- Email: [email protected]
Chorionic villus sampling and limb abnormalities
876
We believe, first, that decisions about priorities for support in developing countries should be made by the countries themselves and not by officers from agencies located in developed countries. In addition to assessment of research design, the role of such agencies should be evaluation of the project, which should clearly show that the work is a local priority. Second, research institutions in developing countries need to collaborate well and strengthen their activities. Third, training of professionals from developing countries should be implemented by local institutions, which would ensure that the focus of training is closer to these countries’ needs, and an easier return home for the trainee. Our centre is presently providing a one-year postgraduate training course on research methodology for physicians in Latin America, supported by IDRC and HRP/WHO. In this course skills in the epidemiological and analytical methods to assess priorities are developed. In 1993, we will initiate a training programme for statisticians. However, every year our centre has to interrupt its activities because funds are obtained through a few institutions that provide non-aligned grants and because support for the training programme has to be renewed yearly. This situation leads to our last recommendation-to promote the long-term institutional support from funding agencies to guarantee the continuity of activities. Centro Rosarino de Estudios Perinatales, Orono 500, 2000 Rosario, Argentina
JOSÉ M. BELIZAN
1. The Commission on Health Research for Development. Health research: essential link to equity in development. Oxford: Oxford University Press, 1990.
Proof of causation SIR,-Mr Meeran (March 14, p 671) states that the standard of proof in legal cases is far less stringent than the standard applied in science. He suggests that experts called to present scientific evidence in court do not unduly restrict themselves to the traditional 95 % confidence interval (CI). The fallacy in his reasoning is that the "balance of probabilities" standard used in civil actions the (" > 50% rule") should not be compared to the CI (or degree-ofcertainty figure) used by epidemiologists. If anything, the burden of proof in civil actions is more comparable with the relative risk (RR) or the aetiological fraction (EF), and several US courts have recognised this. The causation issues dealt with in civil actions and by medical scientists are not the same. In a civil action the causation issue is typically whether the plaintiffs injuries were caused by agent X. Population-based studies by medical experts cannot answer the question of individual causation; they merely show whether an individual is at an increased risk if he or she is a member of a certain group. Since increased risk of future disease is generally not recoverable in American courts (unless the likelihood is > 50%), the broader question of whether a substance causes a statistically significant increased risk of disease is usually not in issue. It has been suggestedl that the > 50% burden of proof in civil cases is met when the EF exceeds 0-5 (RR > 20) because more than half the cases of the disease of interest in an exposed group are caused by exposure to agent X. While this is a more accurate analogy than the comparison to CI, this comparison is also vulnerable. The late Sir Austin-Bradford Hill pointed out many years ago that a statistically significant association is not equivalent to causation. Nor does the RR answer the issue of individual causation at issue in a tort action. If half the cases of the disease of interest in the exposed group may have been caused by agent X, half were not; and the study gives the jury and the judge no clue about which group the plaintiff falls into. Meeran’s suggestion that expert witnesses should consider using relaxed CIs is invalid. Judicial institutions do not and should not proceed oblivious to the well-established practices of scientific institutions. American courts typically require that methods used by testifying experts be generally accepted in the scientific discipline from which the expert comes. In fact, some American courts have specifically held that the 95% CIs must be applied by witnesses before their opinion will be allowed before a jury (eg, Brock v
Merrell Dow Pharm Inc [
1989] and Ealy v Richardson-Merrell Inc
[1990]). Traditional notions of causation in epidemiology (ie, RR > 2.0at 95 % CI) are not at all incompatible with the burden of proof in civil tort actions.2 Moreover, there is no rational public policy which would support the adoption of "second-tier" science in our judicial institutions. Sidley and Austin, One First National Plaza, Chicago, Illinois 60603, USA
TIMOTHY E. KAPSHANDY
1. Muscat JE, Huncharek MS. Causation and disease: biomedical science in toxic ton litigation. J Occup Med 1989, 31: 997-1002. 2. Kaye DH. Apples and oranges: confidence coefficients and the burden of persuasion Cornell Law Rev 1987; 73: 54-77.
Hazards of clinical trials SIR,—Your editorial (Dec 14, p 1495) points out the influence of controlled clinical trials on patient psychology. The possible adverse effects inherent in a trial may be termed trial morbidity, which might have contributed to the difference in mortality between the trial group and control group (who received "usual health care") seen several years after stopping the treatments in the Finnish Study. Another classic setting in which such morbidity might arise is an early intervention trial in patients with acute myocardial infarction. These patients are subjected to the mentally taxing process of decision making about participation in a trial while they are in emotional, physical, and psychological distress during the acute stages of the disease, and recent reports have implicated in-hospital psychological stress as a predictor for long-term cardiac mortality and re-infarction after acute myocardial infarction.2 It would be difficult to estimate trial morbidity, unless the intervention or drug is administered without the knowledge of the patient, which would be unethical. Follow-up of large trials and analysis at varying periods after the completion of the trial (with a knowledge of baseline psychological characteristics of different groups) may help to quantify and trace the causes of trial morbidity. Department of Medicine, Medical College, Calicut, Kerala, India 1.
2.
P. DILEEP KUMAR
Strandberg TE, Salomaa VV, Naukkarinen VA, Vanhanen HT, Sarna SJ, Miettinen TA. Long term mortality after 5 years multifactorial primary prevention of cardiovascular disease in middle aged men. JAMA 1991; 266: 1225-29. Frasure-Smith N. In-hospital symptoms of psychological stress as predictors of long-term outcome after acute myocardial infarction in men. Am J Cardiol 1991; 67: 121-27.
Chorionic villus sampling and limb abnormalities SIR,-Following reports of an association between chorionic villus sampling (CVS) and limb abnormalities,!,2an investigation was made in seven European Registration of Congenital Anomalies and Twins (EUROCAT) registries surveying altogether more than 600 000 births (table). These registries cover geographically defined populations. Malformations were registered for live births, stillbirths, and terminations after the diagnosis of malformation, except in Switzerland, where case-registration was restricted to live births. CVS exposure was recorded on the routine case-report forms of malformed babies, except in Odense, where registry data were matched to a known cohort of exposed women for this study. Where possible, extra information on the timing of exposure was sought. Chromosomal syndromes are not included in this report, since a high rate of exposure to CVS would be expected for such cases. Exposed cases with monogenic syndromes for which CVS could be indicated were also excluded. The exposure rate among cases of limb reduction is compared with the exposure rate among other anomalies in the table. 336 cases of limb reduction were reviewed, of which 4 (1-2%) had CVS exposure. In 11 883 cases of other congenital anomalies, 78 (066%) had CVS exposure, giving an odds ratio of 1 -8 (95% CI 0.66-4.99).
877
BIRTH POPULATIONS SURVEYED, TOTAL NUMBER AND NUMBER EXPOSEDTOCVSAMONG CASES OF LIMB REDUCTION AND OTHER MALFORMATIONS IN SEVEN EUROCAT REGISTRIES
The increase in risk is not statistically significant, the CI being wide. The 4 cases of limb reduction with CVS exposure were: (1) isolated agenesis of 2nd, 3rd, and 4th left (L) toes with hypoplasia of nails of remaining L toes (indication for CVS maternal age, CVS at 76 days of gestation, Paris); (2) partial amputation of four fingers of the L hand, ? amniotic bands (indication for CVS previous child with haemolytic uraemic anaemia, transcervical technique, CVS at 69 days, Switzerland); (3) amniotic disruption sequence with total aplasia of L leg, hypoplasia of L pelvis, L diaphragmatic hernia, imperforate anus, and vagina, pulmonary hypoplasia, kidney anomaly, persistent L superior vena cava, drainage of right (R) pulmonary vein into R atrium (indication for CVS was maternal age, CVS at 65 days of gestation, transcervical technique, Switzerland); and (4) Moebius syndrome with syndactyly of R and L hands, agenesis of the 2nd to 4th fingers of R hand, posterior cleft palate, micrognathism (indication for CVS family history of Duchenne muscular dystrophy, CVS at 79 days of gestation,
Strasbourg). We grouped together all cases of limb reduction for this analysis, but this might conceal an association with specific types of limb reduction. Limb reduction associated with Moebius syndrome is very rare, for example, and was also reported in the original Oxford series.1 3 of the 4 cases were exposed to CVS after 66 days of gestation, whereas Firth et aP hypothesised that the association would be restricted to CVS before 66 days. (Information on the timing of CVS in the seven populations is available from EUROCAT.) It has been suggested that the mechanism by which CVS leads to congenital anomalies might be a vascular insult, and that any defects thought to be due to hypoperfusion of the fetus should be investigated.l,2 Through a search for any additional exposed case with micrognathia, Moebius syndrome, destructive brain anomalies, or Poland anomaly the following were identified: (a) stillbirth with possible Hallermann-Streiff syndrome with microphthahnia, cleft palate, microretrognathia, anal atresia, tetralogy of Fallot, and absence of frontal bone (indication for CVS maternal chromosomal translocation [5;19], CVS at 11 completed weeks, Bouches-du-Rhone), (b) CHARGE syndrome with cleft lip and palate, atrioventricular canal, misshapen ear, syndactyly, facial nerve paralysis, micrognathism, sacral dimple, and deafness (indication for CVS maternal age, CVS at 57 days, transcervical technique [Portex and KC cannula], Groningen); and (c) multiple dysplasia of brain stem and diencephalon and L talipes equinovarus (indication for CVS maternal age, CVS at 63 days, transcervical technique, Switzerland). 1 further case of possible hypoperfusion anomaly (hydranencephaly) was excluded because CVS had been done following ultrasound indication of a brain anomaly. Thus a maximum of 7 exposed cases of possible hypoperfusion anomalies were recorded among the 82 non-chromosomal, nonmonogenic exposed cases (85%; 95% CI 2-5-14-6%). We can estimate from the EUROCAT database 1980-883that these hypoperfusion anomalies, including limb reduction, are expected to account for 8-9% of non-chromosomal anomalies. We cannot claim that CVS exposure was definitely known for all cases of limb reduction or hypoperfusion anomaly. Although reasonable levels of exposure ascertainment are suggested by the
of the CVS exposure figures between populataion estimates and estimates based on malformed cases in some registries, it is possible or even probable that the exposed population is at higher risk of malformation (other than chromosomal or genetic) than the unexposed. Where malformed cases were used as controls, CVS exposure would not be expected to be known less frequently in cases of limb reduction than in cases of other non-chromosomal anomalies. However, we cannot exclude the possibility that the CVS exposed population is at a higher (or lower) risk of limb reduction than of other malformations, independent of CVS. We have not attempted here to derive the prevalence of limb reduction defects in exposed pregnancies, since 1990 registry data were incomplete, and population data on CVS exposure was not known in some registries for an appropriate time period. EUROCAT data do not give any further evidence of a risk associated with early or late CVS, although the wide confidence limits around the odds ratio demonstrate the limited power of the study. They do suggest that, if there is a risk as high as that cited in Oxford (1% for all CVS, 1 ’7% for early CVS), it must be associated either with specific procedures or exposure timings which were infrequent in these EUROCAT registry areas or with specific anomaly types. H. DOLK EUROCAT, F. BERTRAND School of Public Health, M. F. LECHAT, UCL, for the EUROCAT Clos Chapelle-aux-Champs 30, B-1200 Brussels, Belgium working group*
similarity
* S. Aymé (Bouches-du-Rhône), D. Stone (Glasgow), H. de Walle (Groningen), E. Garne (Odense), J. Goujard and C. de Vigan (Paris), C. Stoll and B. Dott (Strasbourg), and T. Pexieder and D. Bloch (Switzerland).
HV, Boyd PA, Chamberlain P, Mackenzie IZ, Lindenbaum RH, Huson SM. Severe limb abnormalities after chorion villus sampling at 56-66 days’ gestation. Lancet 1991; 337: 762-63. 2. Mastroiacovo P, Cavalcanti DP. Limb reduction defects and chorion villus sampling. Lancet 1991; 337: 1091. 3. EUROCAT Working Group. Surveillance of congenital anomalies 1980-1988. EUROCAT Rep 1991; no 4: 24. 1. Firth
Cardiogenic embolism to the brain SIR,-Dr Hart’s contribution to your Stroke Octet (March 7, p 589) contains ambiguous, hazardous, and inaccurate target ranges
optimum oral anticoagulation, although the expression of the prolongation of the prothrombin times in terms of international normalised ratios (INR) is highly welcome. There is ambiguity in the so-called low-density anticoagulation: 2-3 INR is suggested for patients with myocardial infarction after hospital discharge, whereas for patients with a higher risk, the low-intensity range is presented as being 2-4 INR. We believe that the term low-intensity anticoagulation should no longer be used and should be replaced by moderate-intensity regimen as suggested by Hirsh.l Values of 4-7 INR as recommended for patients at high risk are highly hazardous. The use of such values would undoubtedly lead to an unduly large number of severe bleeding complications. The safe upper limit of the target range lies between 45 and 5 INR.2 Finally the quoted recommended range for patients enrolled in the European atrial fibrillation secondary prevention trial is inaccurate. This range is for
2.5-4 INR and not 2-4 INR.2 May I emphasise once again that the combination of oral anticoagulants with dipyridamole as recommended by Hart for many patients with mechanical prosthetic heart valves, has never been shown irrefutably to surpass the effect of anticoagulants
alone.3.4 Department of Haematology, Academisch Ziekenhuis, Postbus 9600, 2300 RC Leiden, Netherlands
E. A. LOELIGER
1. Hirsh J Oral anticoagulants. New Engl J Med 1991; 324: 1865-75. 2. Loeliger EA. Therapeutic target values in oral anticoagulation justification of Dutch policy. Ann Hematol 1992; 64: 60-65. 3. Loeliger EA. Does dipyridamole have antithrombotic potential? Thromb Haemostas 1985; 53: 437 4. FitzGerald GA. Dipyridamole. N Engl JMed 1987; 316: 1247-57.
We believe, first, that decisions about priorities for support in developing countries should be made by the countries themselves and not by officers from agencies located in developed countries. In addition to assessment of research design, the role of such agencies should be evaluation of the project, which should clearly show that the work is a local priority. Second, research institutions in developing countries need to collaborate well and strengthen their activities. Third, training of professionals from developing countries should be implemented by local institutions, which would ensure that the focus of training is closer to these countries’ needs, and an easier return home for the trainee. Our centre is presently providing a one-year postgraduate training course on research methodology for physicians in Latin America, supported by IDRC and HRP/WHO. In this course skills in the epidemiological and analytical methods to assess priorities are developed. In 1993, we will initiate a training programme for statisticians. However, every year our centre has to interrupt its activities because funds are obtained through a few institutions that provide non-aligned grants and because support for the training programme has to be renewed yearly. This situation leads to our last recommendation-to promote the long-term institutional support from funding agencies to guarantee the continuity of activities. Centro Rosarino de Estudios Perinatales, Orono 500, 2000 Rosario, Argentina
JOSÉ M. BELIZAN
1. The Commission on Health Research for Development. Health research: essential link to equity in development. Oxford: Oxford University Press, 1990.
Proof of causation SIR,-Mr Meeran (March 14, p 671) states that the standard of proof in legal cases is far less stringent than the standard applied in science. He suggests that experts called to present scientific evidence in court do not unduly restrict themselves to the traditional 95 % confidence interval (CI). The fallacy in his reasoning is that the "balance of probabilities" standard used in civil actions the (" > 50% rule") should not be compared to the CI (or degree-ofcertainty figure) used by epidemiologists. If anything, the burden of proof in civil actions is more comparable with the relative risk (RR) or the aetiological fraction (EF), and several US courts have recognised this. The causation issues dealt with in civil actions and by medical scientists are not the same. In a civil action the causation issue is typically whether the plaintiffs injuries were caused by agent X. Population-based studies by medical experts cannot answer the question of individual causation; they merely show whether an individual is at an increased risk if he or she is a member of a certain group. Since increased risk of future disease is generally not recoverable in American courts (unless the likelihood is > 50%), the broader question of whether a substance causes a statistically significant increased risk of disease is usually not in issue. It has been suggestedl that the > 50% burden of proof in civil cases is met when the EF exceeds 0-5 (RR > 20) because more than half the cases of the disease of interest in an exposed group are caused by exposure to agent X. While this is a more accurate analogy than the comparison to CI, this comparison is also vulnerable. The late Sir Austin-Bradford Hill pointed out many years ago that a statistically significant association is not equivalent to causation. Nor does the RR answer the issue of individual causation at issue in a tort action. If half the cases of the disease of interest in the exposed group may have been caused by agent X, half were not; and the study gives the jury and the judge no clue about which group the plaintiff falls into. Meeran’s suggestion that expert witnesses should consider using relaxed CIs is invalid. Judicial institutions do not and should not proceed oblivious to the well-established practices of scientific institutions. American courts typically require that methods used by testifying experts be generally accepted in the scientific discipline from which the expert comes. In fact, some American courts have specifically held that the 95% CIs must be applied by witnesses before their opinion will be allowed before a jury (eg, Brock v
Merrell Dow Pharm Inc [
1989] and Ealy v Richardson-Merrell Inc
[1990]). Traditional notions of causation in epidemiology (ie, RR > 2.0at 95 % CI) are not at all incompatible with the burden of proof in civil tort actions.2 Moreover, there is no rational public policy which would support the adoption of "second-tier" science in our judicial institutions. Sidley and Austin, One First National Plaza, Chicago, Illinois 60603, USA
TIMOTHY E. KAPSHANDY
1. Muscat JE, Huncharek MS. Causation and disease: biomedical science in toxic ton litigation. J Occup Med 1989, 31: 997-1002. 2. Kaye DH. Apples and oranges: confidence coefficients and the burden of persuasion Cornell Law Rev 1987; 73: 54-77.
Hazards of clinical trials SIR,—Your editorial (Dec 14, p 1495) points out the influence of controlled clinical trials on patient psychology. The possible adverse effects inherent in a trial may be termed trial morbidity, which might have contributed to the difference in mortality between the trial group and control group (who received "usual health care") seen several years after stopping the treatments in the Finnish Study. Another classic setting in which such morbidity might arise is an early intervention trial in patients with acute myocardial infarction. These patients are subjected to the mentally taxing process of decision making about participation in a trial while they are in emotional, physical, and psychological distress during the acute stages of the disease, and recent reports have implicated in-hospital psychological stress as a predictor for long-term cardiac mortality and re-infarction after acute myocardial infarction.2 It would be difficult to estimate trial morbidity, unless the intervention or drug is administered without the knowledge of the patient, which would be unethical. Follow-up of large trials and analysis at varying periods after the completion of the trial (with a knowledge of baseline psychological characteristics of different groups) may help to quantify and trace the causes of trial morbidity. Department of Medicine, Medical College, Calicut, Kerala, India 1.
2.
P. DILEEP KUMAR
Strandberg TE, Salomaa VV, Naukkarinen VA, Vanhanen HT, Sarna SJ, Miettinen TA. Long term mortality after 5 years multifactorial primary prevention of cardiovascular disease in middle aged men. JAMA 1991; 266: 1225-29. Frasure-Smith N. In-hospital symptoms of psychological stress as predictors of long-term outcome after acute myocardial infarction in men. Am J Cardiol 1991; 67: 121-27.
Chorionic villus sampling and limb abnormalities SIR,-Following reports of an association between chorionic villus sampling (CVS) and limb abnormalities,!,2an investigation was made in seven European Registration of Congenital Anomalies and Twins (EUROCAT) registries surveying altogether more than 600 000 births (table). These registries cover geographically defined populations. Malformations were registered for live births, stillbirths, and terminations after the diagnosis of malformation, except in Switzerland, where case-registration was restricted to live births. CVS exposure was recorded on the routine case-report forms of malformed babies, except in Odense, where registry data were matched to a known cohort of exposed women for this study. Where possible, extra information on the timing of exposure was sought. Chromosomal syndromes are not included in this report, since a high rate of exposure to CVS would be expected for such cases. Exposed cases with monogenic syndromes for which CVS could be indicated were also excluded. The exposure rate among cases of limb reduction is compared with the exposure rate among other anomalies in the table. 336 cases of limb reduction were reviewed, of which 4 (1-2%) had CVS exposure. In 11 883 cases of other congenital anomalies, 78 (066%) had CVS exposure, giving an odds ratio of 1 -8 (95% CI 0.66-4.99).
877
BIRTH POPULATIONS SURVEYED, TOTAL NUMBER AND NUMBER EXPOSEDTOCVSAMONG CASES OF LIMB REDUCTION AND OTHER MALFORMATIONS IN SEVEN EUROCAT REGISTRIES
The increase in risk is not statistically significant, the CI being wide. The 4 cases of limb reduction with CVS exposure were: (1) isolated agenesis of 2nd, 3rd, and 4th left (L) toes with hypoplasia of nails of remaining L toes (indication for CVS maternal age, CVS at 76 days of gestation, Paris); (2) partial amputation of four fingers of the L hand, ? amniotic bands (indication for CVS previous child with haemolytic uraemic anaemia, transcervical technique, CVS at 69 days, Switzerland); (3) amniotic disruption sequence with total aplasia of L leg, hypoplasia of L pelvis, L diaphragmatic hernia, imperforate anus, and vagina, pulmonary hypoplasia, kidney anomaly, persistent L superior vena cava, drainage of right (R) pulmonary vein into R atrium (indication for CVS was maternal age, CVS at 65 days of gestation, transcervical technique, Switzerland); and (4) Moebius syndrome with syndactyly of R and L hands, agenesis of the 2nd to 4th fingers of R hand, posterior cleft palate, micrognathism (indication for CVS family history of Duchenne muscular dystrophy, CVS at 79 days of gestation,
Strasbourg). We grouped together all cases of limb reduction for this analysis, but this might conceal an association with specific types of limb reduction. Limb reduction associated with Moebius syndrome is very rare, for example, and was also reported in the original Oxford series.1 3 of the 4 cases were exposed to CVS after 66 days of gestation, whereas Firth et aP hypothesised that the association would be restricted to CVS before 66 days. (Information on the timing of CVS in the seven populations is available from EUROCAT.) It has been suggested that the mechanism by which CVS leads to congenital anomalies might be a vascular insult, and that any defects thought to be due to hypoperfusion of the fetus should be investigated.l,2 Through a search for any additional exposed case with micrognathia, Moebius syndrome, destructive brain anomalies, or Poland anomaly the following were identified: (a) stillbirth with possible Hallermann-Streiff syndrome with microphthahnia, cleft palate, microretrognathia, anal atresia, tetralogy of Fallot, and absence of frontal bone (indication for CVS maternal chromosomal translocation [5;19], CVS at 11 completed weeks, Bouches-du-Rhone), (b) CHARGE syndrome with cleft lip and palate, atrioventricular canal, misshapen ear, syndactyly, facial nerve paralysis, micrognathism, sacral dimple, and deafness (indication for CVS maternal age, CVS at 57 days, transcervical technique [Portex and KC cannula], Groningen); and (c) multiple dysplasia of brain stem and diencephalon and L talipes equinovarus (indication for CVS maternal age, CVS at 63 days, transcervical technique, Switzerland). 1 further case of possible hypoperfusion anomaly (hydranencephaly) was excluded because CVS had been done following ultrasound indication of a brain anomaly. Thus a maximum of 7 exposed cases of possible hypoperfusion anomalies were recorded among the 82 non-chromosomal, nonmonogenic exposed cases (85%; 95% CI 2-5-14-6%). We can estimate from the EUROCAT database 1980-883that these hypoperfusion anomalies, including limb reduction, are expected to account for 8-9% of non-chromosomal anomalies. We cannot claim that CVS exposure was definitely known for all cases of limb reduction or hypoperfusion anomaly. Although reasonable levels of exposure ascertainment are suggested by the
of the CVS exposure figures between populataion estimates and estimates based on malformed cases in some registries, it is possible or even probable that the exposed population is at higher risk of malformation (other than chromosomal or genetic) than the unexposed. Where malformed cases were used as controls, CVS exposure would not be expected to be known less frequently in cases of limb reduction than in cases of other non-chromosomal anomalies. However, we cannot exclude the possibility that the CVS exposed population is at a higher (or lower) risk of limb reduction than of other malformations, independent of CVS. We have not attempted here to derive the prevalence of limb reduction defects in exposed pregnancies, since 1990 registry data were incomplete, and population data on CVS exposure was not known in some registries for an appropriate time period. EUROCAT data do not give any further evidence of a risk associated with early or late CVS, although the wide confidence limits around the odds ratio demonstrate the limited power of the study. They do suggest that, if there is a risk as high as that cited in Oxford (1% for all CVS, 1 ’7% for early CVS), it must be associated either with specific procedures or exposure timings which were infrequent in these EUROCAT registry areas or with specific anomaly types. H. DOLK EUROCAT, F. BERTRAND School of Public Health, M. F. LECHAT, UCL, for the EUROCAT Clos Chapelle-aux-Champs 30, B-1200 Brussels, Belgium working group*
similarity
* S. Aymé (Bouches-du-Rhône), D. Stone (Glasgow), H. de Walle (Groningen), E. Garne (Odense), J. Goujard and C. de Vigan (Paris), C. Stoll and B. Dott (Strasbourg), and T. Pexieder and D. Bloch (Switzerland).
HV, Boyd PA, Chamberlain P, Mackenzie IZ, Lindenbaum RH, Huson SM. Severe limb abnormalities after chorion villus sampling at 56-66 days’ gestation. Lancet 1991; 337: 762-63. 2. Mastroiacovo P, Cavalcanti DP. Limb reduction defects and chorion villus sampling. Lancet 1991; 337: 1091. 3. EUROCAT Working Group. Surveillance of congenital anomalies 1980-1988. EUROCAT Rep 1991; no 4: 24. 1. Firth
Cardiogenic embolism to the brain SIR,-Dr Hart’s contribution to your Stroke Octet (March 7, p 589) contains ambiguous, hazardous, and inaccurate target ranges
optimum oral anticoagulation, although the expression of the prolongation of the prothrombin times in terms of international normalised ratios (INR) is highly welcome. There is ambiguity in the so-called low-density anticoagulation: 2-3 INR is suggested for patients with myocardial infarction after hospital discharge, whereas for patients with a higher risk, the low-intensity range is presented as being 2-4 INR. We believe that the term low-intensity anticoagulation should no longer be used and should be replaced by moderate-intensity regimen as suggested by Hirsh.l Values of 4-7 INR as recommended for patients at high risk are highly hazardous. The use of such values would undoubtedly lead to an unduly large number of severe bleeding complications. The safe upper limit of the target range lies between 45 and 5 INR.2 Finally the quoted recommended range for patients enrolled in the European atrial fibrillation secondary prevention trial is inaccurate. This range is for
2.5-4 INR and not 2-4 INR.2 May I emphasise once again that the combination of oral anticoagulants with dipyridamole as recommended by Hart for many patients with mechanical prosthetic heart valves, has never been shown irrefutably to surpass the effect of anticoagulants
alone.3.4 Department of Haematology, Academisch Ziekenhuis, Postbus 9600, 2300 RC Leiden, Netherlands
E. A. LOELIGER
1. Hirsh J Oral anticoagulants. New Engl J Med 1991; 324: 1865-75. 2. Loeliger EA. Therapeutic target values in oral anticoagulation justification of Dutch policy. Ann Hematol 1992; 64: 60-65. 3. Loeliger EA. Does dipyridamole have antithrombotic potential? Thromb Haemostas 1985; 53: 437 4. FitzGerald GA. Dipyridamole. N Engl JMed 1987; 316: 1247-57.