- Email: [email protected]
SERUM CHOLESTEROL, BLOOD PRESSURE, AND MORTALITY
1331 Popovic M, Sarngadharan MG, Read E, et al. Detection, isolation and continuous production of cytopathic retroviruses (HTLV-III) from patients with AIDS and pre-AIDS. Science 1984; 224: 497-500. 4. Fuller TC, Trevithick JE, Fuller AA, et al. Antigenic polymorphism of the T4 differentiation antigen expressed on human T helper/inducer lymphocytes. Hum 3.
Immunol 1984; 9: 89-102. 5. Hoxie JA, Flaherty LE, Haggarty BS, et al. Infection of T4 lymphocytes by HTLV-III does not require expression on the OKT4 epitope. J Immunol 1986; 136: 361. 6. Serwadda D, Mugerwa RD, Sewankombo NK, et al Slim disease a new disease in Uganda and its association with HTLV-III infection Lancet 1985; ii: 849-52.
SERUM
CHOLESTEROL, BLOOD PRESSURE, AND MORTALITY
SIR,-Dr Martin and his colleagues (Oct 25, p 933) have been the size of their cohort (361 662 men) but the large
beguiled by
number does not justify the drawing of large conclusions. Had they confined themselves to answering their question-namely, "At what levels should intensive treatment and hypolipidaemic drugs be advised?"-it would have been encouraging to find the longstanding clinical practice of treating those with the highest levels of cholesterol receive such strong epidemiological support. Martin et al succumb, however, to the temptation to go much further. By an apparent broad eyeballing and artificial division of data on coronary heart disease (CHD) death rate, they introduce the concept of "baseline risk". This is arbitrary and misleading. No justification for the use of quintiles is provided, and it is certainly not the case that the first quintile is unique in having no rise in CHD death rate within it. Their histogram shows that this is true of every quintile, apart from the fifth. One is tempted to assume that having a substantial population within the baseline risk makes this a more plausible and realistic aim for the remainder. But there is no baseline risk; the lower the serum cholesterol, the lower the CHD death rate but not the lower total mortality. Indeed, as the other graphs make clear, there is no real variation in total mortality below about the 70th percentile--and little therapeutic potential unless dietary and other advice could be provided to induce everyone to crowd into the magic window of 175-200 mg/dl. Even if the concept of baseline risk is accepted the histogram reveals that, within the 95% confidence limits, there is no increase in CHD death rate until around the 40th percentile, and even then not much. 95% confidence limits are strikingly absent from the tabulated percentage excess risks. Although the excess of deaths at the lowest levels of serum cholesterol may well reflect cancer or other serious illness, Martin et al should surely acknowledge the widespread finding of increased biliary-tract malignancy when serum cholesterol is lowered, irrespective of the method. Although the comparison with CHD death rate related to diastolic blood pressure is interesting, it provides no support for a drive to reduce the population serum cholesterol. Quite the reverse. The bulk of the comparison is with diastolic blood pressures below 94 mm Hg. Although it has been established from actuarial statistics over the past fifty years that increasing blood pressure is associated with increasing mortality, hardly anybody would argue that someone with a blood pressure of 90 mm Hg or less should receive antihypertensive therapy. It has been hard enough to demonstrate any benefit from treating at levels below 110 mm Hg. It is quite wrong to accuse doctors of a lack of awareness of the importance of the two recognised risk factors of serum cholesterol and blood pressure. On the contrary, current practice reflects the profession’s awareness since we treat those with the highest levels of either. People’s recollections about having their blood pressure and serum cholesterol levels measured is as soft a piece of data as one could wish for. How many patients having a "blood test" know in detail what is being measured? This powerful cohort study provides further support for intervention at the highest levels but fails to provide evidence that I
"four-fifths of the population is likely to benefit". Martin et al ignore potential harmful effects and seem to include women in conclusions based on a study of 300 000 men. 88 Kelvin Court, Glasgow G12 0AH
BRIAN C. CAMPBELL
CAPTOPRIL VERSUS FRUSEMIDE IN MODERATE HEART FAILURE
SIR,-Dr Cowley and colleagues (Oct 4, p 770) report that patients with stable, moderate heart failure who were already receiving a low dose of frusemide achieved greater symptomatic benefit by increasing the dose of diuretic rather than by adding captopril in increasing dosage. We have used a double-blind protocol similar to that used by Cowley and colleagues although over 29 patients were treated in parallel and only the frusemide dose was increased over the 3 months of study. 1,2 14 patients received captopril and 15 received frusemide (1 patient in each group was excluded from the 3-month data). Both groups showed similar improvement in exercise tolerance time and functional class-the mean NYHA score was 2-5 (SD 0-4) before treatment with captopril and 1-5 (0-5) at 3 months compared with 2-6 (0-4) and 2-0 (0-9) before and after treatment with increasing doses of frusemide. However, assessed
by M-mode echocardiography, the two had opposite effects. Fractional shortening increased after captopril from a mean of 19-4% (SD 7-5) to 21-0% (8-0) but decreased after frusemide from 19-5% (9-9) to 18-2% (13-1). Endtreatments
systolic wall stress in the left ventricle decreased after captopril from 250 mmHg (SD 67) to 239 mmHg (81) but increased after frusemide from 253 mmHg (80) to 287 mmHg (89). Thus, we feel that captopril provides a more "physiological" approach to the a mean of
treatment
of heart failure.
Department of Cardiology, S. Camillo Hospital, Rome
A. BOCCANELLI S. M. LIBERATORE P. L. PRATI
A, et al. Addition of captopril versus increasing diuretics in moderate but deteriorating heart failure: a double-blind comparative trial. Postgrad MedJ 1986; 62 (suppl 1): 184-87. 2. Boccanelli A, et al. Captopril added to low dose of frusemide as an alternative to increasing diuretics in mild to moderate heart failure. Proceedings of the Tenth World Congress of Cardiology, Washington 1986 Sept 13-20, 1986 (abstract). 1. Boccanelli
COLONOSCOPY AND BARIUM ENEMAS
SIR,-In reply to the letter from Dr Freeman and Professor Sherwood (Nov 1, p 1045), we do indeed recommend colonoscopy as the preferred investigation in patients with persistent rectal bleeding or a change in bowel habit in whom proctoscopy and rigid sigmoidoscopy have failed to reveal an anorectal cause, and this is now established practice in our unit. We appreciate the economic implications of such a policy should it become standard practice but would ask "should patients be denied the better investigation simply on the grounds of cost?" If the answer to this question is "yes" then we have to accept a certain rate of missed cancers and polyps and with it the implications for the patients concerned. If missing a certain proportion of such lesions is not thought acceptable then resources must be diverted into the endoscopy service. The barium enema is probably a little cheaper than colonoscopy but the newer colonoscopes have made endoscopy very cost-effective.’ In reply to Dr James, the double-contrast barium enemas in our study were done by a variety of radiologists of all grades, and colonoscopy was done by one consultant surgeon and one surgical registrar. We made no effort to ensure equality of expertise between radiologist and colonoscopist because we were interested in the "real world" rather than the rigid and unreal confines of a clinical trial. All barium enemas were felt to be of adequate diagnostic quality such that a confident report of either normality or diverticular disease alone could be made by the radiologist. There are adequate data available from comparative studiesu to suggest that endoscopy is more accurate than double-contrast barium enema in the detection of colonic lesions, especially those located in the sigmoid. Our aim was not to compare these two techniques but to determine the frequency with which neoplastic
Immunol 1984; 9: 89-102. 5. Hoxie JA, Flaherty LE, Haggarty BS, et al. Infection of T4 lymphocytes by HTLV-III does not require expression on the OKT4 epitope. J Immunol 1986; 136: 361. 6. Serwadda D, Mugerwa RD, Sewankombo NK, et al Slim disease a new disease in Uganda and its association with HTLV-III infection Lancet 1985; ii: 849-52.
SERUM
CHOLESTEROL, BLOOD PRESSURE, AND MORTALITY
SIR,-Dr Martin and his colleagues (Oct 25, p 933) have been the size of their cohort (361 662 men) but the large
beguiled by
number does not justify the drawing of large conclusions. Had they confined themselves to answering their question-namely, "At what levels should intensive treatment and hypolipidaemic drugs be advised?"-it would have been encouraging to find the longstanding clinical practice of treating those with the highest levels of cholesterol receive such strong epidemiological support. Martin et al succumb, however, to the temptation to go much further. By an apparent broad eyeballing and artificial division of data on coronary heart disease (CHD) death rate, they introduce the concept of "baseline risk". This is arbitrary and misleading. No justification for the use of quintiles is provided, and it is certainly not the case that the first quintile is unique in having no rise in CHD death rate within it. Their histogram shows that this is true of every quintile, apart from the fifth. One is tempted to assume that having a substantial population within the baseline risk makes this a more plausible and realistic aim for the remainder. But there is no baseline risk; the lower the serum cholesterol, the lower the CHD death rate but not the lower total mortality. Indeed, as the other graphs make clear, there is no real variation in total mortality below about the 70th percentile--and little therapeutic potential unless dietary and other advice could be provided to induce everyone to crowd into the magic window of 175-200 mg/dl. Even if the concept of baseline risk is accepted the histogram reveals that, within the 95% confidence limits, there is no increase in CHD death rate until around the 40th percentile, and even then not much. 95% confidence limits are strikingly absent from the tabulated percentage excess risks. Although the excess of deaths at the lowest levels of serum cholesterol may well reflect cancer or other serious illness, Martin et al should surely acknowledge the widespread finding of increased biliary-tract malignancy when serum cholesterol is lowered, irrespective of the method. Although the comparison with CHD death rate related to diastolic blood pressure is interesting, it provides no support for a drive to reduce the population serum cholesterol. Quite the reverse. The bulk of the comparison is with diastolic blood pressures below 94 mm Hg. Although it has been established from actuarial statistics over the past fifty years that increasing blood pressure is associated with increasing mortality, hardly anybody would argue that someone with a blood pressure of 90 mm Hg or less should receive antihypertensive therapy. It has been hard enough to demonstrate any benefit from treating at levels below 110 mm Hg. It is quite wrong to accuse doctors of a lack of awareness of the importance of the two recognised risk factors of serum cholesterol and blood pressure. On the contrary, current practice reflects the profession’s awareness since we treat those with the highest levels of either. People’s recollections about having their blood pressure and serum cholesterol levels measured is as soft a piece of data as one could wish for. How many patients having a "blood test" know in detail what is being measured? This powerful cohort study provides further support for intervention at the highest levels but fails to provide evidence that I
"four-fifths of the population is likely to benefit". Martin et al ignore potential harmful effects and seem to include women in conclusions based on a study of 300 000 men. 88 Kelvin Court, Glasgow G12 0AH
BRIAN C. CAMPBELL
CAPTOPRIL VERSUS FRUSEMIDE IN MODERATE HEART FAILURE
SIR,-Dr Cowley and colleagues (Oct 4, p 770) report that patients with stable, moderate heart failure who were already receiving a low dose of frusemide achieved greater symptomatic benefit by increasing the dose of diuretic rather than by adding captopril in increasing dosage. We have used a double-blind protocol similar to that used by Cowley and colleagues although over 29 patients were treated in parallel and only the frusemide dose was increased over the 3 months of study. 1,2 14 patients received captopril and 15 received frusemide (1 patient in each group was excluded from the 3-month data). Both groups showed similar improvement in exercise tolerance time and functional class-the mean NYHA score was 2-5 (SD 0-4) before treatment with captopril and 1-5 (0-5) at 3 months compared with 2-6 (0-4) and 2-0 (0-9) before and after treatment with increasing doses of frusemide. However, assessed
by M-mode echocardiography, the two had opposite effects. Fractional shortening increased after captopril from a mean of 19-4% (SD 7-5) to 21-0% (8-0) but decreased after frusemide from 19-5% (9-9) to 18-2% (13-1). Endtreatments
systolic wall stress in the left ventricle decreased after captopril from 250 mmHg (SD 67) to 239 mmHg (81) but increased after frusemide from 253 mmHg (80) to 287 mmHg (89). Thus, we feel that captopril provides a more "physiological" approach to the a mean of
treatment
of heart failure.
Department of Cardiology, S. Camillo Hospital, Rome
A. BOCCANELLI S. M. LIBERATORE P. L. PRATI
A, et al. Addition of captopril versus increasing diuretics in moderate but deteriorating heart failure: a double-blind comparative trial. Postgrad MedJ 1986; 62 (suppl 1): 184-87. 2. Boccanelli A, et al. Captopril added to low dose of frusemide as an alternative to increasing diuretics in mild to moderate heart failure. Proceedings of the Tenth World Congress of Cardiology, Washington 1986 Sept 13-20, 1986 (abstract). 1. Boccanelli
COLONOSCOPY AND BARIUM ENEMAS
SIR,-In reply to the letter from Dr Freeman and Professor Sherwood (Nov 1, p 1045), we do indeed recommend colonoscopy as the preferred investigation in patients with persistent rectal bleeding or a change in bowel habit in whom proctoscopy and rigid sigmoidoscopy have failed to reveal an anorectal cause, and this is now established practice in our unit. We appreciate the economic implications of such a policy should it become standard practice but would ask "should patients be denied the better investigation simply on the grounds of cost?" If the answer to this question is "yes" then we have to accept a certain rate of missed cancers and polyps and with it the implications for the patients concerned. If missing a certain proportion of such lesions is not thought acceptable then resources must be diverted into the endoscopy service. The barium enema is probably a little cheaper than colonoscopy but the newer colonoscopes have made endoscopy very cost-effective.’ In reply to Dr James, the double-contrast barium enemas in our study were done by a variety of radiologists of all grades, and colonoscopy was done by one consultant surgeon and one surgical registrar. We made no effort to ensure equality of expertise between radiologist and colonoscopist because we were interested in the "real world" rather than the rigid and unreal confines of a clinical trial. All barium enemas were felt to be of adequate diagnostic quality such that a confident report of either normality or diverticular disease alone could be made by the radiologist. There are adequate data available from comparative studiesu to suggest that endoscopy is more accurate than double-contrast barium enema in the detection of colonic lesions, especially those located in the sigmoid. Our aim was not to compare these two techniques but to determine the frequency with which neoplastic