-a reasonable estimate of the drug’s "cardioprotectiveness" -may influence the choice of management. In the patient with confirmed coronary disease, what is to be made of test results that differ according to whether the patient is on or off treatment? It would be fallacious to label an abnormal test result with the patient off treatment as false positive and misleading to regard a normal result with medication as false negative. Since an exercise-induced drop in ejection fraction or ST-segment depression implies ischaemia, a negative test in a patient on medication indicates that ischaemia is not inducible. What if a test done with a patient on treatment is abnormal? No prospective study with exercise electrocardiography has been done to evaluate this issue. For post-infarction patients taking beta-blockers there is indirect evidence to suggest an adverse outlook if there is at least 2 mm ST depression during exercise electrocardiography. For patients with recent unstable angina (in whom medication withdrawal is unwise) prognosis is worse if the pre-discharge exercise test during drug therapy is positive rather than negative.6 Though the accuracy of the ST-segment/heart rate slope method seems unaffected by beta-blockersthis observation paradoxically limits its usefulness in addressing the effect of medication on the prognostic interpretation of exercise testing. Preliminary radionuclide ventriculographic data show that exercise testing without drug withdrawal may help in the assessment of prognosis in coronary disease.8 In survivors of infarction treated with thrombolysis-a group whose results are often abnormal when exercised without treatment-testing with medication may improve predictive accuracy for recurrent ischaemia.9 In symptom-free patients with residual silent ischaemia during exercise testing after drug withdrawal, suppression of ischaemia by medicaton at repeat testing was associated with improvement in shortterm outcomes These observations support what many clinicians intuitively believe: that an exercise test that is abnormal despite clinically adequate anti-ischaemic medication identifies high-risk patients in whom medical therapy will fail and who will therefore need early revascularisation to improve their prognosis. In retrospect, the prognostic information derived from exercise testing in the randomised studies of surgical versus medical treatment might have been more relevant had testing been done systematically with rather than without medical treatment, since the real issue on trial was early surgery versus continuing with medical treatment, not surgery versus no treatment at all. In conclusion, whether anti-anginal medication should or should not be withdrawn may be influenced by the purpose behind testing. When it is of prime importance to clarify a diagnosis of ischaemic heart disease, we recommend stopping all anti-ischaemic treatment for 3-4 days before testing. When the overriding concern in the patient with a typical history or proven coronary disease is to define the prognosis, we think that testing may give more useful results if done without stopping medication.
2. Gohlke H, Samek L, Betz P, Roskamm H. Exercise testing provides prognostic information in angiographically defined subgroups of patients with coronary artery disease. Circulation 1983; 68: 979-85. 3. Jones RH, McEwan P, Newman GE, et al. Accuracy and diagnosis of coronary artery disease by radionuclide measurement of left ventricular function during rest and exercise. Circulation 1981; 64: 586-601. 4. Ho SWC, McComish MJ, Taylor RR. Effect of beta-adrenergic blockade on the results of exercise testing related to the extent of coronary artery disease. Am J Cardiol 1985; 55: 258-62. 5. Krone RJ, Miller JP, Gillespie JA, Weld FM, and the Multicenter Post-infarction Research Group. Usefulness of low-level exercise testing after acute myocardial infarction in patients taking betablocking agents. Am J Cardiol 1987; 60: 23-27. 6. Wilcox I, Freedman SB, Allman KC, et al. Prognostic significance of a predischarge exercise test in risk stratification after unstable angina pectoris. J Am Coil Cardiol 1991; 18: 677-83. 7. Elamin MS, Boyle R, Kardash MM, et al. Accurate detection of coronary heart disease by new exercise test. Br Heart J 1982; 48: 311-20. 8. Lim R, Dyke L, Dymond DS. Prognostic importance of failure of medical therapy to normalise exercise ejection fraction response in coronary artery disease. J Am Coil Cardiol 1991; 17: 183A. 9. Lim R, Dyke L, Dymond DS. Early prognosis after thrombolysis: value of exercise radionuclide ventriculography performed on anti-ischemic medication. Int J Card Imaging 1991; 7: 125-31. 10. Lim R, Dyke L, Dymond DS. Effect on prognosis of abolition of exercise-induced painless myocardial ischemia by medical therapy. Am J Cardiol 1992; 69: 733-35.
Diet and
causal relation wishful thinking?
cancer:
just
or
The beneficial effects of a prudent diet have been claimed far back as Biblical times (Daniel I, 8-16), but the definition of a healthy diet has varied over generations. In medieval times a lavish intake was favoured, since plump individuals were thought to live longest. At the turn of this century, Rolo Russell began to assert that cancer mortality is highest "in countries that eat more flesh". There was much laboratory activity in the 1940s and 1950sobut in the past three decades the relation between diet and cancer has come under special scrutiny. Existing evidence is based on three sorts of data-indirect relations between the consumption of selected food constituents with cancer incidence and mortality3-’ or time trends in various population groups;8 case control studies;’H1 and laboratory experimentation." The reported correlations are undermined by the fact that human diet does not consist of isolated food components. Furthermore, dietary patterns strongly correlate with social, economic, and political characteristics, all of which could affect cancer risk independently. Therefore, any attempt to isolate one single factor in carcinogenesis may be futile. Long-term prospective studies are not the perfect solution because we cannot determine the time at which
as
carcinogenesis begins. Laboratory experimentations seem much more controllable and more definite, but extrapolation to man is questionable in terms of time, dose, and life patterns. We must also take into consideration
publication bias, whereby "positive" findings are more likely to be printed than "negative". This bias seriously confounds the scientific literature, and cannot be countered by the magic trick of the 1990s-"adjusted for". For example, suppose we interview
REFERENCES 1. Stone
DL, Dymond DS, Elliott AT, Britton KE, Banim SO, Spurrell RAJ. Exercise first pass radionuclide ventriculography in detection of patients with coronary artery disease. Br Heart J 1980; 44: 208-14.
ADDRESS Department of Clinical Epidemiology, Chaim Sheba Medical Center, Tel Hashomer, Israel (Prof B Modan, MD).
aldosterone is responsive to angiotensin and hybrid steroids are not in excess, the mutation may be different, perhaps causing excessive responsiveness to angiotensin (eg, double dose of the promoter region of CYPIIB2). Clinically, we are suggesting that primary aldosteronism is a commoner cause of hypertension than previously suspected, and better screening tests may lead to earlier diagnosis with the normokalaemic variety already becoming common. In patients with efficient antihypertensive mechanisms, a normotensive form may be identifiable by abnormal aldosterone/renin ratio. Such normotensive forms are likely to be found more frequently in relatives of patients with primary aldosteronism. Identification of the genetic abnormalities will permit detection, by genetic screening, of those predisposed to primary aldosteronism, and measures such as a low-salt diet might delay or prevent the development of hypertension and hypokalaemia. In collaboration with groups in Australia and elsewhere we are currently examining the DNA from our patients with primary aldosteronism, seeking mutations in GYP 11 Bl, CYPllB2, and other candidate genes. We are also seeking evidence of disturbed aldosterone biosynthesis in the relatives of patients with primary aldosteronism. REFERENCES 1. Lifton RP,
Dluhy RG, Powers M, et al. A chimaeric 11 &bgr;-hydroxylase/ aldosterone synthase gene causes glucocorticoid-remediable aldosteronism and human hypertension. Nature 1992; 355: 262-65. 2. Sutherland DJA, Ruse JL, Laidlaw JC. Hypertension, increased aldosterone secretion and low plasma renin activity relieved by dexamethasone. Can Med Assoc J 1966; 95: 1109-19. 3. Matsuo K, Kawai K, Tsuchiyama H, Veki V. Glucocorticoidsuppressible hyperaldosteronism: ultrastructural observation of a case.
Acta Pathol Jpn 1985; 35: 1511-19. 4. Miura K, Yoshinaga K, Goto K, et al. A case of glucocorticoid-responsive
hyperaldosteronism.J Clin Endocrinol 1968; 28:
1807-15.
5. Gordon RD, Stowasser M, Tunny TJ, Klemm SA, Finn WL, Krek AL. Clinical and pathlogical diversity of primary aldosteronism including a new familial variety. Clin Exp Pharmacol Physiol 1991; 18: 283-86. 6. Woodland E, Tunny TJ, Hamlet SM, Gordon RD. Hypertension corrected and aldosterone responsiveness to renin-angiotensin restored by long-term dexamethasone in glucocorticoid-suppressible hyperaldosteronism. Clin Exp Pharmacol Physiol 1985; 12: 245-48. 7. Gomez-Sanchez CE, Gill JR, Ganguly A, Gordon RD. Glucocorticoidsuppressible aldosteronism: a disorder of the adrenal transitional zone. J Clin Endocrinol Metab 1988; 67: 444-48. 8. Stowasser M, Gordon RD, Tunny TJ, Klemm SA, Finn WL, Krek AL. Primary aldosteronism: implications of a new familial variety. J Hypertension 1991; 9 (suppl 6): S264-65. 9. Stowasser M, Gordon RD, Tunny TJ, Klemm SA, Finn WL, Krek AL. Familial hyperaldosteronism type II: five families with a new variety of primary aldosteronism. Clin Exp Pharmacol Physiol 1992; 19: 319-22. 10. Hoefnagels WHL, Kloppenborg PWC. Hazards of long-term dexamethasone treatment in primary aldosteronism. N Engl J Med 1982; 306: 427. 11. Gordon RD, Hamlet SM, Tunny TJ, Klemm SA. Aldosteroneproducing adenomas responsive to angiotensin pose problems in diagnosis. Clin Exp Pharmacol Physiol 1987; 17: 175-79. 12. Gordon RD, Gomez-Sanchez CE, Hamlet SM, Tunny TJ, Klemm SA. Angiotensin-responsive aldosterone-producing adenoma masquerades as idiopathic hyperaldosteronism (IHA, adrenal hyperplasia) or low renin essential hypetension. J Hypertension 1987; 5 (suppl 5): S103-06. 13. Tunny TJ, Gordon RD, Klemm SA, Cohn D. Histological and biochemical distinctiveness of atypical aldosterone-producing adenomas responsive to upright posture and angiotensin. Clin Endocrinol 1991; 34: 363-69. 14. Axelrod L, Vickery AL (Jr). Case Records of the Massachusetts General Hospital: a 52-year-old man with hypertension, hypokalemia, and an adrenal mass. N Engl J Med 1992; 326: 1617-23. 15. Hiramatsu K, Yamada T, Yukimura Y, et al. A screening test to identify aldosterone-producing adenoma by measuring plasma renin activity: results in hypertensive patients. Arch Intern Med 1981; 141: 1589-93. 16. Hamlet SM, Tunny TJ, Woodland E, Gordon RD. Is aldosterone/renin ratio useful to screen a hypertensive population for primary aldosteronism? Clin Exp Pharmacol Physiol 1985; 12: 249-52. 17. McKenna TJ, Sequeira SJ, Heffernan A, Chambers J, Cunningham S. Diagnosis under random conditions of all disorders of the renin-
angiotensin-aldosterone axis, including primary hyperaldosteronism. J Clin Endocrinol Metab 1991; 73: 952-57.
VIEWPOINT Should
medication be stopped for exercise testing?
antianginal
dispute the contribution of exercise electrocardiography or radionuclide ventriculography to the management of ischaemic heart disease. Reports of their value were based largely on tests done after withdrawal of antianginal drug therapy for several hours or days.12 This practice stems from the observation that anti-anginal drugs, in particular beta-blockers, can mask otherwise detectable abnormalities—ie, ST-segment depression during exercise electrocardiography or a drop in left ventricular ejection fraction during exercise radionuclide ventriculography.3,4 Few would
Two
points need
be taken into account in the of the value of exercise testing. First, atheromatous coronary disease can be diagnosed only by coronary angiography. Exercise tests simply indicate the functional effect of the obstruction and thus the probability of the presence and extent of the lesion. Secondly, although such testing may also yield prognostic information, the diagnostic value of the test may differ from its prognostic value. In patients in whom there is doubt about the clinical diagnosis, unstable angina or myocardial infarction are
assessment
to
unlikely to occur on withdrawal of therapy. In practice, instability is usually encountered only in patients receiving several drugs for chronic angina, in whom the likelihood of coronary disease is already high. This observation argues against drug withdrawal for testing in such patients, since doing so may not only provoke instability but also be of little value in management. In someone with a typical history, an abnormal test after drug withdrawal is unsurprising. However, a test in the same patient that shows persistent ischaemia despite medication may induce a sense of urgency.
For example, to determine whether early invasive investigation and treatment are indicated after myocardial infarction, it seems more sensible to undertake prognostic testing with the patient on rather than off any medication for secondary prevention. When a patient with typical angina is rendered symptom-free by drugs, testing with the patient on Department of Cardiology, St Bartholomew’s Hospital, London EC1A 7BE, UK (R. Lim, MRCP, D. S. Dymond, FRCP). Correspondence to Dr Richard Lim.
ADDRESS:
162
such medication may give some idea of prognosis while the treatment is continued and whether urgent referral for angiography is needed. When angiography indicates that medical treatment is as justifiable as revascularisation, the result of exercise testing with the patient on medication-a reasonable estimate of the drug’s "cardioprotectiveness" -may influence the choice of management. In the patient with confirmed coronary disease, what is to be made of test results that differ according to whether the patient is on or off treatment? It would be fallacious to label an abnormal test result with the patient off treatment as false positive and misleading to regard a normal result with medication as false negative. Since an exercise-induced drop in ejection fraction or ST-segment depression implies ischaemia, a negative test in a patient on medication indicates that ischaemia is not inducible. What if a test done with a patient on treatment is abnormal? No prospective study with exercise electrocardiography has been done to evaluate this issue. For post-infarction patients taking beta-blockers there is indirect evidence to suggest an adverse outlook if there is at least 2 mm ST depression during exercise electrocardiography. For patients with recent unstable angina (in whom medication withdrawal is unwise) prognosis is worse if the pre-discharge exercise test during drug therapy is positive rather than negative.6 Though the accuracy of the ST-segment/heart rate slope method seems unaffected by beta-blockersthis observation paradoxically limits its usefulness in addressing the effect of medication on the prognostic interpretation of exercise testing. Preliminary radionuclide ventriculographic data show that exercise testing without drug withdrawal may help in the assessment of prognosis in coronary disease.8 In survivors of infarction treated with thrombolysis-a group whose results are often abnormal when exercised without treatment-testing with medication may improve predictive accuracy for recurrent ischaemia.9 In symptom-free patients with residual silent ischaemia during exercise testing after drug withdrawal, suppression of ischaemia by medicaton at repeat testing was associated with improvement in shortterm outcomes These observations support what many clinicians intuitively believe: that an exercise test that is abnormal despite clinically adequate anti-ischaemic medication identifies high-risk patients in whom medical therapy will fail and who will therefore need early revascularisation to improve their prognosis. In retrospect, the prognostic information derived from exercise testing in the randomised studies of surgical versus medical treatment might have been more relevant had testing been done systematically with rather than without medical treatment, since the real issue on trial was early surgery versus continuing with medical treatment, not surgery versus no treatment at all. In conclusion, whether anti-anginal medication should or should not be withdrawn may be influenced by the purpose behind testing. When it is of prime importance to clarify a diagnosis of ischaemic heart disease, we recommend stopping all anti-ischaemic treatment for 3-4 days before testing. When the overriding concern in the patient with a typical history or proven coronary disease is to define the prognosis, we think that testing may give more useful results if done without stopping medication.
2. Gohlke H, Samek L, Betz P, Roskamm H. Exercise testing provides prognostic information in angiographically defined subgroups of patients with coronary artery disease. Circulation 1983; 68: 979-85. 3. Jones RH, McEwan P, Newman GE, et al. Accuracy and diagnosis of coronary artery disease by radionuclide measurement of left ventricular function during rest and exercise. Circulation 1981; 64: 586-601. 4. Ho SWC, McComish MJ, Taylor RR. Effect of beta-adrenergic blockade on the results of exercise testing related to the extent of coronary artery disease. Am J Cardiol 1985; 55: 258-62. 5. Krone RJ, Miller JP, Gillespie JA, Weld FM, and the Multicenter Post-infarction Research Group. Usefulness of low-level exercise testing after acute myocardial infarction in patients taking betablocking agents. Am J Cardiol 1987; 60: 23-27. 6. Wilcox I, Freedman SB, Allman KC, et al. Prognostic significance of a predischarge exercise test in risk stratification after unstable angina pectoris. J Am Coil Cardiol 1991; 18: 677-83. 7. Elamin MS, Boyle R, Kardash MM, et al. Accurate detection of coronary heart disease by new exercise test. Br Heart J 1982; 48: 311-20. 8. Lim R, Dyke L, Dymond DS. Prognostic importance of failure of medical therapy to normalise exercise ejection fraction response in coronary artery disease. J Am Coil Cardiol 1991; 17: 183A. 9. Lim R, Dyke L, Dymond DS. Early prognosis after thrombolysis: value of exercise radionuclide ventriculography performed on anti-ischemic medication. Int J Card Imaging 1991; 7: 125-31. 10. Lim R, Dyke L, Dymond DS. Effect on prognosis of abolition of exercise-induced painless myocardial ischemia by medical therapy. Am J Cardiol 1992; 69: 733-35.
Diet and
causal relation wishful thinking?
cancer:
just
or
The beneficial effects of a prudent diet have been claimed far back as Biblical times (Daniel I, 8-16), but the definition of a healthy diet has varied over generations. In medieval times a lavish intake was favoured, since plump individuals were thought to live longest. At the turn of this century, Rolo Russell began to assert that cancer mortality is highest "in countries that eat more flesh". There was much laboratory activity in the 1940s and 1950sobut in the past three decades the relation between diet and cancer has come under special scrutiny. Existing evidence is based on three sorts of data-indirect relations between the consumption of selected food constituents with cancer incidence and mortality3-’ or time trends in various population groups;8 case control studies;’H1 and laboratory experimentation." The reported correlations are undermined by the fact that human diet does not consist of isolated food components. Furthermore, dietary patterns strongly correlate with social, economic, and political characteristics, all of which could affect cancer risk independently. Therefore, any attempt to isolate one single factor in carcinogenesis may be futile. Long-term prospective studies are not the perfect solution because we cannot determine the time at which
as
carcinogenesis begins. Laboratory experimentations seem much more controllable and more definite, but extrapolation to man is questionable in terms of time, dose, and life patterns. We must also take into consideration
publication bias, whereby "positive" findings are more likely to be printed than "negative". This bias seriously confounds the scientific literature, and cannot be countered by the magic trick of the 1990s-"adjusted for". For example, suppose we interview
REFERENCES 1. Stone
DL, Dymond DS, Elliott AT, Britton KE, Banim SO, Spurrell RAJ. Exercise first pass radionuclide ventriculography in detection of patients with coronary artery disease. Br Heart J 1980; 44: 208-14.
ADDRESS Department of Clinical Epidemiology, Chaim Sheba Medical Center, Tel Hashomer, Israel (Prof B Modan, MD).
