J F Geddes, retired (formerly reader in clinical neuropathology, Queen Mary, University of London)
London ; Email: email@example.com
J Plunkett, forensic pathologist
Regina Medical Center, 1175 Nininger Road, Hastings, MN 55033, USA
It is difficult to understand how Reece et al could interpret our editorial as displaying "a worrisome and persistent bias against the diagnosis of child abuse in general." Child abuse exists, and we know and attest that it exists. In fact, one of us (JP) has testified before the Minnesota State House and Senate on the need for "mandated reporting" laws. The editorial does not discuss "child abuse in general."
Child abuse exists in many forms: our editorial addresses the diagnostic criteria for a specific type of abuse, the so-called shaken baby syndrome. We emphasise, as have Donohoe and Lantz et al, that the literature to support a diagnosis of shaken baby syndrome/inflicted head injury is based on imprecise and ill-defined criteria, biased selection, circular reasoning, inappropriate controls, and conclusions that overstep the data. If it is the questioning of the criteria that is worrisome, we will continue to do so and to cause worry.
We omit "the most important element in this condition: brain injury itself" because there is little scientifically acceptable evidence that "shaking" causes primary traumatic brain injury except under laboratory conditions. Detailed neuropathological studies have shown that apart from the craniocervical junction, the brain is seldom damaged even in those deaths thought to be due to "shaking." Furthermore, the "research" referenced by the authors is misstated, misused, and remarkably selective. The diagnosis of non-accidental trauma in the authors’ cited studies is based on "assessment of the child protection committee," "multidisciplinary child protective team…or the forensic pathologist of the state medical examiner," or "multidisciplinary team consensus." We would no longer even use "the inclusion criteria of Geddes et al" for the reasons stated below. None of the above studies provides objective, reproducible criteria for case selection needed to evaluate methods of case selection, arrive at meaningful conclusions, or allow cross-study comparisons. These are exactly the concerns of Donohoe and Lantz et al., and the focus of our comments. Furthermore, two of the studies stated by the authors to show that "30% to 40% of newly diagnosed shaken baby cases had medical evidence of previously diagnosed head injury" do not even discuss this issue: Alexander et al describe impact trauma in association with shaken baby syndrome, while Kemp et al relate outcome to initial presentation.
We have ourselves previously used confession and criminal conviction as criteria to support a diagnosis of abuse, including "shaking." However, we would no longer do so. A person may confess for a variety of reasons, and even when properly assessed, a confession must be interpreted cautiously if it is used as "evidence" to prove a medical hypothesis. In the field of child abuse, the carer being told that the only way in which injury was possible was by shaking, or that the charge will be reduced if the carer confesses, may influence a confession. In the family courts a confession may prevent children being sent into care. Unless we have full details of the type of legal proceedings involved, the stage at which confession was made, whether a lighter sentence was passed, or whether it was a plea to get parole¾ all of which are highly relevant¾ the use of a confession to support a scientific argument is unsound. How does one reconcile "confession to shaking" with the fact that more than 50% of the infants in the study by Kemp et al had overt evidence for impact head injury?
If, as the authors state, in "crushing head injuries, as in Lantz et al’s report… child abuse is not a consideration" why was this injury initially considered "highly suspicious" for child abuse on the bais of the ocular findings, and why was the older sibling removed from the home because the eye findings were considered pathognomonic of shaken baby syndrome? The current American Academy of Ophthalmology website on ocular involvement in shaken baby syndrome, as in Lantz et al’s report contains the following statement: "When extensive retinal hemorrhage accompanied by perimacular folds and schisis cavities is found in association with intracranial hemorrhage or other evidence of trauma to the brain in an infant, shaking injury can be diagnosed with confidence regardless of other circumstances" (our italics). If shaken baby syndrome is "old news" why does a child abuse specialist believe that a diagnosis of an accidental injury is valid only if the incident is witnessed by a non-family member, and then create a scenario to fit a formulated belief system of infant head injury, shaken baby syndrome?
The primary objective of our editorial was to bring readers’ attention to one new and one recent study relating to infant head trauma, a specific subcategory of child abuse and to voice concerns about the quality of the scientific information available in the literature. We encouraged the readers to evaluate critically the evidentiary basis for a diagnosis of shaken baby syndrome in the light of the questions raised by the two papers. Of course Donohoe’s study was limited and would only retrieve papers that included the words "shaken baby syndrome" in the title, key words, or abstract. Nevertheless, his search did produce a number of articles that are commonly cited as authoritative in the pre-1999 literature. The important thing is that the lack of scientific rigour that he identified is not restricted to infant head injury papers that specifically mention shaken baby syndrome. If Reece et al perform a critical review of the "number of qualified studies" that they assert would have been included by a wider search they will encounter the same "data gaps, flaws of logic, and inconsistency of case definition" that were present in the literature studied Donohoe. We would urge them to look again, for example, at the paper they cite by Alexander et al, where they will find all the above shortcomings.
Evidence-based medicine (EBM) is a tool, not a panacea, to be sure. However, it is a method that encourages rigorous evaluation of what has been published, and what we believe to be true. Does this mean that anecdote cannot be valued? That small non-controlled studies cannot be one basis for evaluating or treating a patient? That "consensus statements" by "experts" cannot guide us? No. It does, however, mean that we must recognise when a diagnosis and treatment recommendation are a belief system, not a scientific truth. It does mean that we must be ever vigilant to recognise and to attest when "the Emperor is wearing no clothes." The authors implied difficulty in performing valid studies on diagnostic specificity or casual mechanisms does not excuse poorly designed observational studies or conclusions that overstep the data. Association does not equal causation.
Finally, we are at a loss to explain or accept the authors’ statement in their penultimate sentence: "Unfortunately, there remains considerable difficulty for some doctors to accept that children are abused." If the authors are suggesting that we are among those doctors, or are encouraging others to do so, their argument is a willful misinterpretation of what we have written. When there is new evidence that challenges an established conviction, medicine has the responsibility to critically evaluate the data, and if verifiable, reflect that change. We must have no vested interest in yesterday’s belief. We are encouraging doctors to think clearly and critically, even in an area as emotive as child abuse. No more. And no less.
Donohoe M. Evidence-based medicine and shaken baby syndrome. Part I: literature review, 1966-1998. Am J Forensic Med Pathol 2003;24:239-42.
Lantz PE, Sinal SH, Stanton CA, Weaver RG, Jr. Perimacular retinal folds from childhood head trauma. BMJ 2004;328:754-6. (27 March.)
Geddes JF, Vowles GH, Hackshaw AK, Nickols CD, Scott IS, Whitwell HL. Neuropathology of inflicted head injury in children. II. Microscopic brain injury in infants. Brain 2001;124:1299-306
Ewing-Cobbs L, Kramer L, Prasad M, et al. Neuroimaging, physical and developmental findings after inflicted and non-inflicted traumatic brain injury in young children. Pediatrics 1998;102:300-7.
Alexander R, Sato Y, Smith W, Bennett T. Incidence of impact trauma with cranial injuries ascribed to shaking. Am J Dis Child 1990;144:724-6.
Jenny C, Hymel KP, Ritzen A, Reinert SE, Hay TC. Analysis of missed cases of abusive head trauma. JAMA 1999;282:621-6.
Kemp AM, Stoodley N, Cobley C, Coles L, Kemp KW. Apnoea and brain swelling in non-accidental head injury. Arch Dis Child 2003;88:472-6.
Esernio-Jenssen DD. Killer televisions. Electronic response to: Perimacular retinal folds from childhood head trauma. http://bmj.bmjjournals.com/cgi/eletters/328/7442/754#55167 (accessed 20 May).
Lantz PE. Re: Killer televisions. Electronic response to: Perimacular retinal folds from childhood head trauma. http://bmj.bmjjournals.com/cgi/eletters/328/7442/754#55418 (accessed 20 May 2004).