495 table we have compared the chance of having an affected fetus Stiller et al. give no information about their presence or of the and women who have positive results after being screened for of sex the An indication absence. patients pregnancy to exclude and would be useful also neural-tube defects with that in four other common groups for would help another z, history whom amniocentesis is currently accepted. source of error. The small numbers of patients on which Stiller Of the women in the five groups, those with high serumet al. based their conclusions make this information essential. A.F.P. levels (>2.5 x median at 16-18 weeks gestation after We found an 18% one-year graft survival in 33 patients receivin transfusions contrast to the have been excluded by ultrasound) have 71% intraoperative only,6 multiple pregnancies ing the highest chance of having an affected fetus. When ultra(5 out of 7) survival reported by Stiller et al. Patients who had been pregnant and those with leucocyte sound is used as an additional check on gestational age some women in this group will be found to have normal levels, and antibodies in their serum were excluded in our study. We believe that only preoperative blood-transfusions have the risk of an affected fetus in the remaining women will be even greater-about 13% on the basis of our own preliminary been shown to have a beneficial effect on renal surresults in Oxford. (The risk of spina bifida alone will be about vival. Our concern is that that practice will be abandoned in favour of the perhaps simpler, but yet unproved, intraopera7%.) In addition, A.F.P. screening can be expected to identify tive transfusion. Our policy of giving at least one blood-transabout 84% of all infants with open neural-tube defects,3 fusion to patients awaiting a renal transplant will remain unwhereas in the other four groups no more than about a third of all affected fetuses will be identified. changed. While the decision to perform an amniocentesis on an indiEurotransplant Foundation, vidual patient will depend on many considerations, the risks c/o Bloodbank, G. G. PERSIJN shown in the table are likely to be an important influence. In University Hospital, Leiden, Netherlands J. J. VAN ROOD the light of these figures it would seem to be inconsistent to regard amniocentesis as unacceptable in women with high serum-A.F.p. levels but acceptable in the other groups. SCREENING FOR NEURAL-TUBE DEFECTS cator.
in
allograft
,
SIR,--Your editorial of Feb. 11 asks if neural-tube defect
screening by measurement of maternal serum alpha-fetoprotein (A.F.P.) should be introduced. The aim of such screening is to identify a group of women with a high enough risk of having an affected fetus to justify amniocentesis. While the risks RISK IN BRITAIN OF HAVING A FETUS WITH A NEURAL-TUBE
D.H.S.S. Cancer Epidemiology and Clinical Trials Unit, Department of Regius Professor of Medicine, Radcliffe Infirmary, Oxford OX2 6HE
NICHOLAS WALD HOWARD CUCKLE
Nuffield Department of Obstetrics and Gynæcology, John Radcliffe Hospital, Oxford
GORDON STIRRAT
DEFECT OR A CHROMOSOMAL ABNORMALITY AMONG FIVE GROUPS ’ OF WOMEN FOR WHOM AMNIOCENTESIS MIGHT BE CONSIDERED
CORONARY-ARTERY BYPASS SURGERY
SIR,-The Veterans Administration paper’ has serious flaws. The survival curves in fig. 2 are labelled wrongly, as comparison with previous V.A. reports shows:2.3 the medical and surgical curves are interchanged until the point of crossover at 30 months (after which the curves are identified correctly) and the medical and surgical curves in fig. 3 are inter-
changed. consideration of the subsets of two and as a single subset on the basis of simiobstruction three-artery lar survival. Addition of the two-artery obstructions to the three-artery group in table IV improves the prognosis for medically treated patients but worsens the outlook for the surgical group. Calculation of the apparent medical and surgical mortality for obstruction of two arteries and an abnormal left ventricle does not conform to survival previously reported at 36 months.2 The numbers presented in the paper refer to randomised but not to compliant patients. Many problems in the V.A. reports would be clarified by a statement giving the numbers of medical and surgical deaths at various specific intervals (e.g., 24 and 48 months) and the numbers of compliant patients eligible to have survived the stated time, both for the whole study and for specific subsets. Reduction of the number of operative deaths would affect survival. During 1973 (the middle year of the V.A. study) our operative mortality for 1400 patients, excluding left-main coronary-artery obstructions, was 0.6%. There was a perioperative infarction-rate of about 4% and graft patency of 87% durThe
*
Serum-A.F.p. >2.5 x median at 16-18 weeks gestation after multiple pregnancy has been excluded by ultrasound; the figures in parentheses, based on our unpublished data, relate to pregnancies which still have high A.F.P. levels after gestational age has been checked by ultrasound. t The 84% comprises an average of 79% relating to open spina bifida only and 88% relating to anencephaly, assuming that each is equally common at birth. t These percentages are based on the frequency of chromomal abnormatmes found in mid-trimester amniotic fluid samples; the other percentages m the table are based on data at birth.
of amniocentesis are still not precisely known they do not appear to be great,I,2 and the procedure is already established clinical practice in several situations. In the accompanying 1. N.I.C.H.D. National
Registry for Amniocentesis Study Group J. Am. med. Ass. 1976, 236, 1471. 2. Simpson, N. E., Dallaire, L., Miller, J. R., Simonovich, L., Hamerton, J. L., Miller, J., McKeen, C. Can. med. Ass. J. 1976, 115, 739. 3. U.K. Collaborative Study Lancet, 1977,i, 1323. 4. Carter, C. O., David, P. A., Laurence, K. M. Med. Genet. 1968, 5, 81. 5. Galjaard, H. Cytogenet. Cell Genet. 1976, 16, 453.
ing
text
that
justifies
period.
V.A.
figures
were
5-6% operative mortality,
Registrar General’s Statistical Review for England and Wales, 1973: part I (A), tables, medical. H.M. Stationery Office. 7. Office of Population Censuses and Surveys Medical Division. Personal com-
6.
munication. 8. Griffith, G. W. Hlth Trends, 1973, 5, 59. 1. Detre, K., Murphy, M. L., Hultgren, H. Lancet, 1977, ii, 1243. 2. Murphy, M. L., Hultgren, H. N., Petre, K. M., Thomsen, J., Takaro, T.
New Engl. J. Med. 1977, 297, 621. 3. Read, R. C., Detre, K. M., Murphy, M. L., thorac. cardiovasc. Surg. 1978, 75, 1.
Hultgren,
H.
N., Takaro, T. J.
496
graft patency 69%.2 3-year survival in our series was 95%. Many of these patients would not have been operated on if the V.A. criteria for the selection of lowrisk cases had been followed. Other groups are reporting figures in the same range. Lower operative mortality in the V.A. studies would have resulted in statistically significant differences in survival, most readily demonstrable in the threeartery obstruction subset. Other V.A. publications have acknowledged the problem of high operative mortality. 2,3 If the operative mortality de-3 creased from 16% for 1970 and 1971 to 5.6% for 1972-74, it is reasonable that still lower mortality might have been achieved in time. One of the major participating V.A. hospitals had a 2% operative mortality during 1972-74.4 The operative mortality for ten of the thirteen V.A. hospitals was 3.4% during these periods, which means that the average mortality for the remaining three hospitals must have been 13%, assuming that all hospitals contributed equal numbers of patients, which was not true. If a lesser number of patients were contributed by the three hospitals having high operative mortalities, the percentage mortality would be even higher. A multicentre study of a surgical procedure cannot tolerate such wide variations in operative mortality. Such variability must be reflected -’n long-term survival, so the experience of various participating hospitals should be indicated individually. The suggestion has been made that it might be necessary to launch a new V.A. study.2 This project would be difficult to implement because of the increasing problems in patient compliance in a society in which freedom of choice is readily available. infarction-rate 18%,’ and
Cleveland Clinic Foundation, Cleveland, Ohio 44106, U.S.A.
detail on pages 1469 and 1470 of the Dec. 29, 1977, issue of the New England Journal of Medicine by both Dr E. Braunwald and the V.A. study group. The general conclusion was that while the challenging claims for the efficiency of surgical therapy in reducing mortality sounded very promising, these claims needed to be confirmed by controlled studies. When this same V.A. study reported a favourable effect of bypass surgery for the patients with left-main disease’ no criticisms of the study were heard. Similarly, Rowe finds fault with the V.A. study 1.2 because of "dismal evidence that the heart was not revascularised", but he welcomes uncritically a positive report4 where the revascularisation was worse, as judged by the higher % prevalence of three-vessel disease in Vismara’s series4 coupled with a similar % frequency of triple-graft surgery :
Imperfections in a controlled study do not lend validity to empirical clinical experiences, nor are data from these experiences usually collected and disclosed in a manner which would permit the scrutiny which is possible for the V.A. study data. Department of Epidemiology, Graduate School of Public Health,
University of Pittsburgh, Pittsburgh, Pennsylvania 15261,
U.S.A.
KATHERINE M. DETRE
WILLIAM L. PROUDFIT
***This letter was shown to Dr Detre.-ED. L. POSTURE IN LABOUR
SIR,-Dr Proudfit is correct: in fig. 2 of our paper the medical and
surgical survival curves are interchanged up to 30 months and in fig. 3 the two treatment curves are interchanged throughout. These mistakes occurred while the figures were being redrawn by The Lancet but they do not affect the nature of our conclusions. Proudfit incorrectly implies that the survival figures of the non-left-main, two and three vessel disease, abnormal left-ventricular-function (L.V.F.) subgroup displayed in table tv of our Lancet paper’ and the survival experience reported in the New England Journal of Medicine2 are inconsistent. In the N.J’.Af. we reported 84% cumulative survival.rates at 36 months follow-up for both the medical and the surgical patients with two-vessel disease abnormal L.v.F. and, for the same follow-up period, cumulative survival-rates of 82% for the medical and 86% for surgical patients with three-vessel disease abnormal L.V.F. The combined survival for multivessel disease abnormal L.V.F. at that time was 82% for medical and 85% for surgical patients, but at four years these figures changed to 81% and 82% respectively (table iv’) due to 1 medical death and 2 surgical deaths during the fourth year. Proudfit should not consider 2% differences in survival (lines 2 and 3 in table iv’) as improvement or worsening in the prognosis when the standard errors of the survival-rates in question range from 3 to 5 %. The remaining comments from Proudfit and those made earlier by Dr Rowe (Feb. 4, p. 264) are not specific to our Lancet paper. They are criticisms of the general issues such as operative mortality, representativeness of the study population, and completeness of revascularisation in the V.A. study. Discussions and replies to such criticisms have been given in 4.
Hultgren, H. N. Paper presented at a meeting of the American Heart Association, in December, 1977. 5. Murphy, M. K., Hultgren, H. N., Detre, K., Thomsen, J., Takaro, T. New Engl. J. Med. 1977, 297, 1470. 1. Detre, K., Murphy, M., Hultgren, H. Lancet, 1977, ii, 1243. 2. Murphy, M., Hultgren, H., Detre, K., Thomsen, J., Takaro, T. New Engl. J. Med. 1977, 297, 621.
SIR,-Dr McManus and Dr Calder (Jan. 14, p. 72)
are cor-
in believing that for many years I have been concerned about the convention of nursing patients in a recumbent position during labour and especially on their backs.$An upright labour and delivery position was considered normal in most societies, including our own, until modern times, the change to recumbency being undertaken mainly for the convenience of the accoucheur.6,7 In 1868 Duncan8calculated that the weight of the fetus at term was sufficient to supply much of the force required to bring the child into the world provided the mother was in an upright position. Mendez-Bauer et al. have demonstrated that the gravitational force is equivalent to a pressure of 25 mm Hg. This force is independent of and likely to be safer than that produced by uterine contraction. Why, then, is it neglected? This is all the more strange because recumbency is an effective means of preventing or treating premature labour. The explanation would seem to lie somewhere between the power of custom and prejudice, and convenience. Apart from McManus and Calder’s, three controlled studies have been published on the effects of maternal posture during labour. In 1974 Mendez-Bauer’s group"’" studied 20 nulliparx, using them as their own controls by comparing the effect of successive alternate adoption of upright and recumbent posrect
3. 4.
Takaro, T., and others. Circulation, 1976, 54, suppl. no. 3, p. 107. Vismara, I. A., Miller, R. R., Price, J. E., Karem, R., DeMaria, A. N., Mason, D. T. Am. J. Cardiol. 1977, 39, 919. 5. Dunn, P. M. Lancet, 1976, i, 790. 6. Englemann, G. J. Labour Among Primitive Peoples. St. Louis, 1882.
7. Atwood, R. J. Acta obstet. gynæc scand. 1976, suppl. 57, p. 6. J. M. Researches in Obstetrics; p. 321. Edinburgh, 1868.
8. Duncan,
9. Mendez-Bauer, C., Arroyo, J., Menendez, A., Salmeau, J., Manas, J., Lavilla, M., Martinez San Martin, S., Villa Elizaga, I., and Zamarriego Crespo, J. in Perinatal Medicine: 5th European Congress of Perinatal Medicine (edited by G. Rooth and L.-E. Bratteby); p. 233. Stockholm, 1976. 10. 11.
and others. Ginec. Obstet. 1974, 1, 221. Mendez-Bauer, C., Arroyo, J., Garcia-Ramos, C., Menendez, A., Lavilla, M., Izquierdo, F., Villa Elizaga, I., Zamarriego, J. J. perinat. Med. 1975, 3, 89.
Arroyo, J.,
in
allograft
,
SIR,--Your editorial of Feb. 11 asks if neural-tube defect
screening by measurement of maternal serum alpha-fetoprotein (A.F.P.) should be introduced. The aim of such screening is to identify a group of women with a high enough risk of having an affected fetus to justify amniocentesis. While the risks RISK IN BRITAIN OF HAVING A FETUS WITH A NEURAL-TUBE
D.H.S.S. Cancer Epidemiology and Clinical Trials Unit, Department of Regius Professor of Medicine, Radcliffe Infirmary, Oxford OX2 6HE
NICHOLAS WALD HOWARD CUCKLE
Nuffield Department of Obstetrics and Gynæcology, John Radcliffe Hospital, Oxford
GORDON STIRRAT
DEFECT OR A CHROMOSOMAL ABNORMALITY AMONG FIVE GROUPS ’ OF WOMEN FOR WHOM AMNIOCENTESIS MIGHT BE CONSIDERED
CORONARY-ARTERY BYPASS SURGERY
SIR,-The Veterans Administration paper’ has serious flaws. The survival curves in fig. 2 are labelled wrongly, as comparison with previous V.A. reports shows:2.3 the medical and surgical curves are interchanged until the point of crossover at 30 months (after which the curves are identified correctly) and the medical and surgical curves in fig. 3 are inter-
changed. consideration of the subsets of two and as a single subset on the basis of simiobstruction three-artery lar survival. Addition of the two-artery obstructions to the three-artery group in table IV improves the prognosis for medically treated patients but worsens the outlook for the surgical group. Calculation of the apparent medical and surgical mortality for obstruction of two arteries and an abnormal left ventricle does not conform to survival previously reported at 36 months.2 The numbers presented in the paper refer to randomised but not to compliant patients. Many problems in the V.A. reports would be clarified by a statement giving the numbers of medical and surgical deaths at various specific intervals (e.g., 24 and 48 months) and the numbers of compliant patients eligible to have survived the stated time, both for the whole study and for specific subsets. Reduction of the number of operative deaths would affect survival. During 1973 (the middle year of the V.A. study) our operative mortality for 1400 patients, excluding left-main coronary-artery obstructions, was 0.6%. There was a perioperative infarction-rate of about 4% and graft patency of 87% durThe
*
Serum-A.F.p. >2.5 x median at 16-18 weeks gestation after multiple pregnancy has been excluded by ultrasound; the figures in parentheses, based on our unpublished data, relate to pregnancies which still have high A.F.P. levels after gestational age has been checked by ultrasound. t The 84% comprises an average of 79% relating to open spina bifida only and 88% relating to anencephaly, assuming that each is equally common at birth. t These percentages are based on the frequency of chromomal abnormatmes found in mid-trimester amniotic fluid samples; the other percentages m the table are based on data at birth.
of amniocentesis are still not precisely known they do not appear to be great,I,2 and the procedure is already established clinical practice in several situations. In the accompanying 1. N.I.C.H.D. National
Registry for Amniocentesis Study Group J. Am. med. Ass. 1976, 236, 1471. 2. Simpson, N. E., Dallaire, L., Miller, J. R., Simonovich, L., Hamerton, J. L., Miller, J., McKeen, C. Can. med. Ass. J. 1976, 115, 739. 3. U.K. Collaborative Study Lancet, 1977,i, 1323. 4. Carter, C. O., David, P. A., Laurence, K. M. Med. Genet. 1968, 5, 81. 5. Galjaard, H. Cytogenet. Cell Genet. 1976, 16, 453.
ing
text
that
justifies
period.
V.A.
figures
were
5-6% operative mortality,
Registrar General’s Statistical Review for England and Wales, 1973: part I (A), tables, medical. H.M. Stationery Office. 7. Office of Population Censuses and Surveys Medical Division. Personal com-
6.
munication. 8. Griffith, G. W. Hlth Trends, 1973, 5, 59. 1. Detre, K., Murphy, M. L., Hultgren, H. Lancet, 1977, ii, 1243. 2. Murphy, M. L., Hultgren, H. N., Petre, K. M., Thomsen, J., Takaro, T.
New Engl. J. Med. 1977, 297, 621. 3. Read, R. C., Detre, K. M., Murphy, M. L., thorac. cardiovasc. Surg. 1978, 75, 1.
Hultgren,
H.
N., Takaro, T. J.
496
graft patency 69%.2 3-year survival in our series was 95%. Many of these patients would not have been operated on if the V.A. criteria for the selection of lowrisk cases had been followed. Other groups are reporting figures in the same range. Lower operative mortality in the V.A. studies would have resulted in statistically significant differences in survival, most readily demonstrable in the threeartery obstruction subset. Other V.A. publications have acknowledged the problem of high operative mortality. 2,3 If the operative mortality de-3 creased from 16% for 1970 and 1971 to 5.6% for 1972-74, it is reasonable that still lower mortality might have been achieved in time. One of the major participating V.A. hospitals had a 2% operative mortality during 1972-74.4 The operative mortality for ten of the thirteen V.A. hospitals was 3.4% during these periods, which means that the average mortality for the remaining three hospitals must have been 13%, assuming that all hospitals contributed equal numbers of patients, which was not true. If a lesser number of patients were contributed by the three hospitals having high operative mortalities, the percentage mortality would be even higher. A multicentre study of a surgical procedure cannot tolerate such wide variations in operative mortality. Such variability must be reflected -’n long-term survival, so the experience of various participating hospitals should be indicated individually. The suggestion has been made that it might be necessary to launch a new V.A. study.2 This project would be difficult to implement because of the increasing problems in patient compliance in a society in which freedom of choice is readily available. infarction-rate 18%,’ and
Cleveland Clinic Foundation, Cleveland, Ohio 44106, U.S.A.
detail on pages 1469 and 1470 of the Dec. 29, 1977, issue of the New England Journal of Medicine by both Dr E. Braunwald and the V.A. study group. The general conclusion was that while the challenging claims for the efficiency of surgical therapy in reducing mortality sounded very promising, these claims needed to be confirmed by controlled studies. When this same V.A. study reported a favourable effect of bypass surgery for the patients with left-main disease’ no criticisms of the study were heard. Similarly, Rowe finds fault with the V.A. study 1.2 because of "dismal evidence that the heart was not revascularised", but he welcomes uncritically a positive report4 where the revascularisation was worse, as judged by the higher % prevalence of three-vessel disease in Vismara’s series4 coupled with a similar % frequency of triple-graft surgery :
Imperfections in a controlled study do not lend validity to empirical clinical experiences, nor are data from these experiences usually collected and disclosed in a manner which would permit the scrutiny which is possible for the V.A. study data. Department of Epidemiology, Graduate School of Public Health,
University of Pittsburgh, Pittsburgh, Pennsylvania 15261,
U.S.A.
KATHERINE M. DETRE
WILLIAM L. PROUDFIT
***This letter was shown to Dr Detre.-ED. L. POSTURE IN LABOUR
SIR,-Dr Proudfit is correct: in fig. 2 of our paper the medical and
surgical survival curves are interchanged up to 30 months and in fig. 3 the two treatment curves are interchanged throughout. These mistakes occurred while the figures were being redrawn by The Lancet but they do not affect the nature of our conclusions. Proudfit incorrectly implies that the survival figures of the non-left-main, two and three vessel disease, abnormal left-ventricular-function (L.V.F.) subgroup displayed in table tv of our Lancet paper’ and the survival experience reported in the New England Journal of Medicine2 are inconsistent. In the N.J’.Af. we reported 84% cumulative survival.rates at 36 months follow-up for both the medical and the surgical patients with two-vessel disease abnormal L.v.F. and, for the same follow-up period, cumulative survival-rates of 82% for the medical and 86% for surgical patients with three-vessel disease abnormal L.V.F. The combined survival for multivessel disease abnormal L.V.F. at that time was 82% for medical and 85% for surgical patients, but at four years these figures changed to 81% and 82% respectively (table iv’) due to 1 medical death and 2 surgical deaths during the fourth year. Proudfit should not consider 2% differences in survival (lines 2 and 3 in table iv’) as improvement or worsening in the prognosis when the standard errors of the survival-rates in question range from 3 to 5 %. The remaining comments from Proudfit and those made earlier by Dr Rowe (Feb. 4, p. 264) are not specific to our Lancet paper. They are criticisms of the general issues such as operative mortality, representativeness of the study population, and completeness of revascularisation in the V.A. study. Discussions and replies to such criticisms have been given in 4.
Hultgren, H. N. Paper presented at a meeting of the American Heart Association, in December, 1977. 5. Murphy, M. K., Hultgren, H. N., Detre, K., Thomsen, J., Takaro, T. New Engl. J. Med. 1977, 297, 1470. 1. Detre, K., Murphy, M., Hultgren, H. Lancet, 1977, ii, 1243. 2. Murphy, M., Hultgren, H., Detre, K., Thomsen, J., Takaro, T. New Engl. J. Med. 1977, 297, 621.
SIR,-Dr McManus and Dr Calder (Jan. 14, p. 72)
are cor-
in believing that for many years I have been concerned about the convention of nursing patients in a recumbent position during labour and especially on their backs.$An upright labour and delivery position was considered normal in most societies, including our own, until modern times, the change to recumbency being undertaken mainly for the convenience of the accoucheur.6,7 In 1868 Duncan8calculated that the weight of the fetus at term was sufficient to supply much of the force required to bring the child into the world provided the mother was in an upright position. Mendez-Bauer et al. have demonstrated that the gravitational force is equivalent to a pressure of 25 mm Hg. This force is independent of and likely to be safer than that produced by uterine contraction. Why, then, is it neglected? This is all the more strange because recumbency is an effective means of preventing or treating premature labour. The explanation would seem to lie somewhere between the power of custom and prejudice, and convenience. Apart from McManus and Calder’s, three controlled studies have been published on the effects of maternal posture during labour. In 1974 Mendez-Bauer’s group"’" studied 20 nulliparx, using them as their own controls by comparing the effect of successive alternate adoption of upright and recumbent posrect
3. 4.
Takaro, T., and others. Circulation, 1976, 54, suppl. no. 3, p. 107. Vismara, I. A., Miller, R. R., Price, J. E., Karem, R., DeMaria, A. N., Mason, D. T. Am. J. Cardiol. 1977, 39, 919. 5. Dunn, P. M. Lancet, 1976, i, 790. 6. Englemann, G. J. Labour Among Primitive Peoples. St. Louis, 1882.
7. Atwood, R. J. Acta obstet. gynæc scand. 1976, suppl. 57, p. 6. J. M. Researches in Obstetrics; p. 321. Edinburgh, 1868.
8. Duncan,
9. Mendez-Bauer, C., Arroyo, J., Menendez, A., Salmeau, J., Manas, J., Lavilla, M., Martinez San Martin, S., Villa Elizaga, I., and Zamarriego Crespo, J. in Perinatal Medicine: 5th European Congress of Perinatal Medicine (edited by G. Rooth and L.-E. Bratteby); p. 233. Stockholm, 1976. 10. 11.
and others. Ginec. Obstet. 1974, 1, 221. Mendez-Bauer, C., Arroyo, J., Garcia-Ramos, C., Menendez, A., Lavilla, M., Izquierdo, F., Villa Elizaga, I., Zamarriego, J. J. perinat. Med. 1975, 3, 89.
Arroyo, J.,
