1005
EDITORIALS
Provocation
injections and their site are seldom well documented polio-prone countries. The mother’s recall is usually the only means of assessing injections, and recall bias makes interpretation of the results almost impossible. 10 The mother of a child who is paralysed is more likely to remember injections during the last month than the mother of a healthy child; in addition she may mistakenly remember the injection site as being in the paralysed limb. The information gap has now been filled by a good study conducted during the recent epidemic in Oman,l1 where high-quality in
paralysis
At the beginning of this century Hochhaus1 noted that the affected limb at onset of paralytic poliomyelitis had commonly been the site of an
injection. Sporadic reports of the phenomenon then appeared up to the 1940s. These observations were made by astute clinicians, but it is noteworthy that mothers in Samoa2 recognised the sequence when paralysis followed a yaws treatment campaign (the physicians on the spot dismissed it as chance). The introduction of widespread vaccination with diphtheria and pertussis vaccines in the 1940s led to an outbreak of reports. 3-5 Although these clearly showed an association between the limb injected and the limb paralysed, researchers could not define whether the association was causal or, if it was, the size of the risk. The UK Medical Research Council set up a committee to investigate the matter, and the two papers that emerged6have until lately been the best evidence. The term provocation seems to have been used first by Bradford Hill,8reporting on some of these data at a conference in 1951. In the first paper6Hill outlined the information that was required to decide whether the association between injections and paralysis was fortuitous or true-"the numbers of children of given ages injected with defined antigens at given points of time and in specified areas, and then the number of these who develop poliomyelitis within subsequent periods". He commented "Such a method is the only means by which those risks can be satisfactorily and effectively measured". This was the study that was conducted by the MRC and it showed, firstly, that injections resulted in an altered distribution of paralysis, with the upper (usually injected) limb being affected more frequently than in natural disease; secondly, that the risk was limited to injections given in the 30 days before paralysis; and, thirdly, that there was no difference in risk whether the injection was subcutaneous or intramuscular. No added risk was found with smallpox inoculation and it was not possible on the numbers studied (6871 children with paralytic polio) to distinguish variation in risk with different diphtheria and pertussis vaccines, or combinations of these. These findings were clearly of great importance to developing countries-particularly with regard to injections other than vaccinations, since these are so commonly given to children in the tropics. There have been anecdotal reports of an association with quinine and penicillin9 but hard evidence is scarce because
the immunisation records allowed accurate of DTP (diphtheria/tetanus/pertussis) vaccinations before onset in children with paralysis and in ten controls for each paralysed child. The findings are similar to those in developed countries in the 1940s and 1950s-namely, that immunisations were significantly more common in the period 7-21 days before onset of paralysis. In this study the researchers also estimated the proportion of cases in the population that could be attributed to these injections; it was 15%. They claim that this is the first estimate of attributable risk, but in fact the MRC committee calculated from their cohort study that the paralysis was causally related to diphtheria or pertussis injections in 13% of children aged 6 months to 2 years with polio-a strikingly similar figure. Both are likely to be underestimates since undocumented non-vaccine injections may have contributed. The mechanism is still unclear. Contamination of vaccine batches and of the needle and syringe have been excluded as possibilities. Two hypotheses are current-first, that the injection induces reflex hyperaemia in the spinal cord associated with the area injected, leading to localisation of poliovirus there;12Z second, that the damage to nerve endings by the injection allows the virus access to the nervous system.13 The mechanism is less important than the public health implications. So far as possible, injections must be avoided in countries with endemic poliomyelitis. If an outbreak is detected there may be a case for suspending the programme of injected vaccines. The risk of paralysis in these circumstances must be weighed against the risk of diseases the vaccination might prevent. In this regard it is unfortunate that there are no data on the risk of paralysis associated with measles immunisation. This picture has been slightly fogged by a study in India.14 Wyatt and co-workers claim that paralysis was associated with injections received as little as 48 hours before onset of paralysis, offering evidence of an association of the site of injection with the limb paralysed and a higher frequency of injection than in controls. Unfortunately the documentation of injections seems to have depended on the mother’s history as recorded in the medical notes and is therefore open to recall bias. In the UK case-control6 and cohort studies’ there was no excess of injections in the 4 days before onset of paralysis; in the Oman assessment
studyll a small excess (not statistically significant) was
1006
found but was put down to the practice of giving vaccines at every opportunity, including consultations for minor febrile illnesses that might be the prodrome of paralytic poliomyelitis. Whether or not the Indian findings are borne out by other work, the public health message is unchanged: avoid unnecessary injections. 1. Hochhaus. Ueber Poliomyelitis acuta. Munch Med Wochenschr 1909; 56: 2353-55. 2. Lambert SM. A yaws campaign and an epidemic of poliomyelitis in Western Samoa. J Trop Med Hyg 1936; 39: 41-46. 3. Geffen DH. The incidence of paralysis occurring in London children within four weeks after immunization. Med Officer 1950; 83: 137-40. 4. McCloskey BP. The relation of prophylactic inoculations to the onset of poliomyelitis. Lancet 1950; i: 659-63. 5. Martin JK. Local paralysis in children after injections. Arch Dis Child 1950; 25: 1-14. 6. Hill AB, Knowelden J. Inoculation and poliomyelitis. A statistical investigation in England and Wales in 1949. BMJ 1950; ii: 1-6. 7. Medical Research Council Committee on Inoculation Procedures and Neurological Lesions. Poliomyelitis and prophylactic inoculation. Lancet 1956; ii: 1223-31. 8. Wyatt HV. Provocation poliomyelitis: neglected clinical observations from 1914 to 1950. Bull Hist Med 1981; 55: 543-57. 9. Townsend-Coles WF, Findlay GM. Poliomyelitis in relation to intramuscular injections of quinine and other drugs. Trans R Soc Trop Med Hyg 1953; 47: 77-81. 10. Guyer B, Bisong AAE, Gould J, Brigaud M, Aymard M. Injections and paralytic poliomyelitis in tropical Africa. Bull WHO 1980; 58: 285-91. 11. Sutter RW, Patriarca PA, Suleiman AJM, Brogan S, et al. Attributable risk of DTP (diphtheria and tetanus toxoids and pertussis vaccine) injection in provoking paralytic poliomyelitis during a large outbreak in Oman. J Infect Dis 1992; 165: 444-49. 12. Bodian D. Viremia in experimental poliomyelitis. II. Viremia and the mechanism of the "provoking" effect of injections or trauma. Am J Hyg 1954; 60: 358-70. 13. Wyatt HV. Incubation of poliomyelitis as calculated from the time of entry into the central nervous system via the peripheral nerve pathways. Rev Infect Dis 1990; 12: 547-56. 14. Wyatt HV. Mahadevan S, Srinavasan S. Unnecessary injections and paralytic poliomyelitis in India. Trans R Soc Trop Med Hyg 1992; 86: 546-49.
Screening for fetal
malformations
discussions about screening the general is assumption that, because a test is possible, testing should be implemented as a service ...".1 This statement appeared in an editorial about prenatal diagnosis of cystic fibrosis but something similar could be said of ultrasound screening for fetal abnormalities. However, unlike testing for cystic fibrosis, ultrasound screening is already done in many maternity units and is not widely regarded as a controversial issue worthy of debate. Thus the offer of routine ultrasound screening for fetal abnormality has been recommended by both the Joint Study Group on Fetal Abnonnalities2 and a study group of the Royal College of Obstetricians and Gynaecologists.3 We need to remember that prenatal diagnosis is a double-edged sword, able to predict and (through abortion) prevent handicapping conditions, but capable also of generating severe and unnecessary anxiety. Take the case of a woman who undergoes a routine ultrasound scan at ten weeks’ gestation. Because she is slim, the embryo is imaged with unusual clarity; an anterior abdominal wall defect is suspected. After two weeks of worry, repeat ultrasound examination shows that the initial "In
most
appearances can be attributed to the normal extrusion of intestine into the umbilical cord in early pregnancy. But now there is obvious nuchal oedema, which is
known to be associated with chromosomal abnormalities.4Amniocentesis is done at fourteen weeks, karyotypic normality is confirmed at seventeen weeks; the offer of a further ultrasound scan is politely but firmly declined at eighteen weeks. Labour is induced at forty weeks because of continuing parental anxiety and a perfectly normal baby is delivered. Many women now seek a detailed scan during mid-pregnancy. Their requests are presumably based on the assumption that ultrasound is effective both in indicating normality and in detecting abnormality. The first reports of the effectiveness of ultrasound for detection of abnormality came from teaching hospitals with a research interest in obstetric ultrasound. Good results were subsequently reported from district general hospitals,5-7 although these may also be "centres of excellence" in the technique; that the information was collected, analysed, and reported indicates an unusual commitment to detection of fetal abnormality and audit of performance. Other accounts point to large differences in detection rates of abnormalities by different departments.8,9 It would be surprising if this were not so, because in many countries obstetric ultrasound has only a vestigial training structure and very little quality control. Data from the UK were sorely in need of regional audit to assess overall performance, to include both the good and the bad. This exercise has been done in the Northern Region with painstaking and praiseworthy thoroughness. The results make uncomfortable reading.10 Wide variation in performance between different units was confirmed, and the variation did not seem to be linked to unit size, type of equipment, or amount of ultrasound work done. There was a trend to improved detection rates during the years of the survey (1985-90), with a consequent lowering of the perinatal mortality rate. For several serious conditions, more than 10% of diagnoses were completely wrong. For example, the diagnosis of duodenal atresia was almost as likely to be wrong as to be correct; this abnormality is strongly associated with Down’s syndrome. One in three and one in five of fetuses diagnosed as having, respectively, bladder outlet obstruction and lethal short-limbed dwarfism were, in reality, unaffected. Diaphragmatic hernia, isolated hydrocephalus, and encephalocele also caused diagnostic difficulties. Exhortations ex cathedra for universal routine anomaly scanning are not supported by this report. We still need to consider carefully the value of this approach. Some sonologists in top-class ultrasound departments are not sufficiently convinced of the merits of routine mid-pregnancy ultrasonography to have adopted this policy; they have retained instead the package of "booking" visit ultrasound with mid-pregnancy alpha-fetoprotein estimation. There are also those who, with good reason, fear the
EDITORIALS
Provocation
injections and their site are seldom well documented polio-prone countries. The mother’s recall is usually the only means of assessing injections, and recall bias makes interpretation of the results almost impossible. 10 The mother of a child who is paralysed is more likely to remember injections during the last month than the mother of a healthy child; in addition she may mistakenly remember the injection site as being in the paralysed limb. The information gap has now been filled by a good study conducted during the recent epidemic in Oman,l1 where high-quality in
paralysis
At the beginning of this century Hochhaus1 noted that the affected limb at onset of paralytic poliomyelitis had commonly been the site of an
injection. Sporadic reports of the phenomenon then appeared up to the 1940s. These observations were made by astute clinicians, but it is noteworthy that mothers in Samoa2 recognised the sequence when paralysis followed a yaws treatment campaign (the physicians on the spot dismissed it as chance). The introduction of widespread vaccination with diphtheria and pertussis vaccines in the 1940s led to an outbreak of reports. 3-5 Although these clearly showed an association between the limb injected and the limb paralysed, researchers could not define whether the association was causal or, if it was, the size of the risk. The UK Medical Research Council set up a committee to investigate the matter, and the two papers that emerged6have until lately been the best evidence. The term provocation seems to have been used first by Bradford Hill,8reporting on some of these data at a conference in 1951. In the first paper6Hill outlined the information that was required to decide whether the association between injections and paralysis was fortuitous or true-"the numbers of children of given ages injected with defined antigens at given points of time and in specified areas, and then the number of these who develop poliomyelitis within subsequent periods". He commented "Such a method is the only means by which those risks can be satisfactorily and effectively measured". This was the study that was conducted by the MRC and it showed, firstly, that injections resulted in an altered distribution of paralysis, with the upper (usually injected) limb being affected more frequently than in natural disease; secondly, that the risk was limited to injections given in the 30 days before paralysis; and, thirdly, that there was no difference in risk whether the injection was subcutaneous or intramuscular. No added risk was found with smallpox inoculation and it was not possible on the numbers studied (6871 children with paralytic polio) to distinguish variation in risk with different diphtheria and pertussis vaccines, or combinations of these. These findings were clearly of great importance to developing countries-particularly with regard to injections other than vaccinations, since these are so commonly given to children in the tropics. There have been anecdotal reports of an association with quinine and penicillin9 but hard evidence is scarce because
the immunisation records allowed accurate of DTP (diphtheria/tetanus/pertussis) vaccinations before onset in children with paralysis and in ten controls for each paralysed child. The findings are similar to those in developed countries in the 1940s and 1950s-namely, that immunisations were significantly more common in the period 7-21 days before onset of paralysis. In this study the researchers also estimated the proportion of cases in the population that could be attributed to these injections; it was 15%. They claim that this is the first estimate of attributable risk, but in fact the MRC committee calculated from their cohort study that the paralysis was causally related to diphtheria or pertussis injections in 13% of children aged 6 months to 2 years with polio-a strikingly similar figure. Both are likely to be underestimates since undocumented non-vaccine injections may have contributed. The mechanism is still unclear. Contamination of vaccine batches and of the needle and syringe have been excluded as possibilities. Two hypotheses are current-first, that the injection induces reflex hyperaemia in the spinal cord associated with the area injected, leading to localisation of poliovirus there;12Z second, that the damage to nerve endings by the injection allows the virus access to the nervous system.13 The mechanism is less important than the public health implications. So far as possible, injections must be avoided in countries with endemic poliomyelitis. If an outbreak is detected there may be a case for suspending the programme of injected vaccines. The risk of paralysis in these circumstances must be weighed against the risk of diseases the vaccination might prevent. In this regard it is unfortunate that there are no data on the risk of paralysis associated with measles immunisation. This picture has been slightly fogged by a study in India.14 Wyatt and co-workers claim that paralysis was associated with injections received as little as 48 hours before onset of paralysis, offering evidence of an association of the site of injection with the limb paralysed and a higher frequency of injection than in controls. Unfortunately the documentation of injections seems to have depended on the mother’s history as recorded in the medical notes and is therefore open to recall bias. In the UK case-control6 and cohort studies’ there was no excess of injections in the 4 days before onset of paralysis; in the Oman assessment
studyll a small excess (not statistically significant) was
1006
found but was put down to the practice of giving vaccines at every opportunity, including consultations for minor febrile illnesses that might be the prodrome of paralytic poliomyelitis. Whether or not the Indian findings are borne out by other work, the public health message is unchanged: avoid unnecessary injections. 1. Hochhaus. Ueber Poliomyelitis acuta. Munch Med Wochenschr 1909; 56: 2353-55. 2. Lambert SM. A yaws campaign and an epidemic of poliomyelitis in Western Samoa. J Trop Med Hyg 1936; 39: 41-46. 3. Geffen DH. The incidence of paralysis occurring in London children within four weeks after immunization. Med Officer 1950; 83: 137-40. 4. McCloskey BP. The relation of prophylactic inoculations to the onset of poliomyelitis. Lancet 1950; i: 659-63. 5. Martin JK. Local paralysis in children after injections. Arch Dis Child 1950; 25: 1-14. 6. Hill AB, Knowelden J. Inoculation and poliomyelitis. A statistical investigation in England and Wales in 1949. BMJ 1950; ii: 1-6. 7. Medical Research Council Committee on Inoculation Procedures and Neurological Lesions. Poliomyelitis and prophylactic inoculation. Lancet 1956; ii: 1223-31. 8. Wyatt HV. Provocation poliomyelitis: neglected clinical observations from 1914 to 1950. Bull Hist Med 1981; 55: 543-57. 9. Townsend-Coles WF, Findlay GM. Poliomyelitis in relation to intramuscular injections of quinine and other drugs. Trans R Soc Trop Med Hyg 1953; 47: 77-81. 10. Guyer B, Bisong AAE, Gould J, Brigaud M, Aymard M. Injections and paralytic poliomyelitis in tropical Africa. Bull WHO 1980; 58: 285-91. 11. Sutter RW, Patriarca PA, Suleiman AJM, Brogan S, et al. Attributable risk of DTP (diphtheria and tetanus toxoids and pertussis vaccine) injection in provoking paralytic poliomyelitis during a large outbreak in Oman. J Infect Dis 1992; 165: 444-49. 12. Bodian D. Viremia in experimental poliomyelitis. II. Viremia and the mechanism of the "provoking" effect of injections or trauma. Am J Hyg 1954; 60: 358-70. 13. Wyatt HV. Incubation of poliomyelitis as calculated from the time of entry into the central nervous system via the peripheral nerve pathways. Rev Infect Dis 1990; 12: 547-56. 14. Wyatt HV. Mahadevan S, Srinavasan S. Unnecessary injections and paralytic poliomyelitis in India. Trans R Soc Trop Med Hyg 1992; 86: 546-49.
Screening for fetal
malformations
discussions about screening the general is assumption that, because a test is possible, testing should be implemented as a service ...".1 This statement appeared in an editorial about prenatal diagnosis of cystic fibrosis but something similar could be said of ultrasound screening for fetal abnormalities. However, unlike testing for cystic fibrosis, ultrasound screening is already done in many maternity units and is not widely regarded as a controversial issue worthy of debate. Thus the offer of routine ultrasound screening for fetal abnormality has been recommended by both the Joint Study Group on Fetal Abnonnalities2 and a study group of the Royal College of Obstetricians and Gynaecologists.3 We need to remember that prenatal diagnosis is a double-edged sword, able to predict and (through abortion) prevent handicapping conditions, but capable also of generating severe and unnecessary anxiety. Take the case of a woman who undergoes a routine ultrasound scan at ten weeks’ gestation. Because she is slim, the embryo is imaged with unusual clarity; an anterior abdominal wall defect is suspected. After two weeks of worry, repeat ultrasound examination shows that the initial "In
most
appearances can be attributed to the normal extrusion of intestine into the umbilical cord in early pregnancy. But now there is obvious nuchal oedema, which is
known to be associated with chromosomal abnormalities.4Amniocentesis is done at fourteen weeks, karyotypic normality is confirmed at seventeen weeks; the offer of a further ultrasound scan is politely but firmly declined at eighteen weeks. Labour is induced at forty weeks because of continuing parental anxiety and a perfectly normal baby is delivered. Many women now seek a detailed scan during mid-pregnancy. Their requests are presumably based on the assumption that ultrasound is effective both in indicating normality and in detecting abnormality. The first reports of the effectiveness of ultrasound for detection of abnormality came from teaching hospitals with a research interest in obstetric ultrasound. Good results were subsequently reported from district general hospitals,5-7 although these may also be "centres of excellence" in the technique; that the information was collected, analysed, and reported indicates an unusual commitment to detection of fetal abnormality and audit of performance. Other accounts point to large differences in detection rates of abnormalities by different departments.8,9 It would be surprising if this were not so, because in many countries obstetric ultrasound has only a vestigial training structure and very little quality control. Data from the UK were sorely in need of regional audit to assess overall performance, to include both the good and the bad. This exercise has been done in the Northern Region with painstaking and praiseworthy thoroughness. The results make uncomfortable reading.10 Wide variation in performance between different units was confirmed, and the variation did not seem to be linked to unit size, type of equipment, or amount of ultrasound work done. There was a trend to improved detection rates during the years of the survey (1985-90), with a consequent lowering of the perinatal mortality rate. For several serious conditions, more than 10% of diagnoses were completely wrong. For example, the diagnosis of duodenal atresia was almost as likely to be wrong as to be correct; this abnormality is strongly associated with Down’s syndrome. One in three and one in five of fetuses diagnosed as having, respectively, bladder outlet obstruction and lethal short-limbed dwarfism were, in reality, unaffected. Diaphragmatic hernia, isolated hydrocephalus, and encephalocele also caused diagnostic difficulties. Exhortations ex cathedra for universal routine anomaly scanning are not supported by this report. We still need to consider carefully the value of this approach. Some sonologists in top-class ultrasound departments are not sufficiently convinced of the merits of routine mid-pregnancy ultrasonography to have adopted this policy; they have retained instead the package of "booking" visit ultrasound with mid-pregnancy alpha-fetoprotein estimation. There are also those who, with good reason, fear the
