Fractures: systematic review

Child Protection Evidence is a resource for clinicians across the UK and internationally to inform clinical practice, child protection procedures and professional and expert opinion in the legal system.


Fractures have been recorded in as many as 55% of young children who have been physically abused,1,2,3 18% of whom have multiple fractures.1  Fractures in physical child abuse denote severe assault and young children are at greatest risk of abusive fractures. Many abusive injuries can be clinically occult and investigative strategies are designed to maximise their early detection, potentially preventing escalation of physical abuse to more serious injury.

This systematic review has been updated and now evaluates the scientific literature published up until February 2017 that compares abusive and non-abusive fractures in children. There are increasing numbers of high quality studies being published in this field, which continue to address the value of skeletal surveys in different populations and fracture patterns associated with abuse. In the 2017 systematic review update, seven new studies have been included.4,5,6,7,8,9,10 These studies compared the yield of skull fractures between 3D computerised tomography (CT) scan and standard skull radiographs,5 provided data on follow up on skeletal survey,7 the diagnostic value of bone scintigraphy4 and added further data regarding metaphyseal fractures9 and forearm fractures.8 Two new studies on fracture dating included very young children,  consequently addressing a previously unevaluated area.6,10

The review aims to answer four clinical questions:

    • Which fractures are indicative of abuse?
    • What is the evidence for radiological dating of fractures in children?
    • What radiological investigations should be performed to identify fractures in suspected child abuse?
    • Does cardiopulmonary resuscitation cause rib fractures in children?

Evidence summary

    • Abusive fractures are more common in children less than 18 months of age than in those older than 18 months.
    • Multiple fractures are more suspicious of abuse than non-abuse.
    • Rib fractures in the absence of major trauma, birth injury or underlying bone disease have a high predictive value for abuse.
    • Multiple rib fractures are more commonly abusive than non-abusive.
    • Abusive femoral fractures are more likely to arise in children who are not yet walking.
    • Mid-shaft fractures are the most common femoral fractures in abuse and non-abuse (analysed for all age groups).
    • Supra condylar humeral fractures in children are associated with accidental injury whilst the commonest abusive humeral fractures in children aged less than five years are spiral or oblique.
    • Humeral fractures in those aged less than 18 months have a stronger association with abuse than humeral fractures in older children.
    • Linear fractures are the commonest abusive and non-abusive skull fractures.
    • Metaphyseal fractures are more commonly described in physical child abuse than in non-abuse.
    • Metaphyseal fractures have been frequently described in fatal abuse.
    • Most children with classic metaphyseal lesions (CML) have other associated injuries which are often multiple.
    • Pelvic, hand, feet and sternal fractures occur in physical abuse and appropriate radiology is required for their detection.
    • The dating of fractures is an inexact science, the radiological features of bone healing represent a continuum, with considerable overlap in timescale.
    • The accuracy of radiological estimates of the time of injury are in terms of weeks rather than days.
    • Radiological investigations of suspected physical abuse include initial and follow up skeletal surveys with specific views to maximise detection of occult injuries particularly in young children.
    • Either skeletal survey (SS) or radionuclide imaging (RNI) alone will miss occult fractures.
    • Studies suggest that up to 12% of contacts under 2 years of age, of children who have been abused with serious injuries may have a positive skeletal survey with twins being a particularly high risk.

This systematic review evaluates the scientific literature that compares abusive and non-abusive fractures in children published until February 2017 and reflects the findings of eligible studies. The review aims to answer four clinical questions:

    • Which fractures are indicative of abuse?
    • What is the evidence for radiological dating of fractures in children?
    • What radiological investigations should be performed to identify fractures in suspected child abuse?
    • Does cardiopulmonary resuscitation cause rib fractures in children?

A literature search was performed using all OVID Medline databases for all original articles and conference abstracts published since 1950. Supplementary search techniques were used to identify further relevant references. See Appendix 1 for full methodology, including search strategy and inclusion criteria.

Potentially relevant studies underwent full text screening and critical appraisal. To ensure consistency, ranking was used to indicate the level of confidence that abuse had taken place as well as for study types.


Fractures found in younger children are often due to abuse, and the included studies state that in children less than one year 25-56% of all fractures were abusive.11,12,13,14

Abusive fractures are more prevalent in the youngest age groups; 85% of accidental fractures occurred in children older than five years whilst 80% of abusive fractures occurred in children younger than 18 months of age.15 In children less than 4 years old, the mean age for accidental fractures was 22.1 months (95%CI 21.2, 24.02 months) while the mean age of abuse cases was 11.7 months (95%CI 10.6, 12.7 months), thus a significant difference between the age of accidental injuries and abuse was observed (p<0.001).16 The greatest risk of abusive fractures was reported by Skellern et al. as being children less than four months of age (p=0.0007).14


Five studies assessed the difference between gender.12,13,14,16,17 Abusive fractures were reported as more common in boys in one study (p=0.024),17; however, no gender difference between abusive and non-abusive fractures was noted in children aged less than one year12,13,14 or in children aged less than four years (p=0.065).16

Influence of ethnicity and socio-economic group

Two studies analysed ethnicity.17,18 Leventhal et al.17 found no difference by ethnicity or socio-economic grouping; however, in children less than three-years-old with fractures there was a greater proportion of abusive fractures in black children than white (p≤0.01).18

Multiple fractures

Abused children were more likely to have multiple fractures than non-abused children (p<0.001),15,17; however, one study found no difference.12

1.1. Rib fractures

In children less than four years of age, rib fractures were more commonly found in those abused than those accidentally injured.16,19

Children with rib fractures were significantly more likely to have been abused, and in those less than 18 months, with rib fractures the odds ratio (OR) for abuse was 23.7 (95%CI 9.5, 59.2) (p<0.001) while in those aged over 18 months was 9.1 (95%CI 3.3, 25) (p<0.001).16

No study addressed disabled children or the influence of ethnicity and socio-economic group.


A meta-analysis was undertaken for children less than 48-months-old.16,20,21,22 The included four studies include children with confirmed or suspected abuse, but showed low heterogeneity (I2=0%). The positive predictive value (PPV) for rib fractures in relation to suspected or confirmed abuse is 66% (95% CI 42.5-89.7).

Number and location of abusive rib fractures

A study investigating fractures in young children found that rib fractures were the most strongly associated with abuse,23 whilst another study found the most prevalent injuries in the child abuse group following non-bony head injury and skull fractures were rib fractures.17

Multiple rib fractures

Abused children had multiple rib fractures.15,19,20,21,24,25,26 One study included a comparison with children with metabolic bone disease.25 Flail chest due to multiple rib fractures in infant physical abuse is reported in one study27 and another study reported that the risk of mortality increases with the number of ribs fractured.23

Location of fractures

Abusive rib fractures were recorded at any location on the rib and were either unilateral or bilateral.15,20,21,24,28,29 Two studies found that anterior fractures were more common in abuse whilst lateral fractures were more common in non-abused children.20,21 Posterior rib fractures were assessed in seven studies and stated that they were the predominant abusive fracture.20,22,26,30,31,32,33 Posterior and postero-lateral rib fractures were found to be equally common in abuse and metabolic bone disease.25

First rib fractures

Two studies were included that reported on first rib fractures.22,34 Fractures were predominantly lateral, one posterior.22,34 Five cases of abusive first rib fractures were reported in one study, four of which had no fracture to adjacent bones and four had associated neurological injury.22

1.2. Costochondral junction fractures

Anterior costochondral fractures are described in abuse,28,34 they may be difficult to visualise radiographically and can occur with associated abdominal injuries.28

Intrathoracic injury in children with rib fractures

One study evaluated intrathoracic injuries in abused and non-abused children less than three years of age with rib fractures. Accidentally injured children had more intrathoracic injuries than abused.19

1.3. Femoral fractures


Two meta-analyses were conducted on children with femoral fractures aged up to 18 months with confirmed or suspected abuse13,35,36,37,38,39,40,41,42 and children aged 12-48 months with confirmed or suspected abuse,35,36,38,40,41,42,43 compared to non-abused children. For children aged up to 18 months there was low heterogeneity between studies (I2=0%) and the positive predictive value (PPV) for suspected or confirmed abuse when a femoral fracture is present is 51.1% (95% CI 34.1-66.1). In children aged 12-48 months there was moderate heterogeneity present between studies (I2=57.4%) and a far lower PPV of 11.7% (95% CI 6.1-17.3) for suspected or confirmed abuse. These findings are consistent with the observation that children who are not independently mobile are far less likely to sustain an accidental femoral fracture.40


Femoral fractures in the abuse group occurred predominantly in children less than one year of age.13,14,15,17,18,35,38,39,41,42,43,44,45,46,47 Abused children who were less than 18 months of age with a fracture were more likely to sustain a femoral fracture (17.5%, 66/377) than those aged 18 months or over (5.7%, 7/123). In contrast, accidentally injured children with a fracture were more likely to sustain a femoral fracture aged 18 months or over than below this age (p<0.001).16 The odds ratio (OR) for abuse in a child aged less than 18 months old with a femoral fracture was 1.8 (95%CI 1.2, 2.7) (p=0.005), compared to 0.3 (95%CI 0.1, 0.7) (p=0.003) in a child aged over 18 months.16

One study showed that a third of isolated femoral fractures in children less than three years of age were abusive.45 The same study showed that a spiral fracture was the most common abusive fracture in children younger than 15 months  (p=0.05); over this age spiral fracture was not significantly associated with abuse.45  However, ther is some conflict as a different study reported   no association between spiral fracture and abuse was confirmed with a population of children less than the age of three years.48

Another study looked at a wider age range (children aged up to three years) where there was no specific fracture type associated with abuse and found there was a highly significant association between femoral fracture in non-ambulant children and abuse.39 A study of femoral fractures in children up to 48 months of age, found that the median age of abuse cases was lower (four months) versus accidental (26.2 months) (p<0.001).36

Fracture type

The commonest fracture location in both abused and non-abused children was mid-shaft of the femur.13,15,41,43,44,45,47,48,49,50 Proximal physeal injuries were described in abused children.44,49,51,52

Of 18 impacted transverse fractures of the distal femoral metadiaphysis in infants, 13 (72%) were non-abusive (from short falls) and five (28%) abused cases. The mean age did not differ significantly between the two groups (abuse: eight months vs non-abuse: 12 months). This was a highly selected case series ascertained through child protection specialists.53

A study of diaphyseal fractures found they were more common in accidental than abusive injury (p=0.007), however distal femoral fractures were more common in abused than accidentally injured children (p=0.01).36

Femoral shaft fractures

Transverse femoral fractures were more commonly associated with abuse than accidental injury.48

Femoral Metaphyseal fractures were recorded in a greater proportion of abusive femoral fractures than non-abusive fractures.15,34,35,44,50

One study reported 10/20 complete distal femoral metaphyseal fractures occurring in children aged less than one year were due to abuse54 and accidental causes of metaphyseal fractures included birth injury, motor vehicle collision and falls.54

1.4. Humeral fractures


Two separate meta-analyses have been conducted, one for children up to 18 months of age13,41,55,56,57 and one for 18-48 months.41,55,56,57

There was low heterogeneity between the studies of children aged 0-18 months (I2=0%). For children aged up to 18 months the positive predictive value (PPV) of a humeral fracture for confirmed or suspected abuse is 43.8% (95% CI 27.6-59.9). For children aged 18-48 months, the heterogeneity between studies was low (I2=28.8%) and the PPV of a humeral fracture due to suspected or confirmed abuse was only 1.8% (95% CI 0-3.9).

The largest case-control study of humeral fractures in less than 48-month-olds highlighted that age less than 18 months, in conjunction with physical or radiology evidence of prior injury and suspicious history are significant indicators of an abusive aetiology.55

Age of children with humeral fractures

Several studies reported on the age of children with humeral fractures, with abusive humeral fractures occurring in 74/111 children under the age of one-year.46

Children less than 15-months-old, with humeral fractures, were more likely to have sustained an abusive humeral fracture (9/25, 36%) than those aged 15-36 months (1/99, 1%) p<0.05.57 Humeral fractures in those aged less than 18 months were significantly more likely to be due to abuse than accidental injury (p<0.001); the odds ratio (OR) for abuse was 2.3 (95% CI 1.3, 4.1) p=0.00416 For a child older than 18 months with a humeral fracture the OR for abuse was 0.29 (95%CI 0.1, 0.7) p=0.005.16 In a study of children under four-years with humeral fracture, a multivariate analysis was used to develop an algorithm to identify abuse55. Abused children tended to be younger than those who were accidentally injured (p>0.001).55

Fracture type

Four studies confirmed that spiral or oblique humeral fractures in children less than five years of age were strongly associated with abuse.15,17,56,57 Humeral shaft fractures were more frequently due to abuse55 and supracondylar fractures were overwhelmingly due to accidents.15,55,57,58

Fracture dislocation of the proximal or distal humeral epiphysis was described in abused and accidentally injured children aged up to seven years (majority under three years).59,60,61 Epiphyseal humeral fractures are difficult to visualise but this may be enhanced using ultrasound or magnetic resonance imaging’scanning.59,60,61

1.5. Skull fractures


A meta-analysis was conducted on studies that included children aged 0-48 months.12,16,18,62,63

There was low heterogeneity between these studies (I2=0%) and the positive predictive value of a skull fracture for suspected or confirmed abuse was 20.1% (95% CI 13.3-26.9). Contrary to the other fractures analysed, age less than 18 months did not appear to be a key variable.16 In one study, in children aged less than 18 months skull fractures were more common in the control subjects than those who were abused.16

Age of children with skull fractures

All studies included pre-school children, five only included children less than one year of age.12,13,14,64,65 Skull fractures of either aetiology (abuse or accident) have an increased prevalence in younger infants.12,13,16,18,62,64,65 In a study of children less than three years of age 80% (75/94) of accidental and 88% (23/26) of abusive skull fractures were in infants under one year of age.17

Skull fractures were more frequent in accidental trauma than abuse in those aged less than 18 months, (p=0.002) and for those older than 18 months (p<0.001).16

Fracture type

The commonest abusive and non-abusive fractures were linear.15,17,62,63 Meservy et al. studied 134 children aged less than two years old (motor vehicle accident excluded), 39 of whom were abused.62 Multiple or bilateral fractures, or those that crossed suture lines, were more common in abused children (p<0.05).62 No significant differences were recorded between the two aetiology groups with relation to nonparietal, depressed, diastatic ≥3mm or complex fractures.62 In a series of alleged short distance fall victims, four children aged between two days and 18 months were found to have bilateral linear skull fractures.66

1.6. Metaphyseal fractures


A study of 215 children aged less than three years old who were examined for a fracture described 13 abusive humeral fractures, including metaphyseal fractures.17

A study of 34 infants less than one year old with 55 fractures; 19 long bone fractures and five metaphyseal fractures were recorded.12 Cause was accidental in 15 infants (44%) and nonaccidental in 19 infants (56%), representing a greater incidence of accidental injury than previously indicated in this age group. However, there was no significant difference between the distribution of fractures between abuse and non-abuse.

The likelihood of classic metaphyseal lesions (CML) fracture in head injured infants aged less than one year of age was assessed in a study of 60 infants. There were 42 children deemed ‘low risk’ of abuse and 18 ‘high risk’ (abuse rank two); 9/18 children in the high risk but no child with a ‘low risk’ had CMLs.67 The CMLs involved all long bones but, more frequently, femur and tibia, 55% of CML were found around the knee.

A study of 63 femoral fracture episodes in children aged less than four-years-old (once pathological fractures and MVC excluded) found 24 fractures were due to abuse/suspected abuse, four of which were distal metaphyseal chip fractures and a further five were distal metaphyseal fractures.44 Of the 39 non-abusive fractures, five were distal metaphyseal fractures, one of which was a chip fracture.44

A study of 826 children with accidental fractures and 35 children less than five-years-old with abusive fractures reported that 17 metaphyseal corner fractures were noted amongst the abused children, all aged less than 18-months-old and that there were no metaphyseal fractures in the accident group.15

A study of infants aged less than one year showed the mean age of those with metaphyseal fractures was four months, 15/50 had bilateral symmetrical lower extremity metaphyseal fractures. There were 42/48 children with CML had positive skeletal survey for occult fractures and one nine-day old infant had CML as a consequence of birth.34

Femoral metaphyseal fractures are far more common amongst abused than non-abused infants.

A study that included 14 children with humeral fractures reported that 11 were abuse cases amongst which there were metaphyseal fractures  but in none of the three accidental falls (all were supracondylar).41 Infants aged less than one year with CML were described and of the 119 children assessed for suspected maltreatment, 111 (93%) were rated ‘highly likely’ to be abused.  The study found that the majority occurred in infants aged upto 2 months and decreased in frequency as age increased.9 More than 95% of subjects with CMLs had at least one other injury.

Fatally abused children with metaphyseal fractures

Five studies by Kleinman et al.68,69,70,71,72 delineated the histologic/radiologic correlates of classical metaphyseal fractures in 31 fatally abused infants with skeletal injuries. Children ranged in age between three weeks and 10.5 months (mean age: three months). There were 165 fractures in the 31 children; 72 long bone fractures; 64 (89%) of which were classical metaphyseal lesions in 20 children.  Metaphyseal fractures accounted for 39% of the total fractures and were found in 20/31 children. The ’commonest site’ for CMLs was the tibia whilst metaphyseal fractures were commonly bilateral and symmetrical. Specimen radiography increased the yield of fractures noted on skeletal survey from 58% to 92%.68,69,70,71,72

Tibial and fibular fractures

Tibial and fibular fractures were attributed to abuse in 1/8 fractures in children less than three years,18 when assessing abusive tibial fractures alone, 7/12 were metaphyseal.15

In assessing the age of children with fractures, one study noted 96% (23/24) of all tibial or fibular fractures were abusive in children less than 18 months.37 Whilst another found that in children less than three-years-old, 14/35 tibial and fibular fractures were from abuse.17 Worlock reported abusive tibial and fibular fractures in children less than 18 months of age but not in older toddlers.15 Tibia and fibular fractures were more common in abuse than accidental injury (p< 0.001) in children under four years old .16 The OR for abuse in a child older than 18 months was 2.1 (95%CI 0.7, 6.2) p=0.172.16 The odds ratio (OR) for abuse in a child aged less than 18 months with a tibial/fibular fracture was 12.8 (95% CI 5.1, 32.6) p<0.001.16

Radial and ulnar fractures

Children with injuries attributed to abuse were significantly younger than those with non-abusive injuries.8 Most non-abusive fractures were greenstick, Worlock identified metaphyseal ulnar fractures in the abused group only.15,16,17,18 Radial/ulnar fractures were more common in abused than control group children aged less than 18 months (p=0.001),16 whilst in those less than four years of age with a radius/ulnar fracture, the OR for abuse was 5.8 (95%CI 2.4, 14.3 (p<0.001)).16 Forearm fractures in 135 children were studied and there were 11 cases attributed to abuse. These children were significantly younger than those with non-abusive injuries (p<0.001).8 Buckle fractures were the most common fracture type in both groups followed by transverse fractures. There were no significant differences in the distribution between the two groups.  The study reported young age, additional injuries and an absent or inconsistent explanation should increase concern that the fracture was caused by child abuse.

*The criteria for the ranking of abuse was lowered for this section that relies largely on case studies of rarer fracture types, some of which have a lower abuse ranking.

1.7. Pelvic fractures

Nine included studies described pelvic fractures.46,73,74,75,76,77,78,79,80 Most children with pelvic fractures had suffered from multiple additional injuries.46,73,74,75,76,77,78,79 Two studies reported three infants. One child had up to 29 additional fractures recorded,74,75,76,77,78,79 whilst another presented two case reports: one child had multiple burns and pelvic fractures; one was fatally abused due to associated intra-abdominal injuries.78

Five studies included children with pelvic fractures with associated suspected or confirmed sexual abuse. 73,75,76,78,80  In one study, a child was disabled and non-verbal.75

A report of fourteen cases of pubic radio-lucency are described, seven were cases of confirmed abuse, three had fractures of the superior pubic ramus, three had pelvic normal variants and one had indeterminant findings (all seven had multiple associated fractures including metaphyseal).77 A study described a four-month-old infant with leg pain and swelling had spinal fractures, comminuted fracture of distal left femoral metaphysis and a right ischial tuberosity fracture. These were not evident on the initial plain films but were seen on magnetic resonance imaging (MRI) and on the two-week follow-up skeletal survey.74

1.8. Fractures of the hands and feet

Seven studies described abusive fractures to hands and feet.16,34,80,81,82,83,84 One comparative study found no significant difference in the rate of fractures to the hands and feet between abused and control children less than four years of age.16

Two cross-sectional studies showed that fractures to the hands and feet were present in 1.4% of  all children less than ten years of age being evaluated for child abuse,82 and in 5-5.5% of children less than two years old undergoing skeletal survey.34,82 Abusive fractures to hands and feet were recorded in infants and toddlers; 47/56 children with abusive fractures to the hands and feet were aged less than two years.34,82 The mean age of the children with abusive fractures to the hands was 14.1 months(range 5.6-22.4 months). The mean age of children with abusive fractures to the feet was 10 months (range 1.3-13.6 months).82 One study showed the mean age of abusive fractures to the hands or feet was 5 months (range 1-10 months).34 The mean number of hand fractures per child was 2 (range 1-4), the mean number of fractures to the feet was 1.5 (range 1-3). The most common fractures were metacarpal or metatarsal.82

Another study described 22 fractures of the hands and feet are in 11 abused infants,83 Torus fractures predominated and 7/11 infants had additional fractures of the ipsilateral extremity.

A study of 16 infants with abusive fractures to hands or feet, metatarsels were the most common site of injury and transverse or buckle type fractures were present in 75%.34 All infants with hand or feet fractures had additional fractures. There were two case reports included which described a six-month-old with fractures of second to fifth metatarsals bilaterally, and associated fractures of radius and ulna81 and a ten-year-old with multiple bilateral phalangeal fractures of different ages thought to be secondary to hyper-extension.84

1.9. Mandibular fractures

Two included studies describe mandibular fractures.85,86 One study describes a mandibular fracture after a direct blow in a six-month-old.85 A second study addressed abusive mandibular fractures occurred in all age groups86 and found that in contrast to other abusive fractures, mandibular fractures were not associated with other injuries.86

1.10. Sternal fractures

One included study87 of 12 children (age range: 1.5 – 15.5 years) with sternal fracture, two were abused.87 Both abusive cases were less than three years of age, one had associated long bone fractures.

1.11. Clavicular fractures

Three included studies described clavicular fractures in children aged up to 5 years.15,16,34 In children with fractures, clavicular fractures were more common in abused children than those who were accidentally injured (p<0.001) for children less than 18 months and for those 18 – 48 months.16 For a child aged less than four years with a clavicular fracture the odds ratio (OR) for abuse is 4.4 (95% CI 1.9, 10.2) (p=0.001).16 Clavicular fractures were identified in 18% of abused children less than 18 months versus 5% of those accidentally injured.15 For those aged 18 – 60 months, 14% of the abused children versus 12% of the accidentally injured children had clavicular fractures.15 One study noted that 4/24 abused infants with clavicular fractures had bilateral clavicular fractures 10/ 24 infants had further occult fractures on skeletal survey.34

1.12. Vertebral fractures

Fifteen included studies contain findings on vertebral fractures.34,88,89,90,91,92,93,94,95,96,97,98,99,100,101 We conducted a spinal injuries systematic review since injury to the spine can consist of both spinal fractures and injury to the spinal cord. Specific details relating to the studies of vertebral fractures can be found in the Spinal Injuries section.

1.13. Limitations of review findings

    • Considerable heterogeneity between studies.
    • Small number of comparative studies, in particular relating to less common fractures such as sternum, mandible, scapular, feet and hands.
    • Inadequate analysis of data by child’s developmental stage, and a relative lack of data relating to non-mobile infants.
    • Few studies gave details of the number and type of co-existent fractures in individual children, which would have been of value to clinicians.
    • A lack of radiological detail is given in some of the studies particularly relating tometaphyseal fractures.
    • No data relating to disabled children.

Seven studies were included of children aged upto 17-years-old.102,103,104,105,106,107,108  There were no studies addressing fracture dating in disabled children or the influence of ethnicity or socio-economic group.

Two studies found no correlation between the age and gender of the child and fracture dating102,104 and another two studies ascertained children who had been abused where the authors felt they could identify the timing of the injury.103,106 Halliday et al. concluded that if there is no subperiosteal new bone formation (SPNBF) on x-ray, then the fracture is likely to be less than 11 days old,103 whilst Sanchez et al., studying 16 infants, evaluated callus formation as an indicator of healing.106

Four studies defined different radiological criteria for fracture dating. The studies considered fractures of the immobilised forearm,104 femoral region,107 long bones105 and clavicular fractures in the newborn.107

Three studies found hard callus and early remodelling is seen at eight weeks in most cases.104,108 The peak period for a hard callus is at three weeks or greater and remodelling at five weeks or greater.105 A study exploring the rate of radiological healing in newborn infants evaluated 131 infants with clavicular fractures aged 0-93 days and developed a timetable of healing according to standardised criteria.107

Cumming’s study estimated the earliest calcification at the fracture site in 23 newborns, calcified periosteal reaction was noted as early as seven days.102

Levels of agreement between three radiologists regarding the timing of radiological features was only deemed ‘moderate’, apart from the recognition of periosteal reaction103. In another, levels of agreement between three radiologists were high amongst all radiographs, however the presence of a plaster cast limited interpretation for some images.105

Table 1. Summary of the dating characteristics identified6,10,102,103,104,105,106,107,108


* drawn from a population of children who were abused, with assumed date of injury
With thanks to Dr Kath Halliday for providing original data from her study103

2.1. Research implications

    • Future studies could include fractures that are not routinely immobilised.
    • Future research should use previously defined features of radiological dating to enable comparisons between study populations.
    • Further studies establishing a timetable for healing for fractures in children less than three years of age would be of value.

2.2. Limitations of review findings

    • Comparison between studies was hampered as:
      • Different bones were studied
      • There were variable time intervals between radiographs
      • Differing numbers of radiographs per fracture.
    • In many studies, fractures were immobilised with plaster, this may limit visualisation of features.
    • Although one study has included children whose fractures were not immobilised, these children had been abused and thus the precise time of the injury may be questionable.
    • A study has now been conducted determining the rate of healing for infant fractures this was necessarily based on clavicular fractures which may not heal at the same rate as long bones etc.
    • No metaphyseal fractures were included in these studies and data cannot therefore be generalised to the healing of CMLs.

3.1. Skeletal surveys


Nine studies including children aged up to 10 years were included.1,109,110,111,112,113,114,115,116

Table 2: Details of included studies that evaluated follow up skeletal survey


Author, year, study design

# children undergoing repeat imaging/total children

Imaging performed

Time interval

Results of follow up SS




Harper et al., 2013112

Cross sectional

 796/1038 (76.7%)

20 centres included full SS according to AAP guidance 2009. At follow-up all centres excluded skull, 6 centres excluded spine, five excluded pelvis films

Not available

124/796 (15.6%) had a new fracture, fractures included rib, long bone, CML, hands or feet, clavicle, vertebra and scapula. 18/252 (7.1%) of children with a normal initial SS had fractures on follow-up. 6.5% of subjects had fractures of hands and feet on follow-up and 1.6% vertebral fractures. No new fractures of the pelvis were identified. Concerning features on the initial SS were confirmed as normal in 55 subjects Prospective study across 20 centres. Indications for SS not given. 24% of subjects did not return for follow-up imaging


Singh et al., 2012114

Retrospective cohort


Full SS including oblique views initially, omitted skull and spine for follow up imaging

Mean 19 +/- 11 days

24/169 (14%) had previously unrecognised healing fractures on follow up

6/24 (25%) of these subjects had a negative initial SS. In 8 cases findings on follow up influenced abuse diagnosis

Retrospective review from 2002-2009, 88% < 1 year. Significant increase in number of follow up SS 2005-2009. Except for two fractures that were present, but missed, on the initial SS, all of the fractures identified on the follow-up SS were rib or metaphyseal fractures of the extremities.  Only 11% of initial cohort underwent repeat imaging. Noted new, and newly recognised, metacarpal fractures on follow up, argues against the proposal to omit hands/feet on follow up
Bennett et al., 2011110

Case series


Initial and repeat were full SS according to American College of Radiology (ACR) standards, 19 images. Oblique views of ribs not routinely obtained.

9-56 days

All had normal initial SS, 4 (8.5%) had abnormal follow up SS. 3 rib fractures, 1 proximal humerus


Unusual inclusion criteria of only those with a completely negative SS, yet still showed additional forensically relevant fractures. No detail as to why these children underwent repeat imaging


Karmazyn et al., 20111

Cross sectional

930 children
109/116 equivocal fractures re-imaged

Full SS (31 views), including oblique views initially. Repeat imaging only for equivocal findings


Nine hundred thirty children (515 boys and 415 girls) were included. Fractures were detected in 317 children (34%),

166 (18%) had multiple fractures. Common sites for fractures were the long bones (21%), ribs (10%), skull (7%), and clavicle (2%). Ten children (1%) had fractures in the spine (n = 3), pelvis (n = 1), hands (n = 6), and feet (n = 2).

All 10 children had other signs of physical abuse 29/116 (25%) definite fractures confirmed in previously equivocal findings

Retrospective study children < 2 years, 2003-2009

124/930 had new fractures on follow up. Main aim to propose reduced imaging for initial SS, propose excluding spine, pelvis, hands and feet, unless superficial injury to this area, as they accounted for 1% of fractures found. Cases described would suggest that some of these fractures were significant findings however

Sonik et al., 2010115

Case series


Full SS, no oblique views ribs initially.

11/22 follow up full SS, 11 no repeat skull imaging

11-29 days (mean 16.7)

New fractures identified in 3/22 patients (13.6%), one in whom initial SS was normal Retrospective study children <2 years undergoing repeat SS, 2003-2007. No details as to why these children underwent repeat imaging. Propose omitting AP pelvis and lateral spine. No oblique views, small numbers with no power calculation to support recommendation
Anilkumar et al., 2006109

Case series


Initial SS (including oblique views if age <1 year), follow up chest x-ray +/- oblique views

10 days – 3 weeks

3/59 (5.1%) had additional rib fractures noted on follow up

2/59 (3.4%) had rib fractures identified for the first time

Dating information was obtained in 3/59 patients (5.1%)

Retrospective study of children <2 years, 1998-2003, routinely invited for follow up from 1/1/2000

Only 59/200 cases returned for follow up

Zimmerman et al., 2005116

Prospective cohort


Initial and repeat were full SS, 19 images

Skull excluded from repeat survey

Mean 21.4 +/-9.7 days

22/48 children had additional information, 11 of whom had additional fractures identified. Additional fractures included rib, classic metaphyseal, clavicular, scapular, fibular and ulnar. In one child abuse was excluded by follow-up imaging Prospective review between 1998 and 2000. Indications included all infants with suspected physical abuse who had multiple fractures, fractures of varying ages, fractures inconsistent with history, concern for abuse not diagnosed initially, abnormal initial SS. Only 48/74 (65%) of those called for follow-up attended
Kleinman et al., 1996113

Retrospective cohort


Initial and repeat were full SS

Skull excluded from repeat survey, no obliques in first survey

14 days

13 (40.6%) children had 32 additional fractures identified. One initial SS was negative. Additional fractures included classic metaphyseal lesions, rib, spinal, pelvic and hand. Contributed to the dating of the injury for 13 of 70 fractures. One repeat SS confirmed original findings as a normal variant Retrospective review between 1990 and 1995, indications for repeat SS included high suspicion of abuse, original imaging inconsistent with history
Hansen et al., 2014111

Cross sectional


Full ACR skeletal survey at baseline and follow-up compared to limited view follow-up (excluding spine, pelvis, and skull)

10–42 days

The limited view follow-up would have missed eight spinal fractures in five children not visible on the original skeletal survey. All of these infants had additional fractures. Two infants had spinal fractures visible on chest view, two had further spinal abnormalities on initial skeletal survey and one had no other indications of spinal fracture. No pelvic fractures were identified on the follow-up SS that were not present on the initial SS. Those with pelvic fractures had a median of 7.5 other fractures. This study suggests that omitting pelvic images from the follow-up SS does not miss further fractures. Authors suggest omitting spinal views on the follow-up carries a low risk of missing significant spinal fractures

Key findings

Whilst the reason for undertaking follow up SS varied between studies, and the combination of films included in initial and follow up SS varied; additional findings to the initial SS were reported in all included studies (Table 2), this was true when the intial SS was normal or abnormal. Follow up SS identified new fractures and clarified equivocal findings and radiological findings in children who had normal initial SS.

Additional fractures identified on follow up SS included rib, long bone, CML, hands or feet, clavicle, fibular and ulnar fractures, vertebra and scapula, but not pelvic fractures.

Which radiology views should be included in SS

Twenty studies were included with children aged up to 16 years, most of the children were less than two years of age.1,26,73,75,76,79,81,82,83,89,91,94,99,117,118,119,120,121,122

Benefit of oblique views of the chest

Three studies showed significant benefit of oblique views of the chest,117,118,121 including a comparison of two view chest X-ray (anteroposterior and lateral) with a four view assessment including two additional oblique views of the ribs in 73 children.118 In this study, sensitivity improved by 17% (95% CI 2-36%, p=0.18) and specificity improved by 7% (95% CI 2-13%, p=0.004).118 The average improvement for diagnostic accuracy between three radiologists was 9% (CI 5-16%, p=0.005).118 Three children had rib fractures that were only seen on oblique films.118 The addition of oblique views increased detection rate by 19%121 in a study evaluating the benefit of oblique views in addition to a standard ACR skeletal survey in infants who all had at least one rib fracture.121 The four-view chest series, posteroanterior, lateral, right oblique, and left oblique radiographs, adds information to that obtained from the two-view chest series and increases the accuracy of diagnosing rib fractures in cases of possible physical abuse. The additional rib fractures noted or excluded using the four-views were predominantly posterior and lateral.117

3.2. Skeletal survey (SS) versus radionuclide imaging (RNI)

Eighteen studies of children aged up to 16 years were included.21,32,60,73,83,122,123,124,125,126,127,128,129,130,131,132,133 Eight studies compared the diagnostic yield in 509 children who had both investigations.122,123,126,127,129,130,131,132 In the included studies RNI was performed between 24 to 96 hours of the SS, however the number of images included in SS varied. No study included oblique views of the ribs and all studies, except for Pickett et al.,131 confirmed that using either investigation in isolation would miss some fractures.

Nine studies highlight additional findings on RNI not identified on SS.21,32,60,73,83,124,125,128,133 Five cases had RNI findings confirmed on repeat plain films,32,83,133one of which describes costo-vertebral fractures seen on RNI but not on SS.32RNI was more sensitive, overall, at identifying bony abnormalities than SS122,123,126,127,132 whilst two studies stated that SS had the greatest sensitivity.129,131 SS identifies metaphyseal fractures and skull fractures significantly better than RNI,129 however RNI had an increased sensitivity in detecting soft tissue trauma as well as bone trauma.123

Studies mostly found that neither SS or RNI is as good as the two investigations combined.122,123,126,127,129,132 RNI predominately missed skull, metaphyseal and epiphyseal fractures whereas skeletal survey missed rib fractures.60,122,126,127,129,132

3.3. Identifying less common abusive fractures

Four studies questioned the value of screening for rarer fractures.1,26,82,119,134 Children with these rarer (e.g. pelvis, hands and feet, sternum) fractures were reported to have either clinical findings or multiple additional fractures elsewhere.1 This study included 11 additional views to those carried out during standard SS, limiting its applicability. In 530 screened children, no pelvic, one spinal and nine fractures were identified to the feet, the radiation dose of pelvic imaging was highlighted and omitting the pelvis was recommended.119

Of 365 children 5.5% had fractures to the spine, hands and feet. Of the 25 spinal fractures identified in ten children (from Hangmans fracture at C2 to sacral fractures), all but one had associated non-spinal injuries. 1.4% of SS had hand fractures and 1.6% fractured feet.82 Spinal fractures were also reported in 14/751 children with skeletal surveys and or neuroimaging, 22 spinal fractures were observed. Four children had co-existent injuries, 4/14 were aged two to four years and seven had multiple spinal fractures.26

Benefits of lateral views in addition to standard frontal views of long bones

Significantly more metaphyseal fractures were seen on combined frontal and lateral views (p<0.01) but there was no significant difference for diaphyseal fractures.120 The levels of agreement between radiologists was found to improve with the addition of lateral views, especially in metaphyseal fractures. The study recommended the inclusion of coned metaphyseal views of knees and ankles within skeletal survey.

Which children with suspected abuse should be investigated for occult fractures?

Eighteen studies were included.1,17,75,80,122,135,136,137,138,139,140,141,142,143,144,145,146,147

Influence of ethnicity and socio-economic group

Ethnicity and socio-economic group has been reported to influence whether a child undergoes skeletal survey. African/American infants with unwitnessed head injury were more likely to have a SS than white infants (90.5% v 69.3%, p=0.01),146 however non-white or Hispanic children or those without private insurance were more likely to undergo screening.141 White children with private insurance were much less likely to have a SS than white children with no insurance/government insurance (50% v 88%, p<0.001).146,147 The introduction of a screening guideline reduced the inequities in SS conducted in white versus African/American children (rates of SS in white children increased from 69.3% to 84.6%, (p=0.05) but stayed the same amongst African/American children).141,146 

Influence of age on fracture detection

The diagnostic yield from SS correlates inversely with age and is significantly greater for children less than two years of age.17,75,122,135,136,138,144,148 The rate of occult fractures detected in children aged less than two years in two large-scale studies was 10% – 13%.1,138,149

A higher diagnostic rate for SS was identified in children less than six months of age than in those aged two years and six months,138 whilst two studies showed no difference in the yield of occult findings on SS in those aged less than one year and those aged one – two years.139,144 Significantly more rib fractures140 and classical metaphyseal lesions140,144 were found in children aged less than one year than those aged one to two years. A retrospective case series of skeletal survey found 14/17 (82%) of positive SS were in children less than one year of age,136 14/55 (25%) of children aged less than one year had positive SS.136

Screening siblings/household contacts

Siblings were assessed in three studies.136,143,145 There were six skeletal surveys performed on siblings (less than three years old) of children with abuse, SS was positive in one sibling.136 Skeletal survey was conducted in 134 household contacts aged less than two years, 16/134 had fractures, 9 of whom were aged less than six months;145 eight children had an isolated fracture and eight multiple fractures, none of which had clinical signs.145 Fractures were found in 9/16 twins on their SS giving an odds ratio of 20.1 (95% CI 5.8-69.9) for identifying a fracture in a twin of an abused child.145 Of 75 twin/triplet siblings screened, twins were more likely to have an occult fracture identified than non-twins.143

Indications for SS

Skeletal survey identified that 14% of children with abusive burns had occult fractures.139,148 The mean age of children with positive SS and burns was significantly older than non-burns cases (p=0.03),148 the fractures in children with burns included rib, CML, long bone, skull, and clavicular.139 Positive SS findings were also seen in 29% of infants aged less than one year with an unwitnessed head injury warranting child protection investigations.146 If the injury severity score was more than 15, cases were more likely to have positive findings than those with lower scores (OR 3.4, p<0.01).146

Abusive head trauma (AHT) was significantly associated with a positive SS, compared to children who underwent SS for other reasons (p<0.00).138 Three children presenting with an isolated skull fracture and no other signs of abuse had a positive SS.138 Skeletal survey was carried out in 86% of children aged less than 18 months presenting with an isolated skull fracture, of whom 6% had an additional fracture identified. Only one of nine with additional fractures was older than 6 months, eight of nine had a simple skull fracture.142

Further fractures were found in 12/201 (5.5%) children presenting with skull fracture in 201 children aged less than one year with a normal Glasgow Coma Scale (GCS).137 A lower incidence of additional fractures was found in 141 infants aged less than one year presenting with a skull fracture (non-MVC) who underwent SS with only two additional fractures identified, each had risk factors for child abuse.147 Additional fractures were lower limb metaphyseal fractures.147

Children presenting with apparent life threatening event (ALTE)/apnoea/seizures had a significantly higher rate of positive SS than those presenting for other reasons (p=0.05).138

What other imaging modalities may enhance the diagnosis of occult fractures?

16 studies with children aged up to 8 years were included.29,60,61,77,143,150,151,152,153,154,155,156,157,158,159,160,161

Computerised Tomography (CT)

Older studies suggested that Computerised tomography scan (CT) will miss skull fractures.150,158,159 however, use of 3D reconstruction is valuable in interpreting skull fractures.156

CT of the head with 3D reconstruction was compared to skull radiographs in children with suspected abusive head trauma5. In this group of 177 children, who were aged under one year, skeletal surveys were carried out including two views, anteroposterioir and lateral. Sixty-two (35%) had skull fractures detected by radiography whilst CT with 3-D reconstruction identified 67 (38%) skull fractures and was 97% sensitive and 94% specific for skull fracture. There was no significant difference between findings from plain radiographs and 3-D CT scans (p= 0.18).  In this study CT with 3-D reconstruction was shown to be equivalent to skull radiographs in identifying skull fractures.

Computerised tomography of chest may show rib fractures missed on two view chest radiography161 or on four view chest radiographs.157

Magnetic Resonance Imaging (MRI)

MRI may be valuable additional investigations for physeal or epiphyseal injures61 and  may identify trauma where plain films are equivocal.152

Whole body MRI (WB-MRI) was compared to initial and repeat skeletal survey combined. WB-MRI had a high specificity (95%) but low sensitivity (40%) for detecting fractures and was poor at identifying rib and metaphyseal fractures in particular.77

Ultrasound (U/S)

Studies assessing the use of ultrasound have found they can show metaphyseal fractures around the knee29,61,155 and also highlight periosteal haematoma of the tibia which was later confirmed as a fracture, and also a femoral fracture.160 Ultrasound can be helpful in identifying distal humeral epiphysiolysis,153 in the diagnosis of costo-chondral dislocation of the lower ribs29,48 and if performed on the chest can show acute rib fractures not apparent on plain film, including oblique views.154

Other imaging modalities
    • The use of (18F-NaF) positron emission tomography (PET)was compared to initial SS in 22 children (follow-up in 14) less than two years of age. PET versus SS had a sensitivity of 85% versus 72% for detecting all fractures, 92% versus 68% for thoracic fractures, and 67% versus 80%  for detecting classic metaphyseal lesions.162
    • PET identified three spinal fractures in one child missed on SS and confirmed on MRI.162
    • Bone scintigraphy (BS) was evaluated in the diagnosis of non-accidental injury and was found to be similar to SS in 74% of cases.4 In 4% (7/166) of the patients BS demonstrated a new finding and had a false-positive and false-negative rates of 2% and 13%, respectively. The two false-positive cases were because of a presumed infection of a clavicle and increased bony uptake. Of the 21 false negative cases, 16 had skull vault fractures, and five had metaphyseal fractures, which could be seen on film but not seen on BS. Routine use of BS as an adjunct to SS in the investigation of non-accidental injury aids confidence in diagnosis or identifies new findings in 12% of cases.

3.4. Research implications

Further research is needed in the following areas:

    • To evaluate the diagnostic yield in household contacts of childen with suspected abuse
    • To assess the role of radiological screening of disabled children with suspected abuse.

3.5. Limitations of review findings

Which investigation has a higher yield, skeletal survey (SS) or radionuclide imaging (RNI)?

    • Nine studies were published pre-1990; few studies published since 1998.
    • No study included oblique views of the ribs in the SS.
    • Rationales for choosing both investigations were not stated.

Does repeat SS enhance detection?

    • Oblique views of ribs not included in early studies.
    • Further clarification as to which infant’s merit follow-up skeletal survey would be of value.

What views should be included in a SS?

    • Older studies did not operate to current American College of Radiology/Royal College of Radiologists – Royal College of Paediatrics and Child Health national guidelines.
    • Decisions to include additional views (e.g. coned views) appear to be clinically determined, thus limiting their generalisability.

Which children with suspected abuse should be investigated for occult fractures?

    • All studies were retrospective.
    • The methodology of SS varied between studies and the indications for requesting a SS were ill-defined in some studies.
    • It was not possible to define the diagnostic yield of SS for two/three-year-olds.

What other imaging modalities may enhance detection of occult fractures?

    • Highly selected case series.

Thirteen studies were included with children aged up to 18 years.25,31,163,164,165,166,167,168,169,170,171,172,173 Data were not analysed by gender, disability, ethnicity or socio-economic group.

4.1. Cardiopulmonary resuscitation-related fractures

Rib fractures are a rare complication of cardiopulmonary resuscitation (CPR) in children, with the reported incidence of between 0-4.3% in 11 studies,31,163,164,165,166,167,169,170,171,172,173 six of which showed no rib fractures.31,164,167,169,171,172 One study of 80 children recorded an incidence of 18.7%.166

The fractures associated with CPR, where reported, were all multiple, anterior/anterolateral, unilateral or bilateral and there were no posterior rib fractures reported due to CPR. Two infants with rib fractures ascribed to CPR, one anterior and one postero-lateral following bimanual chest compression.25 A rise in incidence of CPR related fractures was been noted and hypothesised that this was associated with the introduction of two-thumb rather than two-finger resuscitation.170,174,175


Rib fractures were sought at post-mortem in ten studies. No authors referred to the use of specimen radiography.163,164,165,166,167,168,170,171,172,173 Nine studies explored the prevalence of rib fractures following CPR.31,163,164,165,166,167,169,171,172

CPR related fractures have been reported. Three children, a two-month-old; a three-month-old who died of sudden infant death syndrome; a five-year-old who drowned, all were resuscitated by trained personnel and sustained fractures. One child, the three-month-old, was also resuscitated by untrained personnel for a total of 75 minutes. There was no concern about child abuse in any of these cases163,165. A further assessment of post-resuscitative clinical features found that 15/80 children aged 0-18 years sustained CPR induced rib fractures on early chest x-ray, in these cases abuse had been excluded, although admission was trauma-related.166 A further included study found 24/546 SUDI cases sustained rib fractures, 15 were healing fractures, ten of which had features of abusive injury.173 Of nine acute fractures, seven had no features of abusive injury and were thought to be CPR related. The fractures were multiple, anterolateral, and involved 3rd – 6th ribs; bilateral in one case.173 Almost all (14/15) healing rib fractures were visible on post-mortem skeletal survey (SS), one case had only anantemortem chest radiographs.173 Only 2/9 acute rib fractures were visible on post-mortem SS, found only by inspection of ribs following stripping of the pleura.173

In contrast, a larger study, of 382 children undergoing CPR, found no cases developed rib fractures, they all had a post-mortem examination including rib palpation and stripping of the parietal pleura.167 Resuscitation of 153 children for 1 – 540 minutes resulted in no cases sustaining rib fractures.171

The location of fractures was described in a case series of five children sustaining rib fractures post-CPR noted that all children had by-stander CPR using one-handed and two-handed techniques, fractures found were anterolateral, 3rd–6th ribs, 2/5 bilateral.168 A further 9/571 infants aged 0-6 months that exhibited rib fractures attributed to CPR lasting 21-260 minutes, fractures were all anterior to lateral. Almost three times as many fractures occurred between 2006-2008 versus 1997-2005.170,174,175 It is hypothesised that these findings might correlate with the change in CPR technique as recommended by the International Liaison Committee on Resuscitation (ILCOR), however no attempt was made to determine what method of CPR had been conducted.170,174,175

4.2. Research implications

With the change in International Liaison Committee on Resuscitation (ILCOR) guidelines for resuscitation of infants recommending a two-thumb technique, further studies will be required to determine the incidence of rib fractures with this technique, using more sensitive radiological and autopsy techniques post-CPR, e.g. four view chest films/radionuclide imaging/specimen radiography.

4.3. Limitations of review findings

The included studies were retrospective and included a wide age range of children, however the more recent studies have included infants less than one year old. The methodology of determining rib fractures was suboptimal in some studies where either early radiology or post-mortem without reflection of the pleura has been undertaken. The inclusion of trauma patients is problematic in determining the aetiology of the rib fractures.

The review identified several interesting findings that were outside of the inclusion criteria, these are as follows:

Clinical question 1: Which fractures are indicative of abuse?

Rib fractures

    • Post-mortem studies should include reflection of the pleura to ensure that all fractures are accurately identified.176

Imaging strategies

    • Posterior rib fractures may not be seen on a skeletal survey but have been identified at post-mortem.176


    • Rib fractures have been described as a consequence of chest physiotherapy for bronchiolitis in France.177 The method of physiotherapy was not described and all children received this therapy unsupervised in their own homes.177
    • However, a study of 647 children undergoing chest physiotherapy resulted in no rib fractures.178

Femoral fractures

    • There has been a decrease in the incidence of femoral shaft fractures in children, reduction of 42% between 1987–2005.179
    • The commonest cause of femoral shaft fracture in children less than four years of age is a fall of less than one metre.179
    • A fracture classification system to distinguish transverse, oblique or spiral fractures was developed and validated. There was moderate inter-observer reliability.180

Skull fractures

    • Biparietal skull fractures might result from a single blow to the occiput, as described following accidental injury.181
    • An influential study of severely abused children stated that depressed, diastatic, growing, and multiple fractures were more common in abuse than in non-abuse. The study included a considerable number of fatally abused children, and excluded unintentional fractures from motor vehicle injury and was not eligible for inclusion in this systematic review.182

Tibial fracture

    • Undisplaced spiral fracture of the tibia without a concomitant fibular fracture is most likely to be an unintentional toddler fracture particularly if the child is a boy less than 2.5 years of age.183

Humeral fracture

    • A case series of seven infants aged 4-7 months with isolated humeral fractures where the explanation was when the infant rolled over. Skeletal surveys were negative, all cases arose as a consequence of court proceedings.184
    • Radiological identification of distal humeral epiphyseal separation is aided by the use of ultrasound in young infants.185

Metaphyseal fractures

    • A 1984 study reported that passive excercise physiotherapy administered, particularly to preterm infants, had caused metaphyseal fractures.186
    • Metaphyseal fractures have been recorded in serial casting of clubfoot.187
    • Birth trauma can cause metaphyseal fractures in breech delivery.188
    • External cephalic version for breech presentation may result in CML.189
    • CML occurring following birth was associated with pain expressed as irritability, lack of spontaneous movement of the affected leg and poor feeding in an infant.189
    • A study of high resolution CT defines the precise fracture plane that occurs in metaphyseal fractures, contributing to an understanding of the biomechanics of these fractures.190


    • Post-mortem CT has a low sensitivity for rib fractures in comparison to autopsy.191
    • Proposedautopsy techniques to maximise identification of fractures are described.192
    • It is proposed that cone-beam CT performed during post-mortem may aid in the dating of fractures.193
    • Vertebral clefts may be visible through the vertebral body and be confused with fractures. A post-mortem study discusses ten affected foetuses.194

Important clinical differentials

    • Sternum ossification centres that projected over ribs on a chest film have been mistaken for fractures. Follow up imaging revealed real cause.195
    • Diffuse cortical thickening on the medial aspect of the tibia, mimicking periosteal reaction may result from intraosseous needle insertion, but has been mistaken for abuse.196
    • An overview of clinical variants that are important when reviewing skeletal surveys.197
    • Heterotopic ossification from the ischium and sacrum to proximal femur posteriorly secondary to physical abuse.198
    • Metaphyseal fragmentation may be noted in infants, simulating metaphyseal fractures.199
    • It is important to distinguish suture variants within the occiput from fractures.200
    • An unossified membranous strip within the parietal bone may be mistaken for skull fracture in infants.201
    • Widespread medullary necrosis and periosteal reaction, with epiphyseal sparing, described as a complication of traumatic pancreatitis.202
    • Useful infomation is described about the histopathology of vitamin D deficiency compared to NAI fractures.203


    • Descriptions of fractures determined by type and rates of stress and strain applied to a bone.183,204,205,206,207
    • Reviews of biomechanics highlight numerous variables relating to childhood fractures.183,204,205,206,207
    • Studies of falls down stairs highlighted that the peak age are children aged one year, sustaining predominantly minor injuries with 4/18 sustaining skull fracture and 10% with limb fractures.208

Consequences of abusive fractures

    • Compartment syndrome may occur in the lower limbs as a consequence of abusive fractures.209

Presenting features

    • Twenty-one percent of children with abusive fractures were missed at initial presentation.210
    • Boys with abusive extremity fractures attending non-pediatric emergency department or primary care were most commonly missed.210
    • Delay (more than eight hours) in presentation with an extremity fracture was evaluated in 206 children211. Although the median time to presentation was one hour, 21% presented after eight hours, 15% showed no external sign of injury and 12% used the injured extremity normally. However, all children had at least one sign or symptom.211
    • A study of the fracture pattern at diagnosis of 68 children with Osteogenesis Imperfecta (OI) showed that although rib fractures were identified in 21% none of these occurred in infancy. ‘All subjects with more than two fractures were diagnosed prenatally or in the immediate newborn period. Seventeen (25%) infants were diagnosed after one week of age but before 12 months of age. None of these infants had either rib fractures or more than one fracture at the time of diagnosis.212

Birth related fractures

    • A study of birth related fractures, including metaphyseal, confirms that these fractures are painful and may be associated with tenderness and swelling.213
    • Multiple rib fractures described in macrosomic infant with shoulder dystocia.214,215
    • Posterior rib fractures described as birth injuries, some macrosomic216,217 including those with shoulder dystocia.214,215
    • Classic metaphyseal lesions of the femur have been noted after caesarian section. Two cases were breech presentation.218
    • Femoral fractures are a rare birth injury (incidence 0.13/1000) as recorded in Ireland between 1996–1999. The typical fracture occurring was a spiral fracture of the proximal half of the femur which was held in an extended position. There were 5/7 affected infants that were delivered by caesarean section. In 6/7 cases, no evidence of femoral injury was noted on immediate post-natal examination.219
    • Birth injury can cause depressed skull fractures.220
    • Rib fractures have been identified in approximately 2% of ex pre-term infants (less than 37 weeks gestation).221

Age and likelihood of abuse

    • Fracture requiring orthopaedic management was present in 28% of 11,554 inpatients with a diagnosis of child abuse. The relative risk for child abuse in children aged less than one year was 11.46; 3.07 for those aged one to two years.222
    • The proportion of fractures rated as abusive in children aged less than three years attending a single centre fell by up to 50% over 24 years.223

Growth Recovery lines

    • Zapala et al. compared the likelihood of encountering growth recovery lines on skeletal surveys in 52 infants at low and 21 at high risk of abuse and suggested that abused infants are prone to a temporary disturbance in endochondral ossification as a result of episodic physiological stresses.224
    • Prevalence of growth recovery lines was significantly different between groups, the low-risk group was 38% versus 71% in the high-risk group (p<0.001).224

Clinical question 2: What is the evidence for radiological dating of fractures in children?

    • Repeat skeletal survey may aid in fracture dating.113
    • Bone scans have no place in fracture dating as they become positive within seven hours and can remain positive for up to one year.225
    • Study of animal models for dating rib fractures.226
    • A small study of cone-beam CT on post-mortem specimens suggested that this may be an accurate estimate of the age of the fracture.193
    • A study comparing whole body magnetic resonance imaging (WB-MRI) with skeletal survey noted that WB-MRI provided little information regarding fracture age.227
    • Study of radiological features of healing accidental fractures – with very clear time frame (because all sustained at birth).6

Clinical question 3: What radiological investigations should be performed to identify fractures in suspected child abuse?

Practice guidelines

    • American and British guidelines for practice have specific guidance on SS views to be taken.228,229,230,231
    • Oblique views of the ribs are recommended when rib fractures are evident, and consideration should be given to including oblique views in the standard survey protocol.228
    • Follow-up Skeletal Survey (SS) is recommended when there are abnormal or equivocal findings on the initial SS, or when abuse is suspected clinically.228
    • Studies addressing concerns about radiation dosage.232,233
    • Audit of UK SS three years after the British Society of Paediatric Radiology published their standards: revelaed that only 15% included all appropriate views and technical quality has considerably improved (score 9.7/11).234
    • Forty studies in the US were assessed for the performance of SS or Radionuclide Imaging of children with suspected physical abuse aged less than two years or infants less than on year with non-vehicle associated head injury or femoral fractures:235
      • Eight-three percent of children less than two years old underwent appropriate screening, 68% of those aged less than one with a head injury and 77% of those less than one with a femoral fracture. Influential variables for appropriate screening were injury severity and year of admission235
    • A helpful study in guiding day to day use of skeletal survey using various theoretical clinical scenarios.236
    • Information on the radiation dose delivered by skeletal survey which is a common concern for parents/carers. The information is valuable to clinicians counselling families about skeletal survey.237
    • Study relates to follow up skeletal survey recommendations and completion rates. Significant factors for new injury identification are stated. New fractures identified in 16%. Possibly warrants further evaluation/research to develop objective guidelines for recommendations for follow up SS.7
    • Important conclusions relating to sentinel injuries and their investigation are presented.238
    • 3D reconstructions of head CT are useful at evaluating possible skull fractures.239

Alternative imaging techniques

    • Ultrasound (US) has been shown to be useful in detecting occult rib fractures in adults. These studies show an increased sensitivity of US over standard radiography, particularly in the cartilaginous portion of the rib.240,241
    • Evaluation of separation of the distal humerus epiphysis is well defined byUS, particularly in neonates where ossification is minimal.242
    • The use of 18F-NaF positron emission tomography (PET) whole-body imaging has been shown to demonstrate additional subtle fractures including classic metaphyseal lesion of the humerus and iliac crest fractures not seen on initial SS.162
    • Multi-planar computerised tomography scan (CT) and 3D image reconstructions may enhance the visualisation of rib fractures243. However, even these imaging modalities may miss rib fractures as detailed in post-mortem studies.244,245
    • Post-mortem CT may be of value in detecting incomplete buckle rib fractures.246
    • A study of 605 CT images of children aged 0-3 years highlighted that 53% had Wormian bones, the majority of which were multiple.247
    • CT scan of chest has been shown to be more sensitive than the initial SS in demonstrating acute rib fractures.248

Other potential indications for imaging

    • Multiple birth infants appear to be at higher risk of fractures or abdominal injuries than other siblings.249
    • An exploration of 320 children less than 2 years of age presenting with a single extremity fracture, 37% of whom underwent neuroimaging, identified only 5 children (aged less than one year) with traumatic findings but none had clinically significant head injury.250
    • In 146 infants less than six months of age presenting to child abuse physicians with an isolated bruise, 23.3% had occult fractures identified on skeletal survey.251

Clinical question 4: Does cardiopulmonary resuscitation cause rib fractures in children?

    • Histological dating of fractures may be crucial to distinguish abusive from cardiopulmonary resuscitation (CPR) related fractures.31,252,253
    • Posterior rib fractures in a child who has been resuscitated on a firm surface would appear inconsistent with the biomechanics of resuscitation.254
    • One case study proposed that a 21-month-old with fatal head injuries sustained a posterior rib fracture as a consequence of CPR performed by trained personnel:255
      • However, this 21-month child had a co-existent unexplained pelvic fracture and absence of rib fracture prior to CPR was only determined by an abdominal computerised tomography scan255
      • The paper itself contains much debate as to whether the case was abused or not and the reader can not confidently conclude that this was a confirmed case of abuse.255
      • CPR technique utilising a mannikin demonstrates that the majority of practitioners are likely to over-compress in the absence of real time feedback.256

We performed an all-language literature search of original articles, their references and conference abstracts published since 1950. The initial search strategy (it should be noted that Q4 required a separate search strategy) was developed across OVID Medline databases using keywords and Medical Subject Headings (MeSH headings) and was modified appropriately to search the remaining bibliographic databases.  The search sensitivity was augmented by the use of a range of supplementary ‘snowballing’ techniques including consultation with subject experts and relevant organisations, and hand searching selected websites, non-indexed journals and the references of all full-text articles.

Standardised data extraction and critical appraisal forms were based on criteria defined by the National Health Service’s Centre for Reviews and Dissemination.257 We also used a selection of systematic review advisory articles to develop our critical appraisal forms.134,258,259,260,261 Articles were independently reviewed by two reviewers. A third review was undertaken to resolve disagreement between the initial reviewers when determining either the evidence type of the article or whether the study met the inclusion criteria. Decisions related to inclusion and exclusion criteria were guided by Cardiff Child Protection Systematic Reviews, who laid out the basic parameters for selecting the studies.

Our panel of reviewers included paediatricians, paediatric radiologists, orthopedic surgeons, reserach officers, designated and named doctors and specialist nurses in child protection. All reviewers underwent standardised critical appraisal training, based on the CRD critical appraisal standards.134

Inclusion criteria

General criteria


Inclusion Exclusion
Papers with all evidence types Personal practice
English and non-English papers Review papers
Papers from conferences – paediatric, radiology, orthopaedic Management of fractures papers
Patients between 0-17 years of age Papers where the population included adults and children
Studies of non-abusive data only

Additional inclusion criteria specific review questions

1. Which fractures are indicative of abuse?


Inclusion Exclusion
General inclusion criteria plus: General exclusion criteria plus:
Comparative studies of children less than 17 years of age with fractures either abusive or non-abusive Rank of abuse is 4 or 5 or mixed rank where relevant cases can not be extracted
Non-comparative studies of other fracture types found in abuse where comparative data was unavailable Studies addressing abusive/non-abusive fractures only
Single case reports
Studies of outcome or management of abusive fractures

2. What is the evidence for radiological dating of fractures in children?


Inclusion Exclusion
General inclusion criteria plus: General exclusion criteria plus:
Primary research addressing how you can date fractures radiologically in children up to 17 years of age Underlying bone disease was present
Criteria for dating was not detailed

3. What radiological investigations should be performed to identify fractures in suspected child abuse?


Inclusion Exclusion
General inclusion criteria plus: General exclusion criteria plus:
Children (0-17) who had radiological investigations to identify bone fractures in suspected child abuse Studies where details on the yield from the investigations were not available
Fatal abuse

4. Does cardiopulmonary resuscitation cause rib fractures in children?


Inclusion Exclusion
General inclusion criteria plus: General exclusion criteria plus:
External closed cardiopulmonary resuscitation (CPR) Studies relating to complications or outcomes of CPR (other than rib fractures)
No underlying bone disease or child abuse as the cause of collapse Inadequate quality of confirmation of fractures
Incidence of associated rib fractures was recorded

Ranking of abuse

Distinguishing abuse from non-abuse is central to our review questions. As our reviews span more than 40 years, standards for defining abuse have changed markedly.  We have devised the following ranking score where “1” indicates the highest level of confidence that abuse has taken place.  These rankings are used throughout our systematic reviews (where appropriate).


Ranking Criteria used to define abuse
1 Abuse confirmed at case conference or civil or criminal court proceedings or admitted by perpetrator
2 Abuse confirmed by stated criteria including multidiscpilinary assessment
3 Abuse defined by stated criteria
4 Abuse stated but no supporting detail given
5 Suspected abuse

Ranking of evidence by study type


T1 Randomised controlled trial (RCT)
T2 Controlled trial (CT)
T3 Controlled before-and-after intervention study (CBA)
O1 Cohort study/longitudinal study
O2 Case-control study
O3 Cross-sectional
O4 Study using qualitative methods only
O5 Case series
O6 Case study
X Formal consensus or other professional (expert) opinion (automatic exclusion)

Additional criteria for specific review questions

1. Which fractures are indicative of abuse?

As above

2. What is the evidence for radiological dating of fractures in children?

Studies were graded for quality based upon study design, accurate documentation of time of injury and by standardised criteria for radiological dating

3. What radiological investigations should be performed to identify fractures in suspected child abuse?

We also used the following ranking of skeletal survey (SS)


 Ranking Criteria used to define SS
1 SS to British Society of Paediatric Radiology/American College of Radiology standards, including oblique views of ribs
2 SS of all bones: axial/limbs/hands/feet/skull/pelvis/spine.  Views taken specified
3 SS of skull/long bones/chest/pelvis. No mention of hands or feet
4 X-ray of skeleton including multiple bone radiology.  No definition of what this included
5 Baby-gram
4. Does cardiopulmonary resuscitation cause rib fractures in children?

The authors were careful to give the cause of cardiorespiratory collapse and ranks 4/5 were excluded prior to abuse.

Definition of levels of evidence and grading practice recommendations

Practice recommendations.257,260 This classification is based on the Bandolier system adapted to include the Centre for Reviews and Dissemination’s guidance for undertaking reviews.257


 Grade Level Type of evidence
A Ia Evidence obtained from a well designed randomised controlled trial of appropriate size (T1)
B Ib Evidence obtained from a well designed controlled trial without randomisation (T2, T3)
B IIa Evidence obtained from a well designed controlled observational study e.g. cohort, case-control or cross-sectional studies. (Also include studies using purely qualitative methods) (O1, O2)
C IIb Evidence obtained from a well designed uncontrolled observational study (O3, O4)
C III Evidence obtained from studies that are case study or case series (O5, O6)

Search strategy

Medline search strategy used for the 2017 fractures search questions 1-3:

1. Which fractures are indicative of abuse?
2. What is the evidence for radiological dating of fractures in children?
3. What radiological investigations should be performed to identify fractures in suspected child abuse?


1. Child

2. child

3. (battered child or shaken baby or battered baby).mp.

4. 1 or 2 or 3

5. (child: or infant: or baby or toddler:).mp.



8. 5 or 6 or 7

9. non-accidental

10. (non-accidental trauma or nonaccidental trauma).mp.

11. (non-accidental: and injur:).mp.

12. soft tissue

13. physical

14. (or/9-13) and 8

15. 4 or 14

16. Fractures, Ununited/or Radius Fractures/or Fractures, Malunited/or Tibial Fractures/or Fractures, Bone/ or Rib Fractures/ or Femoral Neck Fractures/or Femoral Fractures/or Humeral Fractures/or Shoulder Fractures/or Fractures, Compression/or Fractures, Cartilage/or Hip Fractures/or Intra-Articular Fractures/or Fractures, Open/or Fractures, Closed/or Fractures, Comminuted/


18. Fractures, Bone/

19. rib

20. (multiple skull fractur: or eggshell fractur: or skull fractur:).mp.

21. femoral

22. humeral

23. pelvic

24. (spiral fractur: or transverse fractur:).mp.

25. metaphyseal

26. (corner fractur: or bucket handle fractur:).mp.

27. metaphyseal chip

28. classic metaphyseal

29. or/16-28

30. (investigat: adj3 fract:).mp.

31. (radiolog: adj3 fractur:).mp.

32. (roentgen: adj3 fract:).mp.

33. skeletal

34. ((paediatric or pediatric) adj3 radiolog:).mp.

35. ((paediatric or pediatric) adj3 nuclear medicine).mp.

36. Tomography, X-Ray Computed/


38. (bone scan or X rays).mp.

39. skeletal

40. isotope bone

41. or/30-40


43. (timing adj3 healing).mp.

44. (pattern: adj3 fractur:).mp.

45. ((dating or date) adj3 fractur:).mp.

46. (ag: adj3 fractur:).mp.

47. or/42-46

48. 41 or 47

49. 15 and 29 and 48

50. 8 and 29 and 47

51. 49 or 50

52. limit 51 to (humans and yr=”2014 -Current”)

53. limit 52 to humans

Medline search strategy used for the 2017 fractures search question 4: Does cardiopulmonary resuscitation cause rib fractures in children?


1     exp child/


3     exp Infant, Newborn/

4     (child* or babies or baby or toddler*).mp.

5     (infancy or infant* or neonat*).tw.

6     (pediatric* or paediatric*).tw.

7     or/1-6

8     metaphyseal fractur*.mp.

9     rib fractur*.mp

10     clavicle fractur*.mp.

11     Rib Fractures/

12     Diaphragm/in [Injuries]

13     “Wounds and Injuries”/

14     Wounds, Nonpenetrating/

15     Thoracic Injuries/

16     thoracic Injur*.tw.

17     ((rupture or trauma or injur*) adj3 (chest or lung or rib)).tw.

18     or/8-17

19     exp Cardiopulmonary Resuscitation/

20     cardio-pulmonary


22     cardiac

23     heart

24     or/19-23

25     7 and 18 and 24

26     limit 25 to yr=”2014-Current”


Please note that the final search string is missing from the search strategy published in the following publication

Kemp AM, Dunstan F, Harrison S, Morris S., Mann M, Rolfe K, Datta S, Thomas DP, Sibert JR, Maguire S. Patterns of skeletal fractures in child abuse: systematic review. British Medical Journal. 2008;337:a1518

The correct search strategy (at the time of publication of this paper) is found below:

    1. Child
    2. child
    3. (battered child or shaken baby or battered baby).mp.
    4. 1 or 2 or 3
    5. (child: or infant: or baby).mp.
    6. non-accidental
    7. non-accidental
    8. (non-accidental: and injur:).mp.
    9. soft tissue
    10. physical
    11. (or/6-10) and 5
    12. 4 or 11
    14. 12 and 13
    15. (investigat: adj3 fract:).mp.
    16. (radiolog: adj3 fractur:).mp.
    17. (roentgen: adj3 fract:).mp.
    18. skeletal
    19. bone
    20. Isotope Bone
    23. Tomography, X-Ray Computed/
    24. ((paediatric or pediatric) adj3 radiolog:).mp.
    25. ((paediatric or pediatric) adj3 nuclear medicine).mp.
    26. or/15-25
    27. (ageing adj3 fractur:).mp.
    28. ((dating or date) adj3 fractur:).mp.
    29. (pattern: adj3 fractur:).mp.
    31. or/27-30
    32. 26 or 31
    33. 14 and 32

Databases searched

Databases searched for both cardiopulmonary resuscitation and fractures review:


Databases Time period searched
ASSIA (Applied Social Sciences Index and Abstracts) 1987 – 2017
Child Data 1958 – 2009
CINAHL (Cumulative Index to Nursing and Allied Health Literature) 1982 – 2017
Cochrane Central Register of Controlled Trials (CENTRAL) 1996 – 2017
EMBASE 1980 – 2017
MEDLINE 1950 – 2017
MEDLINE In-Process and Other Non-Indexed Citations 1951 – 2017
Open SIGLE (System for Information on Grey Literature in Europe) 1980 – 2005*
Scopus 2009 – 2017
Social Care online (previously Caredata) 1970 – 2015
Trip Plus 1997 – 2005
Web of Knowledge — ISI Proceedings 1990 – 2017
Web of Knowledge — ISI Science Citation Index 1981 – 2017
Web of Knowledge — ISI Social Science Citation Index 2008 – 2017
* ceased indexing
 institutional access terminated
 no yield so ceased searching
Journals ‘hand searched’ Time period searched
Child Abuse and Neglect 1977 – 2015
Child Abuse Review 1992 – 2015
Websites searched Date accessed
The Alberta Research Centre for Health Evidence (ARCHE) January 2015
Child Welfare Information Gateway (formerly National
Clearinghouse on Child Abuse and Neglect)
January 2015

Pre-review screening and critical appraisal

Papers found in the database and hand searches underwent three rounds of screening before they were included in this update. The first round was a title screen where papers that obviously did not meet the inclusion criteria were excluded. The second was an abstract screen where papers that did not meet the inclusion criteria based on the information provided in the abstract were excluded. In this round the pre-review screening form was completed for each paper. These first two stages were carried out by systematic reviewer at the RCPCH and a clinical expert. Finally a full text screen with a critical appraisal was carried out by members of the  panel of expert reviewers. Critical appraisal forms were completed for each of the papers reviewed at this stage. Examples of the pre-review screening and critical appraisal forms used in previous reviews are available on request (


We analysed data according to fracture sites. Where it was possible to obtain overall estimates of the probability of abuse for a particular fracture type from cross-sectional studies, we undertook a meta-analysis and presented the result as forest plots.

We pooled estimates from individual studies using the method of De Simonian and Laird, deriving a confidence interval for the pooled estimate and testing for heterogeneity between studies262. Given the varied nature of the studies, our work was inevitably a pooled estimate of different populations rather than a single one. Insufficient detail was shown in most papers to be able to analyse by age and so in the forest plots the studies were ranked in increasing order of mean age in the abused cases, as far as was possible. The results were generally summarised as the proportion of children with a given fracture type who are classed as abused – that is the predictive value of the fracture type for identifying abuse. Proportions were compared between groups using the chi-square test or Fisher’s exact test where that was appropriate.

Publications arising from fractures review

Prosser I, Maguire S, Harrison S, Mann M, Sibert J, Kemp A. Review. How old is this fracture? Radiological dating of fractures in children: a systematic review. American Journal of Roentgenology. 2005;184(4):1282-1286

Maguire S, Mann M, John N, Ellaway B, Sibert JR, Kemp AM. Does cardiopulmonary resuscitation cause rib fractures in children? A systematic review. Child Abuse and Neglect. 2006;30(7):739-751

Kemp AM, Bulter A, Morris S, Mann M, Kemp KW, Rolfe R, Silbert J, Maguire S. Which radiological investigations should be performed to identify fractures in suspected child abuse? Clinical Radiology. 2006;61(9):723-736

Kemp AM, Dunstan F, Harrison S, Morris S., Mann M, Rolfe K, Datta S, Thomas DP, Sibert JR, Maguire S. Patterns of skeletal fractures in child abuse: systematic review. British Medical Journal. 2008;337:a1518

Kemp AM. Fractures in physical child abuse. Paediatrics and Child Health. 2008;18(12):550-553

Maguire S, Cowley L, Mann M, Kemp A. What does the recent literature add to the identification and investigation of fractures in child abuse: an overview of review updates 2005 – 2013. Evidence Based Child Health: A Cochrane Review Journal. 2013;8(5):2044-2057.

Primary studies arising from fractures review

Prosser I, Lawson Z, Evans A, Harrison S, Morris S, Maguire S, Kemp AM. A timetable for the radiologic features of fracture healing in young children. American Journal of Roentgenology. 2012;198(5):1014-1020

Cosway B, Mathura N, Mott A, Bredow M, Fraser J, Rawlinson A, Wei C, Thyagarajan M, Harrison S, Kemp A. Occult rib fractures: Defining the cause. Child Abuse Review. 2013.

Disclaimer: This is a summary of the systematic review findings up to the date of our most recent literature search. If you have a specific clinical case, we strongly recommend you read all of the relevant references as cited and look for additional material published outside our search dates.

Original reviews and content © Cardiff University, funded by NSPCC

Published by RCPCH April 2018

While the format of each review has been revised to fit the style of the College and amalgamated into a comprehensive document, the content remains unchanged until reviewed and new evidence is identified and added to the evidence-base. Updated content will be indicated on individual review pages.



Murphy R., Kelly D.M., Moisan A., et al. Transverse fractures of the femoral shaft are a better predictor of nonaccidental trauma in young children than spiral fractures are. Journal of Bone & Joint Surgery, American Volume 2015; 97(2): 106-111.


Jones J.C., Feldman K.W., Bruckner J.D. Child abuse in infants with proximal physeal injuries of the femur. Pediatr Emerg Care 2004; 20(3): 157-161.


Haney S.B., Boos S.C., Kutz T.J., et al. Transverse fracture of the distal femoral metadiaphysis: a plausible accidental mechanism. Pediatr Emerg Care 2009; 25(12): 841-844.


Arkader A., Friedman J.E., Warner W.C., Jr., et al. Complete distal femoral metaphyseal fractures: a harbinger of child abuse before walking age. J Pediatr Orthop 2007; 27(7): 751-753.


Pandya N.K., Baldwin K.D., Wolfgruber H., et al. Humerus fractures in the pediatric population: an algorithm to identify abuse. J Pediatr Orthop B 2010; 19(6): 535-541.


DeLee J.C., Wilkins K.E., Rogers L.F., et al. Fracture-separation of the distal humeral epiphysis. J Bone Joint Surg Am 1980; 62(1): 46-51.


Lindberg D.M., Harper N.S., Laskey A.L., et al. Prevalence of abusive fractures of the hands, feet, spine, or pelvis on skeletal survey: perhaps “uncommon” is more common than suggested. Pediatr Emerg Care 2013; 29(1): 26-29.


Hechter S., Huyer D., Manson D. Sternal fractures as a manifestation of abusive injury in children. Pediatr Radiol 2002; 32(12): 902-906.


Bode K.S., Newton P.O. Pediatric nonaccidental trauma thoracolumbar fracture-dislocation: posterior spinal fusion with pedicle screw fixation in an 8-month-old boy. Spine (Phila Pa 1976) 2007; 32(14): E388-393.


Feldman K.W., Avellino A.M., Sugar N.F., et al. Cervical spinal cord injury in abused children. Pediatr Emerg Care 2008; 24(4): 222-227.


Gille P., Bonneville J., Francois J., et al. Fracture des pédicules de l’axis chez un nourrisson battu (Fractures of axis pedicles in battered infant). Chirurgie Pediatrique 1980; 21(5): 343-538.


Oral R., Rahhal R., Elshershari H., et al. Intentional avulsion fracture of the second cervical vertebra in a hypotonic child. Pediatr Emerg Care 2006; 22(5): 352-354.


Singh R., Squires J., Fromkin J.B., et al. Assessing the use of follow-up skeletal surveys in children with suspected physical abuse. J Trauma Acute Care Surg 2012; 73(4): 972-976.


Williams G., Treves S.T. A second radiographic skeletal survey for child abuse triggered by bone scintigraphy found positive after the initial survey was called negative. Clin Nucl Med 2007; 32(1): 29-31.


Critical Appraisal Skills Programme (CASP).


Deye K.P., Berger R.P., Lindberg D.M. Occult abusive injuries in infants with apparently isolated skull fractures. J Trauma Acute Care Surg 2013; 74(6): 1553-1558.


Higginbotham N., Lawson K.A., Gettig K., et al. Utility of a child abuse screening guideline in an urban pediatric emergency department. J Trauma Acute Care Surg 2014; 76(3): 871-877


Hicks R.A., Stolfi A. Skeletal surveys in children with burns caused by child abuse. Pediatr Emerg Care 2007; 23(5): 308-313.


Lindberg D.M., Berger R.P., Reynolds M.S., et al. Yield of skeletal survey by age in children referred to abuse specialists. J Pediatr 2014; 164(6): 1268-1273 e1261.


Drubach L.A., Johnston P.R., Newton A.W., et al. Skeletal trauma in child abuse: detection with 18F-NaF PET. Radiology 2010; 255(1): 173-181.


Warkentine F.H., Horowitz R., Pierce M.C. The use of ultrasound to detect occult or unsuspected fractures in child abuse. Pediatr Emerg Care 2014; 30(1): 43-46.


Matshes E.W., Lew E.O. Do resuscitation-related injuries kill infants and children? Am J Forensic Med Pathol 2010; 31(2): 178-185.


Matshes E.W., Lew E.O. Two-handed cardiopulmonary resuscitation can cause rib fractures in infants. Am J Forensic Med Pathol 2010; 31(4): 303-307.


Heideken J., Svensson T., Blomqvist P., et al. Incidence and trends in femur shaft fractures in Swedish children between 1987 and 2005. J Pediatr Orthop 2011; 31(5): 512-519.


Sawyer J.R., Kapoor M., Gonzales M.H., et al. Heterotopic ossification of the hip after non-accidental injury in a child: case report. J Pediatr Orthop 2009; 29(8): 865-867.


Greeley C.S., Donaruma-Kwoh M., Vettimattam M., et al. Fractures at diagnosis in infants and children with osteogenesis imperfecta. J Pediatr Orthop 2013; 33(1): 32-36.


Rajegowda BK, Chintalapalli P, Peralta D, et al. Unexpected detection of a newborn with multiple rib and clavicular fractures: birth injury? . Neonatal Intensive Care 2007; 20(1): 19-20.


Bullock D.P., Koval K.J., Moen K.Y., et al. Hospitalized cases of child abuse in America: who, what, when, and where. J Pediatr Orthop 2009; 29(3): 231-237.


Zumwalt R., Fanizza-Orphanos A. Dating of healing rib fractures in fatal child abuse. Advanced Pathology 1990; 3: 193-205.


Practice Guideline for Skeletal Surveys in Children (PDF).


Non-accidental injury standard for skeletal surveys.


Standards for Radiological Investigations of Suspected Non-accidental Injury. Joint document produced in collaboration with the Royal College of Paediatrics and Child Health. London: RCR;


Parisi M.T., Wiester R.T., Done S.L., et al. Three-dimensional computed tomography skull reconstructions as an aid to child abuse evaluations. Pediatric Emergency Care 2015; 31(11): 779-786.


Bixby S.D., Abo A., Kleinman P.K. High-impact trauma causing multiple posteromedial rib fractures in a child. Pediatr Emerg Care 2011; 27(3): 218-219.


Sanchez T.R., Grasparil A.D., Chaudhari R., et al. Characteristics of Rib Fractures in Child Abuse-The Role of Low-Dose Chest Computed Tomography. Pediatric Emergency Care. 2016; 12.


Systematic Reviews: CRD’s Guidance for Undertaking Reviews in Health Care. University of York; 2009.


Polgar A., Thomas S. Critical Evaluation of Published Research. In: 3rd edition ed. Introduction to Research in the Health Sciences. Melbourne: Churchill Livingstone; 1995. p.


A schema for evaluating evidence on public health interventions (version 4). Melbourne: National Public Health Partnership; 2002.


Health Evidence Bulletins Wales: A systematic approach to identifying the evidence. Cardiff: Information Services UWCM; 2004