680
for the future. Treatment for Wilms’ tumour is being and the elimination of radiotherapy in many patients will reduce the risks of many long-term effects, perhaps including those on the remaining kidney. Cure is cure for life; and we are still 30-40 years away from counting the final cost of that cure.5 PETER J. SHAW Children’s Hospital, MICHAEL STEVENS Camperdown, Sydney 2050, Australia treatment
continuously improved,
DM, Morgan ER, Vozar MA, Langman CB. Renal function reserve in long-term survivors of unilateral Wilms tumor. J Pediatr 1991; 118: 698-702. 2. Barrera M, Roy LP, Stevens M. Long-term follow-up after unilateral nephrectomy and radiotherapy for Wilms’ tumour. Pediatr Nephrol 1989; 3: 430-32. 3. Makipernaa A, Koskimies O, Jaaskelainen J, Teppo A-M, Siimes MA. Renal growth and function 11-28 years after treatment of Wilms’ tumour Eur J Pediatr 1991; 1. Bhisitkul
150: 444-47. 4. Evans AE, Norkool P, Evans I, Breslow N, D’Angio GJ. Late effects of treatment for Wilms’ Tumor. Cancer 1991; 67: 331-36. 5. Craft AW, Pearson ADJ. Three decades of chemotherapy for childhood cancer: from cure ’at any cost’ to cure ’at least cost’. Cancer Surv 1989; 8: 605-29.
Screening for osteoporosis SIR,-Dr Stevenson and his colleagues (Feb 8, p 370) claim that screening for osteoporotic fracture by bone mineral density (BMD) measurement would have a sensitivity ranging from 80% down to 58% for cut-off levels equal to or below the lower quartile. These estimates are said to derive from the results of prospective studies which indicate a 1 5-3-0 relative risk of fracture with each standard deviation reduction in BMD, together with a 15-30% lifetime fracture risk. However, such a high sensitivity would imply much greater relative risks. For example, suppose 1000 women with a 15% lifetime risk were screened with the lower quartile cut-off. If the sensitivity was 80%, 120 of the 150 with future fractures would be screened positive as would 212 (25%) of the remaining 850 women. Thus the lifetime risk in those with positive results would be 120(120 + 212) or 36% compared with 30 - (30 + 638) or 4% in the negatives, a relative risk of 9. For an 80% sensitivity and 75% specificity the average BMD in those with future fracture must be Istandard deviations below normal. All the available evidence points to a much smaller separation in BMD distribution. Academic Unit of Obstetrics and Gynaecology, Department of Clinical Medicine, and Institute of Epidemiology and Health Services Research, St James’s University Hospital, Leeds LS9 7TF, UK
HOWARD CUCKLE
SIR,-Dr Stevenson and colleagues point out that the Effective Health Care report on screening for osteoporosis drew heavily on our paper, and they challenge our conclusions that bone mineral density (BMD) measurement is a poor screening test for future hip fracture and that the protection from hip fracture imparted by HRT is largely lost several years after replacement stops.’1 They do not dispute the point that measurement of BMD at the time of fracture does not distinguish women with and without a hip fracture. (Among women of the same age the separation between the two groups is about half a standard deviation so that, for example, only about 13% of all hip fractures would occur in the 5% of people with the lowest BMD; prospective studies yield a similar estimate.’) Their point that "some of the controls will later be cases" is true, but not the implication that allowance for this would improve the performance of screening. The distribution of BMD in "new cases" that arise over the next five or ten years is likely to be similar to that in the existing cases; indeed it will be curtailed to a small extent because people with lower BMD have a higher mortality and so die younger. Their assertion that there is a similar degree of overlap in blood pressure between people who do or do not develop a stroke is wrong; the overlap is less. The argument that the poor case for screening applies only to hip fracture and not to other osteoporotic fractures that are more frequent is not valid because detection rate and false-positive rate (sensitivity and specificity) do not depend upon disease prevalence, and there is no evidence that screening performance is better for other fractures. None of the points made by Stevenson et al alters our conclusion that screening is not worthwhile.
Stevenson and colleagues are incorrect in stating that, after hormone replacement therapy (HRT) "epidemiological studies show that protection persists into extreme old age". The data from studies of hip fracture are conclusive: the four studies showed substantial protection with current use but little effect after about five years from stopping.’The explanation may be that bone loss following cessation of hormone replacement is more rapid than normal postmenopausal loss ("catch-up" loss), there is evidence both for3 and against this (Stevenson’s letter). Whether or not catch-up loss occurs is not critical, however. Both a low BMD and a fall are needed for a high risk of fracture. BMD becomes low by age 65 but falls are not common until after about 75, so that fractures become a serious problem only after this age. Although BMD declines rapidly after the menopause it declines much more slowly after 65; it is only about 4% less at 75 than at 65.4 Delaying this rapid loss by taking oestrogens for ten years will have little effect on BMD after 75 when falls become frequent. To be effective in preventing fracture, HRT must be started soon after the menopause and continued for many years. We are surprised that Stevenson and colleagues dispute the view that exercise and stopping smoking influence osteoporotic fractures, since the evidence is so consistent. We cited 26 studies supporting an effect of exercise and 13 for smoking.
stopping
Wolfson Institute of Preventive Medicine, Medical College of St Bartholomew’s Hospital, London EC1 M 6BQ, UK
M. R. LAW
N. J. WALD T. W. MEADE
MR, Wald NJ, Meade TW. Strategies for prevention of osteoporosis and hip fracture. Br Med J 1991; 303: 453-59. 2. Browner WS, Seeley DG, Vogt TM, Cummings SR. Non-trauma mortality in elderly women with low bone mineral density. Lancet 1991; 338: 355-58. 3. Law MR, Wald NJ, Meade TW Preventing osteoporosis Br Med J 1991, 303: 922 4. Mazess RB, Barden HS, Ettinger M, et al. Spine and femur density using dual-photon absorptiometry in US white women. J Bone Min Res 1987; 2: 211-19. 1. Law
Hypertension, in black and white SIR,-Your editorial of Jan 4 (p 28) draws attention to the recent mapping of two genes that may regulate blood pressure (BP). This complicated interplay of nature and nurture in the aetiology of hypertension was highlighted by referring to the pattern of blood pressure and hypertension in African-Americans in the USA, where several studies have examined the relation between racial factors and predisposition to hypertension.’ Studies on indigenous west Africans were not discussed, even though US blacks originated from the west coast of Africa. Wholesale comparisons ofBP patterns in blacks should be advanced with some caution, especially since the populations in some studies cited are from ethnic groups very different from those of the ancestors of the US blacks of today. In Nigeria, over the past two decades, we have studied population blood pressure patterns2 and some related biochemical data. 5 With a population of over 100 million, Nigeria is home to the largest concentration of the world’s blacks. Some of our findings are at variance with those cited in the editorial but offer interesting insights into the nature of BP in negroid populations. Urban-rural differences in BP are one way of predicting the effects a change in lifestyle may have on blood pressure. We found no difference in the casual systolic and diastolic BP in urban and rural populations in NigeriaThe usual rise in BP with age was present in both populations and this rise was comparable to that seen in non-negroid populations. Recently an urban blood pressure survey reaffirmed the prevalence (8%) of hypertension (BP > 160/95 mm Hg) in Nigeria.’ Of interest was the significant (p
for the future. Treatment for Wilms’ tumour is being and the elimination of radiotherapy in many patients will reduce the risks of many long-term effects, perhaps including those on the remaining kidney. Cure is cure for life; and we are still 30-40 years away from counting the final cost of that cure.5 PETER J. SHAW Children’s Hospital, MICHAEL STEVENS Camperdown, Sydney 2050, Australia treatment
continuously improved,
DM, Morgan ER, Vozar MA, Langman CB. Renal function reserve in long-term survivors of unilateral Wilms tumor. J Pediatr 1991; 118: 698-702. 2. Barrera M, Roy LP, Stevens M. Long-term follow-up after unilateral nephrectomy and radiotherapy for Wilms’ tumour. Pediatr Nephrol 1989; 3: 430-32. 3. Makipernaa A, Koskimies O, Jaaskelainen J, Teppo A-M, Siimes MA. Renal growth and function 11-28 years after treatment of Wilms’ tumour Eur J Pediatr 1991; 1. Bhisitkul
150: 444-47. 4. Evans AE, Norkool P, Evans I, Breslow N, D’Angio GJ. Late effects of treatment for Wilms’ Tumor. Cancer 1991; 67: 331-36. 5. Craft AW, Pearson ADJ. Three decades of chemotherapy for childhood cancer: from cure ’at any cost’ to cure ’at least cost’. Cancer Surv 1989; 8: 605-29.
Screening for osteoporosis SIR,-Dr Stevenson and his colleagues (Feb 8, p 370) claim that screening for osteoporotic fracture by bone mineral density (BMD) measurement would have a sensitivity ranging from 80% down to 58% for cut-off levels equal to or below the lower quartile. These estimates are said to derive from the results of prospective studies which indicate a 1 5-3-0 relative risk of fracture with each standard deviation reduction in BMD, together with a 15-30% lifetime fracture risk. However, such a high sensitivity would imply much greater relative risks. For example, suppose 1000 women with a 15% lifetime risk were screened with the lower quartile cut-off. If the sensitivity was 80%, 120 of the 150 with future fractures would be screened positive as would 212 (25%) of the remaining 850 women. Thus the lifetime risk in those with positive results would be 120(120 + 212) or 36% compared with 30 - (30 + 638) or 4% in the negatives, a relative risk of 9. For an 80% sensitivity and 75% specificity the average BMD in those with future fracture must be Istandard deviations below normal. All the available evidence points to a much smaller separation in BMD distribution. Academic Unit of Obstetrics and Gynaecology, Department of Clinical Medicine, and Institute of Epidemiology and Health Services Research, St James’s University Hospital, Leeds LS9 7TF, UK
HOWARD CUCKLE
SIR,-Dr Stevenson and colleagues point out that the Effective Health Care report on screening for osteoporosis drew heavily on our paper, and they challenge our conclusions that bone mineral density (BMD) measurement is a poor screening test for future hip fracture and that the protection from hip fracture imparted by HRT is largely lost several years after replacement stops.’1 They do not dispute the point that measurement of BMD at the time of fracture does not distinguish women with and without a hip fracture. (Among women of the same age the separation between the two groups is about half a standard deviation so that, for example, only about 13% of all hip fractures would occur in the 5% of people with the lowest BMD; prospective studies yield a similar estimate.’) Their point that "some of the controls will later be cases" is true, but not the implication that allowance for this would improve the performance of screening. The distribution of BMD in "new cases" that arise over the next five or ten years is likely to be similar to that in the existing cases; indeed it will be curtailed to a small extent because people with lower BMD have a higher mortality and so die younger. Their assertion that there is a similar degree of overlap in blood pressure between people who do or do not develop a stroke is wrong; the overlap is less. The argument that the poor case for screening applies only to hip fracture and not to other osteoporotic fractures that are more frequent is not valid because detection rate and false-positive rate (sensitivity and specificity) do not depend upon disease prevalence, and there is no evidence that screening performance is better for other fractures. None of the points made by Stevenson et al alters our conclusion that screening is not worthwhile.
Stevenson and colleagues are incorrect in stating that, after hormone replacement therapy (HRT) "epidemiological studies show that protection persists into extreme old age". The data from studies of hip fracture are conclusive: the four studies showed substantial protection with current use but little effect after about five years from stopping.’The explanation may be that bone loss following cessation of hormone replacement is more rapid than normal postmenopausal loss ("catch-up" loss), there is evidence both for3 and against this (Stevenson’s letter). Whether or not catch-up loss occurs is not critical, however. Both a low BMD and a fall are needed for a high risk of fracture. BMD becomes low by age 65 but falls are not common until after about 75, so that fractures become a serious problem only after this age. Although BMD declines rapidly after the menopause it declines much more slowly after 65; it is only about 4% less at 75 than at 65.4 Delaying this rapid loss by taking oestrogens for ten years will have little effect on BMD after 75 when falls become frequent. To be effective in preventing fracture, HRT must be started soon after the menopause and continued for many years. We are surprised that Stevenson and colleagues dispute the view that exercise and stopping smoking influence osteoporotic fractures, since the evidence is so consistent. We cited 26 studies supporting an effect of exercise and 13 for smoking.
stopping
Wolfson Institute of Preventive Medicine, Medical College of St Bartholomew’s Hospital, London EC1 M 6BQ, UK
M. R. LAW
N. J. WALD T. W. MEADE
MR, Wald NJ, Meade TW. Strategies for prevention of osteoporosis and hip fracture. Br Med J 1991; 303: 453-59. 2. Browner WS, Seeley DG, Vogt TM, Cummings SR. Non-trauma mortality in elderly women with low bone mineral density. Lancet 1991; 338: 355-58. 3. Law MR, Wald NJ, Meade TW Preventing osteoporosis Br Med J 1991, 303: 922 4. Mazess RB, Barden HS, Ettinger M, et al. Spine and femur density using dual-photon absorptiometry in US white women. J Bone Min Res 1987; 2: 211-19. 1. Law
Hypertension, in black and white SIR,-Your editorial of Jan 4 (p 28) draws attention to the recent mapping of two genes that may regulate blood pressure (BP). This complicated interplay of nature and nurture in the aetiology of hypertension was highlighted by referring to the pattern of blood pressure and hypertension in African-Americans in the USA, where several studies have examined the relation between racial factors and predisposition to hypertension.’ Studies on indigenous west Africans were not discussed, even though US blacks originated from the west coast of Africa. Wholesale comparisons ofBP patterns in blacks should be advanced with some caution, especially since the populations in some studies cited are from ethnic groups very different from those of the ancestors of the US blacks of today. In Nigeria, over the past two decades, we have studied population blood pressure patterns2 and some related biochemical data. 5 With a population of over 100 million, Nigeria is home to the largest concentration of the world’s blacks. Some of our findings are at variance with those cited in the editorial but offer interesting insights into the nature of BP in negroid populations. Urban-rural differences in BP are one way of predicting the effects a change in lifestyle may have on blood pressure. We found no difference in the casual systolic and diastolic BP in urban and rural populations in NigeriaThe usual rise in BP with age was present in both populations and this rise was comparable to that seen in non-negroid populations. Recently an urban blood pressure survey reaffirmed the prevalence (8%) of hypertension (BP > 160/95 mm Hg) in Nigeria.’ Of interest was the significant (p
