About cookies
We want you to enjoy your visit to our website. That's why we use cookies to enhance your experience. By staying on our website you agree to our use of cookies. Find out more about the cookies we use.
Judicial scalesJustice For Families
Campaign Group
Thursday 11th February 2016
Company Details: Justice for Families (2010) Ltd, 1772 Coventry Road, Birmingham, B26 1PB | Company No: 7303996
Mobile Site |  Desktop Site
The Misdiagnosis of Metaphyseal Fractures: a potent cause of wrongful accusations of child abuse

The Misdiagnosis of Metaphyseal Fractures: a potent cause of wrongful accusations of child abuse

The Misdiagnosis of Metaphyseal Fractures: a possible cause of wrongful accusations of child abuse

James LeFanu




The following scenario illustrates how parents who take their child to casualty seeking medical help can find themselves accused of abusive assault.


  • Parents take child to casualty with head injury or fractured limb for which there may not be an immediately obvious or satisfactory explanation
  • This raises the possibility of NAI hence the request for a skeletal survey.  This shows several metaphyseal ‘fractures’ – though there are no clinical signs of pain or bruising that would support such a diagnosis.  No further steps are taken to confirm the diagnosis by, for example, seeking an orthopaedic opinion or performing subsequent follow-up x-rays to assess the rate of healing as is standard procedure in the management of bony fractures in children
  • Social Services request an expert report from one of a small group of radiologists who confirm the presence of metaphyseal fractures which, in the context of the presenting injury, they attribute to child abuse.  Their reports, however, do not comment on the lack of confirmatory clinical signs, nor suggest the possibility of a differential diagnosis
  • Several months later parents are convicted of grievous bodily harm and their child is taken into foster care


There are several strong clinical grounds for doubting the specificity of these alleged metaphyseal fractures as being characteristic of abuse.  The diagnosis of fracture must be highly improbable in the absence of the relevant clinical signs of injury.  It seems highly improbable that a small baby who has allegedly been the victim of repetitive physical assault should nonetheless appear well with no physical stigma of injury such as bruises or soft tissue injury other than the presenting injury.  Further, as Justice Judge noted in his appeal court ruling in the Angela Canning case, the diagnosis of abuse must be ‘extremely unlikely’ (his words) in the absence of any reasonable explanation for why respectable parents, with no history of mental illness or psychopathology, should seek to inflict these injuries on their children. (1)


Nonetheless paediatric radiologists suggest  these metaphyseal fractures have certain unique attributes that circumvent these challenges to their diagnosis.   Thus Helen Carty and Alan Sprigg in Diagnostic Radiology (Churchill Livingstone 2002) argue:  Metaphyseal fractures are usually asymptomatic and not evident clinically” – despite the ‘direct wrenching or twisting of the limbs’ necessary to cause them.  Further, they claim that metaphyseal fractures ‘do not always result in callus formation’ so there is no reason to perform follow-up x-rays to confirm/refute the diagnosis of a fracture by looking for evidence of healing. (2)  Thus their diagnosis of these metaphyseal fractures as being due to child abuse is, in effect, irrefutable.






I           Background


In 1946 John Caffey suggested the combination of fractures of the long bones in children together with subdural haematoma were probably due to abusive injury. (3)  A decade later he described the two specific radiological findings of the ‘Classic Metaphyseal Lesion’ (CML) – the ‘bucket handle’ and ‘corner fracture’ that he attributed to ‘bits of cartilage and bone being torn off and displayed in a variety of patterns’. (4)


In 1986, Paul Kleinman’s influential paper that defines current understand of metaphyseal fractures rejected Caffey’s interpretation of the aetiology and pathogenesis of the CML.  His careful radiological-histological correlation study in four children with unequivocal evidence of abuse identified the fundamental lesion as ‘a subepiphyseal planar layer of microfractures through the most immature portion of metaphyseal bone’.  The radiological ‘bucket handle’ and ‘corner fracture’ appearances did not represent, as Caffey had believed, large arcs of bone being torn from a zone of provisional calcification but rather represented ‘a radiolucent zone indicating the plane of disruption in the primary spongiosa’.  Further, Kleinman suggested the biomechanics necessary to produce these lesions was not as Caffey had claimed was due to direct physical force, but rather ‘violent shaking if an infant is held by the trunk and extremities’. (5) Kleinman has subsequently confirmed this radiological- histopathological interpretation of the CML in a series of regional studies of the long bones. (6,7,8)



II         Problems arising from Kleinman’s interpretation of the significance of metaphyseal fractures


Kleinman’s authoritative studies provide a sound scientific basis for evaluation of the significance of metaphyseal fractures.  Nonetheless they leave unanswered several outstanding questions.


(i)                  The legitimacy of extrapolation


The question of whether it is legitimate to extrapolate the findings in definitively abused children to infer they have the same significance in the child with no clinical signs of physical abuse is central to the problem of wrongful diagnosis – as shown by analogy with Shaken Baby Syndrome.  Caffey in his general description of SBS, suggested the whiplash effect of vigorous shaking offered ‘a reasonable explanation’ for the presence of subdural and retinal haemorrhages in severely abused children. (9)  The imagery of how the violent to-and-fro movement of a baby’s head could cause bleeding of the vessels of the eye and brain was very persuasive and it seemed logical to infer that any child presenting with these findings must have been shaken – despite the absence of any other circumstantial evidence of abuse. (10,11,12)  This is now acknowledged to be incorrect with the recognition that several mechanisms other than shaking can give rise to such injuries – such as a marked rise in retinal venous pressure from an acute intracranial bleed (Terson’s syndrome), or a hypoxic disruption of the cerebral microcirculation caused by respiratory collapse for any reason. (13,14,15,16) 


Similarly there must be doubts whether Kleinman’s interpretation of the CML in his studies of definitively abused children has the same significance as those ‘picked up’ by skeletal surveys in children in whom there is no evidence of abusive assault.  There are two specific reasons why the extrapolation might be unsafe.


  • The histological-radiological correlation


Kleinman does not make clear how the histological ‘microfractures’ should be visible on the x-ray as ‘a radiolucent zone’, though the prevailing view would be that this represents a failure to deposit calcium secondary to devascularisation.  Nonetheless Dr T K Thomsen of the University of Odense, though agreeing with Kleinman’s general interpretation, observes how ‘none of the cases had macroscopic evidence of fracture, which is surprising in the light of the ‘corner’ fractures seen on radiography’.  Contrariwise microfractures do not necessarily cause ‘a radiolucent zone’ as he notes that radiological appearances may be normal ‘in spite of histologic change’. (17)


  • The ambiguity of the radiological findings


The CML with its ‘bucket handle’ and ‘corner fracture’ appearance are referred to in the literature as being ‘characteristic’ or indeed diagnostic of child abuse – but Kleinman acknowledges that these appearances are present ‘in only a minority of cases’, while in the rest there is simply ‘metaphyseal irregularity’. (18)  The question naturally arises whether he believes these ‘metaphyseal irregularities’ are specific for abuse or whether they might have some alternative explanation in the child with no physical stigma of assault – particularly since the radiological diagnosis of a fracture is not always confirmed at autopsy.


(ii)        Might the x-ray be normal?


There is no case control study comparing the skeletal surveys of injured children with a random group of healthy crontols because of the ethical problem of exposing them to the potential hazards of unnecessary radiation.  This clearly poses a major difficulty when trying to distinguish metaphyseal injury from a normal variant of ossification of growing bones.  It is, however, possible to get some sense of the range of ‘normal variants’ might be from a study also performed by Kleinman of 78 children who died of SIDS where post mortem skeletal survey revealed a pattern of ‘suspicious’ appearances ‘not to be confused with infant abuse’ in over 50%. (19)


Subsequently Alan Oestreich, a radiologist at the Children’s Hospital Medical Centre in Cincinatti, reported that one of Kleinman’s normal variants, the ‘step-off’ was present in ‘a large majority of normal infants’ and emphasises the danger of their radiological misinterpretation: “when radiographed at an angle tilted from the axial plane [the appearance] may be bizarre to such an extent that a true orthogonal view may be required to confirm normality rather than a fracture.” (20)  Similarly Kleinman argues that ‘high detailed radiographs in several projections’ are necessary to distinguish normal variants from fractures, it is thus of extreme concern that this is not current practice in Britain.  Rather the diagnosis of an abusive metaphyseal fracture is often based on a single x-ray view taken as part of a skeletal survey with its (acknowledged) poor level of diagnostic specificity is well recognised.


(iii)               The disputed biomechanics of injury


Biomechanics is central to the diagnosis of metaphyseal fractures for it is clearly necessary what force is required to cause these injuries if the radiologist is to have an opinion whether mild but non-intentional trauma might have caused these lesions (if real) or alternatively, whether the degree of force required is such that one would reasonably expect there to be further circumstantial evidence of physical abuse.  It is thus important to note here that there are no biomechanical studies that would shed light on the forces necessary to produce these injuries while the inference as to how they may have occurred is clearly contradictory: Kleinman attributes them to violent shaking (5) while Carty and Sprigg insist they are due to ‘twisting and wrenching of the limbs’. (2)



In summary, Kleinman’s studies are an important contribution to the interpretation of metaphyseal findings but their limitations should also be recognised.  Of particular concern should be:


  • The validity of extrapolating the significance of radiological findings in abused children to those where there are no signs of neglect or abusive assault
  • Doubts as to whether (or how) the subtle histological lesions of microfractures should be visible as a ‘radiolucent zone’ on x-ray
  • The lack of case control studies that would make the distinction between the ‘normal’ and the ‘abnormal’
  • The failure to perform the necessary radiological investigations that would permit that distinction to be made
  • The lack of biomechanical studies that would support one or other of the (contradictory) inferences as to the nature of the injury necessary to cause these metaphyseal fractures.


There seems little doubt, in light of the above, that the radiological interpretation of the significance of metaphyseal fractures being due to NAI might be in error.  This necessarily raises the question of what else might account for the radiological appearances that would appear to fall into two quite distinct patterns which in turn correlate the parental account of events leading up to their child’s presentation to hospital.  This raises the possibility of the following differential diagnosis:  normal variants, birth injury, bone disease (cause unknown).



III        The differential diagnosis of metaphyseal fractures


(i)                  Normal Variants


The first type of clinical presentation is as in the scenario outlined in the Introduction where the child presents with an injury for which there may not be an immediately satisfactory explanation – such as a chronic subdural or a long bone fracture.  The skeletal survey reveals numerous symmetrical bilateral fractures or irregularities on both surfaces medially and laterally of the long bones.  The reason why these findings might be normal variants have already been outlined: the lack of clinical signs of abusive assault, the lack of case control studies that would permit the distinction between the normal and abnormal, and the inadequate radiological investigation that might permit such a distinction to be made.  There are two further reasons why this pattern of findings might be normal variants.  First, the normal variants in Kleinman’s study ‘not to be confused with abuse’ were, as here, symmetrical and bilateral.  Second, the claim that these symmetrical bilateral lesions are indeed due to abusive assault presupposes that many apparently respectable parents with no history of psychopathology should all systematically ‘twist or wrench’ the limbs of their babies in precisely the same way so as to produce the same patterns of injury – but which consistently leave no external signs of injury.  This must be, in Justice Judge’s words ‘extremely unlikely’.


(ii)                Birth trauma


The second possible differential diagnosis is birth trauma.  Here the child can present as in (i) but a skeletal survey shows in addition evidence of healing of old fractures that might be supportive of the diagnosis of NAI.  Here again there is no control study that would reveal the incidence and type of metaphyseal and other fractures that might be sustained during birth.  There is, however, no doubt that birth trauma can produce a wide spectrum of injuries including subdural haemorrhages (present in 10% of 111 infants who had an MRI scan within 48 hours of delivery) and fractures of the rib, clavicle and long bones. (21,22)  There is no difficulty in recognising how readily a fracture sustained at birth but not identified on a skeletal survey till two or three weeks later could be wrongly attributed to child abuse.  Thus, Barry and Hocking in their discussion of a birth induced rib fracture comment, “had the child not had a radiograph taken on the first day of life there would have been doubts as to the cause of the injury, and the incorrect diagnosis of child abuse considered”. (23)


(iii)               Bone disease (cause unknown)


Here both the clinical presentation and the pattern of radiological abnormality is quite different from (i) and it is certainly very significant that this important distinction is not mentioned in the literature.  The presentation is usually with a ‘spontaneous’ fracture or following minimal trauma but the skeletal survey reveals not only multiple fractures of the ribs and long bones but in addition gross metaphyseal abnormalities at the end of the long bones.  There are no stigma of the (very severe) assault that would be necessary to cause such symptoms – so it is reasonable to suppose there must be some form of underlying bone disease, of which osteogenesis imperfecta is clearly the best recognised.


When, however, the relevant tests for OI prove negative, the parents are likely to be accused of causing the injuries despite the lack of circumstantial evidence of assault.  The possibility remains however that the child has some form of bone disease whose cause has not yet been identified.  There is, for example, substantial evidence that minor non-abusive trauma can result in fractures noticeably in the premature infant (osteopenia of infancy) or even in apparently ‘normal’ babies or children receiving chest physiotherapy or undergoing orthopaedic manipulation. (24,25)


This supposition is of course speculative and indeed one of the protagonists of such a view, Colin Paterson, has recently been disciplined by the General Medical Council.  Nonetheless the substantial question remains of how an infant might sustain severe multiple bone fractures in the absence of other circumstantial evidence of abusive assault – and it would be wrong to exclude on prima facie grounds the possibility that there may be conditions that predispose to fractures in infants that are not yet recognised.  This is illustrated by the recent finding of a mutation in the LRP5 gene that influences the effects of mechanical stimulation on bone formation that is currently under investigation by Professor Nick Bishop of Sheffield’s Children’s Hospital. (26)





The misinterpretation of metaphyseal fractures identified on skeletal surveys as being characteristic of child abuse is clearly a possible source of false allegations against parents.  Specifically the reliance on a radiological diagnosis in the absence of other circumstantial evidence of abuse fails to acknowledge three possible other explanations for the x-ray findings – namely normal variants, birth trauma and bone disease (cause unknown) each of which has a distinct clinical presentation and radiological appearance.





1.         R v Canning 2004. EWCA Crim 1, 19 Jan 2004


2.         Helen Carty, Alan Sprigg.  The radiology of non-accidental injury, in Grainger & Allison’s Diagnostic Radiology, ed. R G Grainger, D Allison.  Churchill Livingston 2002


3.         Caffey J.  Multiple fractures in the long bones of infants suffering from chronic subdural haematoma.  AJR 1946.  Ref 56, pp163-173


4.         Caffey J.  Some traumatic lesions in growing bones other than fractures and dislocations: Clinical and radiological features.  BRJ Radiol 1957; 30: 225-238


5.         Kleinman P, Marks S C and Blackbourn B.  The metaphyseal lesion in abused infants: A radiologic-histopathologic study.  AJR 1986; 146: 895-905


6.         Kleinman P K, Marks SC Jr.  A regional approach to the classic metaphyseal lesion in abused infants: A proximal tibia.  AJR 1996; 166: 421-6


7.         Kleinman P K, Marks S C Jr.  A regional approach to the classic metaphyseal lesions in abused infants: The distal fibia. AJR 1996; 166: 1207-1212


8.         Kleinman P K, Marks S C Jr.  A regional approach to the classic metaphyseal lesions in abused infants: The distal femur. AJR 1996; 170: 43-47


9.         Caffey J. The whiplash shaken infant syndrome.  Paediatrics 1974; 54: 396-403


10.       Carter J E, McCormick A Q.  Whiplash shaking syndrome: Retinal haemorrhages and computerised axial tomography of the brain.  Child Abuse Negl 1983; 7: 279-286


11.       Duhame A C, Christian C W, Roke L B, Zimmerman R A.  Non accidental injury in infants: The Shaken Baby Syndrome. N Engl J Med 1998; 38: 182-9


12.       Wilkins B.  Head injury – abuse or accident.  Arch Dis Childhood 1997; 76: 393-7


13.       Geddes J F, Plunkett J.  The evidence base for Shaken Baby Syndrome.  BMJ 2004; 328: 719-720


14.              Christian C, Taylor A, Hurtle R et al.  Retinal haemorrhages caused by accidental household trauma.  J Pediatr 1999; 135: 125-7


15.              Geddes J, Tasker R, Hackshaw A et al.  Dual haemorrhage in non-traumatic infant death.  Does it explain the bleeding in ‘Shaken Baby Syndrome’?  Neuropathol Appl Neurobiol 2003; 29: 14-22


16.              Adams G, Ainsworth J, Butler L et al.  Update from the ophthalmology child abuse working party: Royal College of Ophthalmologists.  Eye 2004; 18: 795-8


17.              Thomsen T K, Elle B, Thomsen J L.  Post-mortem radiological examination in infants: Evidence of child abuse?  Forensic Science International 1997; 90: 223-230


18.              Kleinman P K.  Diagnostic imaging of child abuse (second edition) Mosby


19.       Kleinman P K, Belanger, Karellas A, Spevak M R.  Normal metaphyseal radiologic variants not be confused with infant abuse.  AJR 1991; 156: 781-783


20.       Oestreich A E, Borhaan S A.  The periphysis and its effects of the metaphysic.  Skeletal Radiol 1992; 21: 283-286


21.       Whitby E H.  Frequency and natural history of subdural haemorrhages in babies and relation to obstetric factors.  Lancet 2003; 362: 846-851


22.       Cumming W A.  Neonatal skeletal fractures.  Birth trauma or child abuse?  J Can Ass Radiol 1979; 30: 32-33


23.       Barry P W, Hocking.  Infant rib fracture – birth trauma or non-accidental injury.  Archives of disease in childhood 1993; 68: 250


24.       Chalumeau M et al.  Rib fractures after chest physiotherapy for bronchiolitis.  Paediatric Radiol 2003; 33: 733-4


25.       Grayev A M, Boal D K B, Wallach D M et al.  Metaphyseal fractures mimicking abuse during treatment for club foot.  Pediatr Radiol 2001; 31: 559-563


26.       Hartikka H, Makitie O, Mannikko M et al.  Heterozygous mutations in the LRP5 gene are associated with primary osteoporosis in children.  Journal of Bone and Mineral Research 2005; 20: 783-789.