Editor,—The question, outlined by Meadow,1-1 of which infant deaths have an unnatural cause is primarily a problem for the coroner’s pathologist rather than the paediatrician. The great majority of these infants never come to the attention of a paediatrician before or after death, and it is the pathologist who is required to advise the coroner as to what should be entered on the death certificate. As someone who has had a specialised interest in this task for over a quarter of a century, the frequency of suspicion of homicide in the absence of physical injury is probably no more than 1% of the total.

I am aware of the argument that if I do not know the cause of death I cannot know that the infant has not been gently smothered. On the other hand, the principle that guilt has to be proven is a good one. Moreover, the curious uniformity of the clinical history and lack of pathology in these cases is a strong argument in favour of regarding sudden infant death syndrome (SIDS) as an entity distinct from homicide. Where either the history or the pathology is not typical of SIDS, it is always prudent to record that the cause of death is “unascertained”. For these relatively rare cases, it is usually impossible for the pathologist, police or coroner to take matters further. Infant smothering is an unwitnessed crime. Confessions are seldom volunteered and such is parental guilt surrounding any child’s death that it is foolish to accept a confession without corroborative evidence. I remain sceptical that paediatricians are better at detecting homicide than an experienced member of the police service, especially as in most cases the child has never been a hospital patient. I wonder whether Professor Meadow would be enthusiastic about involving physicians in the investigation of possible homicide in unexplained deaths in adults. Dr Watson never showed the same aptitude for the task as Sherlock Holmes!

  Editor,—The warning to meta-analysts published at the end of Professor Meadow’s paper2-1 was intriguing. The paper reports a highly selected, uncontrolled case series collected over 18 years with no reference to a denominator population. The paper would be excluded from any meta-analysis with appropriate inclusion–exclusion criteria independent of whether the same cases appeared in similar uncontrolled, highly selected case series.2-2 Uncontrolled case series collected by tertiary specialists occupy the lowest level in the hierarchy of evidence based medicine. They have been responsible for misleading conclusions which, when taken up by the media, have had unfortunate consequences for child health. Two of the most notorious examples are the pertussis vaccination controversy in the 1970s2-3 and the more recent publication of a highly selected case series claiming to show a link between measles–mumps–rubella vaccination and autism.2-4

Professor Meadow’s clinical descriptions and insights are fascinating and potentially useful to paediatricians. His suggestion that all unexpected infant deaths should be fully investigated including a “death scene investigation” is appropriate and justified on the basis of the cases he describes. However, his statement that “Currently, many paediatric units are failing to heed warning signs and failing to protect some very vulnerable children” is impossible to interpret in the absence of denominator data in his study and the protracted period over which the cases were collected. Professor Meadow’s case series is a seriously flawed basis from which to mount an attack on the integrity of SIDS as a diagnostic category. Unfortunately, it is these latter conclusions of his paper that have received media attention and which may have the effect of causing unnecessary further distress to bereaved parents. A sound estimate of the proportion of sudden unexpected infant deaths attributable to, or suspicious of, infanticide can be obtained only by properly designed controlled studies within a defined population such as the CESDI (confidential enquiry into still-births and deaths in infancy) study.2-5 Unjustified and sweeping conclusions based on highly selected and uncontrolled case series are not only scientifically flawed but are potentially seriously misleading.

Professor Sir Roy Meadow comments:

As Professor Davis infers, it would be convenient and less stressful for us all to let sleeping dogs lie, and to avoid confronting the problem of infanticide. However, paediatricians have a responsibility to the other children in these families. Half of the 50 families whom I reported included more than one dead child, and within most of the other families there were children who incurred serious non-fatal abuse. Moreover, most of the 81 dead children had, in retrospect, features that could have led to intervention preventing their death. Therefore, we have to improve our recognition of children at great risk to protect them; and improve our recognition of infanticide to protect the siblings.

Professor Davis and Dr Barson agree with me about the importance of necropsy being performed by a skilled pathologist experienced with infants. However, we have moved on from the era when morbid pathology provided the gold standard for diagnosis. Today, clinical diagnoses are based primarily on the history, and supplemented by findings on examination and investigation. Indeed, many paediatricians base their diagnoses far more on the history than on examination or investigation findings. It is illogical to jettison that central basis of diagnosis, the history, when considering why a child has died. The history should be assessed by a paediatrician, because paediatricians are familiar with histories of childhood illness. Pathologists who are most expert at necropsies and tissue morphology may have had only eight weeks’ experience as an undergraduate learning about paediatric diagnosis and illness. Criminal investigation department officers and coroners officers, for all their skills in other areas, have had even less experience. It is improper and unfair to give pathologists the responsibility for providing a quick diagnosis without the aid of a competent clinical history.

Amidst the hate mail that I have received recently have been some thoughtful letters from bereaved parents. Some have written of their surprise at the “casual and lax way” in which the death of their baby was investigated: “we really did feel that experts and the police should have looked into our house and our affairs much more thoroughly than they did and then we would have felt absolved of guilt.” Others have written of their greatest sadness that they still do not understand why their child died: “if only there had been a clear reason we would have felt so much better.” Those clear reasons, usually natural, but sometimes unnatural, are much more likely to come to light if all sudden deaths are subject to a multidisciplinary assessment, including a visit to the death scene, with involvement of not just an experienced pathologist, but also a paediatrician, general practitioner, health visitor, and police officer.

My report was aimed at paediatricians to alert them to some of the features that may enable earlier detection of young children who are likely to be killed by their parents. The note at the end about the dangers of meta-analysis was based on past experience of having some of my observational studies inappropriately included in meta-analysis reports, and incorrectly reported in review articles. I have made clear in many publications, and to the media, that my work does not allow conclusions to be drawn about the proportionof infant deaths that are caused by parents. The most recent estimate from CESDI and other studies suggest that parental acts may be the main, or a contributory, cause of between 8% and 18% of infant deaths. That is a substantial number of deaths. Not only those children but also their vulnerable siblings, deserve better assessment and care.