tion before being labelled as sudden infant death syndrome ... 3 ELncry Jl. Itfanticide, lilicidc, and cot death. ... I wish the term "sudden infant death syndrome".
literary dysplasia). The reaction on detecting such a wrong varies from a mild sense of self satisfaction to a feeling of horror at the actual and potential consequences of that erroneous quotation. In "The ABC of Child Abuse" Professor Roy Meadow has written on suffocation,' stating that "Between 2 and 10% of babies currently labelled as dying from the sudden infant death syndrome have probably been smothered by their mothers." This sentence contains two linked errors. The quotation would appear to have arisen (although this is not apparently acknowledged2) in an article by John Emery in which he states "In a random series of 50 deaths, there were 5 in which the question of unnatural death was raised. In a more superficial assessment of a further 120 deaths presenting as cot deaths in other centres the suspicion occurred in only two. Thus where one is looking at the cases in great detail the rate could be as high as one in 10, and where looked at on a more superficial level it seems to be around one in 50. These are not court proved cases, simply deaths that required further study" (my italics).' Somehow this statement has become transmogrified into something quite different and has acquired a certain statistical sanctity. What Emery may be saying is that a small number of children whose deaths are sudden and unexpected require unusually detailed investigation before being labelled as sudden infant death syndrome, not that they are victims of homicide perpetrated by their mothers or anyone else. I find it difficult to react to a distortion of this nature, and the potentially catastrophic consequences to parents arising therefrom, with anything other than horror. We must carefully check the original statement before misusing its content to further our own arguments. J M N HILTON State Hcalth Laboratorv Services, Queen Elizahbeth 11 Mcdical Ccntrc,
Ncdlands, WVA 60)09. Austrlial 1 Meadow R. Sutff6oationt. Br Aled 7 1989;298:1572-3. 1 ()J C.) 2 M4eadlow R. Rccurretit cot dcath altdl suiflfocation. Arch Dis (Cild 1989;64: 179-80. 3 ELncry Jl. Itfanticide, lilicidc, and cot death. Arch I)is (ithld 1985 ;60:505-7.
of cot deaths caused by parents their answer has always been within that 2-10% range. Unsurprisingly, their estimates have varied; the proportion will vary from country to country and from time to time. Southall and colleagues have made an important contribution in filming some of the ways in which mothers suffocate their children. However, their work is of limited epidemiological value in relation to cot death. They are studying recurrent apnoea, which precedes less than 10% of cot deaths. Moreover, within the group of mothers who repetitively smother to cause apnoea only a minority will do so while in hospital and thus could be caught in the act. Most child abuse occurs at home. None the less their 2% prevalence is worryingly high. The upset to innocent parents caused by media reports of suffocation and cot death is very sad. It raises questions about how one can teach about, and prevent, child abuse without causing distress. Unfortunately, a small proportion of all childhood illness, fractures, burns, scalds, and deaths is caused directly by parents. Unless we record that, and teach medical students and junior doctors about it, the suffering will not be detected or prevented. I wish the term "sudden infant death syndrome" could be abolished. Created to help parents, it has become a barrier to preventing death and disability of children. A syndrome is a characteristic collection of symptoms and signs, yet this particular label is given when there are no symptoms and no signs. It is a non-diagnosis. The incidence of cot death (2 per 1000) is about the same as it was 20 years ago, while most other indices of childhood morbidity and mortality have improved greatly during that period. One reason is that the syndrome label is being used inappropriately as a diagnosis: it ought to be the starting point for sensible (and sensitive) inquiry and more research. Nearly a quarter of all infant deaths in the United Kingdom are sudden and unexplained. In contrast to many of the conjectured causes, suffocation is worth detecting because successful intervention is often possible. But we do have to face up to its existence. ROY MEADOW
AUTHOR'S REPLY,-That suffocation of children occurs is in fact not conjecture. The evidence from court reports and the medical literature verifies the experience of those of us who have to deal with these sad families. Mr Tallon's incredulity is understandable, but it does seem to be the mother who is the perpetrator. From my own experience of over 25 proved cases of suffocation it was always the mother. An international survey of parentally induced apnoea found the probable perpetrator to be the mother in 22 cases and the father in one (M J Light, M S Sheridan, unpublished data). This unusual sex incidence (for child abuse) is similar to that for parentally induced illness (Munchausen syndrome by proxy). There is no doubt that suffocation is causing death, brain damage, and some terrible ordeals for children. It is a serious form of child abuse which we all hope is uncommon. Its extent is uncertain, as is its importance in relation to sudden infant death syndrome (cot death). Though well documented cases of cot death are subsequently found to be caused by suffocation, the extent to which it occurs is best addressed by those who have studied the epidemiology of cot deaths. The most authoritative review of this difficult subject is by Emery,' who refers to his own work and that of others and concludes that "the figures for filicide (child homicide by a parent) as a major factor in unexplained cot deaths, are between one in 10 and one in 50." In recent years a number of experts on cot death from several countries have visited me to discuss suffocation, and when I have asked them for their confidential opinion about the proportion
BMJ
VOLUME
299
15
JULY
1989
Department of 1'aediatrics and Child Health, St James's University Hospital, Leeds LS9 TIE I Meadow R. Not so sudden infant death [Abstracti. Arch Dis Child (in press). 2 MNleadow R. MIunichaLtsen syndrome by proxy. Arch Di's Child
1982;57:92-8. 3 Emery JL. Inf:anticide, filicide, and cot death. Archi Dis Child 1985;60:505-7.
Myocardial infarction in Asian men SIR, - Dr L 0 Hughes and his colleagues estimated that the incidence of first myocardial infarction in Harrow was 4 9 times higher in south Asian than in European men during 1985-7. 'If true, the implications of this would be startling. A relative risk of 4 9 applied to national mortality would mean that 37% of south Asian men could expect to die of coronary heart disease by age 65 unless the case fatality rate was lower in this group. We reported both morbidity2 and mortality' from coronary heart disease in the south Asian (mainly Gujarati) population of Brent and Harrow during 1979-83. We found that the rate of admission for myocardial infarction was 1 4 times higher and the mortality from coronary heart disease 1 6 times higher in south Asian men than in the general population. Although these rates are exceptionally high, the relative risks are less extreme than those reported by Dr Hughes and his colleagues. We re-examined Hospital Activity Analysis data for North West rhames region to try to identify
reasons for the discrepancy between our results and those of Dr Hughes and his colleagues. Using country of birth and religion to assign ethnic group, we estimated that the ethnic breakdown of men aged 30-69 admitted to Northwick Park Hospital in 1982-4 with a principal diagnosis of myocardial infarction was as follows: 12% south Asian, 66%/0 of other ethnic origin, and 22% unknown. This contrasts with their report that south Asians accounted for 41% of men presenting with a first myocardial infarction. The discrepancy may lie in the methods used to exclude patients with evidence of a previous infarction. We estimate that south Asian men aged 30-64 resident in Harrow have a relative risk of admission for myocardial infarction of 1 8 compared with the general population of the borough. Although we agree that the prevalence of coronary heart disease in Gujaratis in north west London is among the highest in the world, the estimate of a fivefold relative risk is not compatible with other population based data. P M McKEIGUE MICHAEL MARMOT
I)epartment of Community Medicine, University College and Middlesex School of Medicine, London WC I E 6EA I Hughes LO, Raval U, Raftery EB. First myocardial infarctions in Asian and white men. BrWcd_7 1989;298:1345-50. (20 Mav.) 2 McKeigue PM, Marmot M(wi, Adelstein AM, et al. Diet and risk factors for coronary heart disease in Asians in north-west London. Lancet 1985;ii: 1086-90. 3 McKeigue P'M, Mlarmot MG. Mortality from coronary hcart Med 7 disease in Asian communitics in London. Br 19A5; 297:903.
AUTHORS'REPLY, - Dr McKeigue and Professor Marmot raise an important point. The calculation of relative rates of infarction using either Hospital Activity Analysis or the methods used in our study is subject to the sources of error of the method. Hospital Activity Analysis can be mistaken in defining ethnic origin: indeed, had we used it, five Asian patients in our study would (on the basis of their place of birth, religion, and surnames) have been misclassified as European. Also, the documented diagnosis may be wrong, and as we pointed out in our publication, the frequent atypical presentation of infarction in Asians could lead to underdiagnosis. Finally, any method that fails to classify 22% of patients (another indication of how punctiliously house officers, nursing staff, and ward clerks complete Hospital Activity Analysis forms) must be viewed with caution. Even so we fully accept the dangers of extrapolating the results of a small hospital based study to the country as a whole. Moreover, the discrepancy between Hospital Activity Analysis and our methods may, of course, be in part due to our own inaccuracy. In particular, we calculated relative rates from patients who were admitted to the coronary care unit, which allowed us to be certain of the diagnosis and ethnic origin but may have been a source of bias in our case selection. Although an attempt was made to assess bias of general practitioners, the heightened awareness of coronary disease in Asians (largely due to the work of Dr McKeigue and Professor Marmot) may have led to preferential referral of Asians. Moreover, within hospital referral to the coronary care unit of Asians may have been more likely because of their younger age at presentation. Nevertheless, during the period of study 36% of all men admitted to the unit were of southern Asian origin, and the finding of 41% of all new patients with infarction being Asian is compatible with this figure. Other hospital based studies (namely, the small study by Padhani and Dandona,' and the Birmingham study presented to the Heart Foundation workshop by Professor Littler) both estimated rates in excess of those calculated from population based data, which suggests a consistent discrepancy between these two approaches. A more exact quantification of the excess morbidity from
179