read and response to question in details

D ow

nl oa

de d

fr om

w w

w .a

jr on

li ne

.o rg

b y

T ho

m as

J ef

fe rs

on U

ni v

on 0

9/ 23

/1 4

fr om

I P

a dd

re ss

1 47

.1 40

.2 33

.1 6.

C op

yr ig

ht A

R R

S . F

or p

er so

na l

us e

on ly

; al

l ri

gh ts

r es

er ve

d

Pediatric Imaging Prosser et al. Radiologic Dating of Pediatric Fractures

1282

0361–803X/05/1844–1282

© American Roentgen Ray Society

Review

AJR 2005;184:1282–1286

Ingrid Prosser1 Sabine Maguire1 Sara K. Harrison2 Mala Mann3 Jonathan R. Sibert1 Alison M. Kemp1 Welsh Child Protection Systematic Review Group Prosser I, Maguire S, Harrison SK, Mann M, Sibert

JR, Kemp

AM

Received August 5, 2004; accepted after revision September 9, 2004.

Supported by the National Society for the Prevention of Cruelty to Children of the United Kingdom.

1Department of Child Health, Cardiff University, Wales College of Medicine, Academic Centre, Llandough Hospital, Penarth CF64 2XX, Wales, United Kingdom. Address correspondence to A. M. Kemp. 2Department of Radiology, Cardiff University, Wales College of Medicine, Heath Hospital, Heath Park, Cardiff CF14 4XN, Wales, United Kingdom. 3Duthie Library, Cardiff University, Wales College of Medicine, Heath Hospital, Heath Park, Cardiff CF14 4XN, Wales, United Kingdom.

How Old Is This Fracture? Radiologic Dating of Fractures in Children: A Systematic Review

OBJECTIVE. We conducted a systematic review of the literature to define the evidence for radiologic dating of fractures in children in the context of child protection.

CONCLUSION. Radiologic dating of fractures is an inexact science. Most radiologists date fractures on the basis of their personal clinical experience, and the literature provides little consistent data to act as a resource. There is an urgent need for research to validate the criteria used in the radiologic dating of fractures in children younger than 5 years.

ractures occur in up to 52% of child abuse cases [1, 2]. In con- trast to accidental fractures, most abusive fractures occur in chil-

dren younger than 3 years; 80% of such frac- tures occur in children younger than 18 months [3]. Abusive fractures may be multi- ple and of different ages [4, 5], a point that can only be determined from their dating. Dating fractures may also highlight inconsis- tencies between the timing of an injury and the history given, thus aiding in the diagnosis of child abuse [6].

Police and lawyers are particularly interested in the timing of injuries in child abuse to identify or exclude potential perpetrators. In the court setting, radiologists are frequently asked to date fractures to narrow down the time of injury. We have conducted what we believe to be the first systematic review of the literature to define the evidence for radiologic dating of fractures in children in the context of child protection.

Materials and Methods We performed an all-language literature search

of original articles published from 1966 through March 2004 as shown in Figure 1. We searched the Applied Social Science Index and Abstracts (AS- SIA) [7], CareData [8], MEDLINE [9], Child Data

[10], Cumulative Index to Nursing and Allied Health Literature (CINAHL) [11], EMBASE [12], PsychINFO [13], System for Information on Grey Literature in Europe (SIGLE) [14], Social Science Citation Index [15], and Turning Research into Practice (TRIP) [16] databases. In addition, we per- formed an appropriate hand-search of literature pub- lished from 1947 to 23rd February 2004. Key words used in our search are listed in Appendix 1. Each ar- ticle underwent two independent reviews by mem- bers of a group of 27 specialist reviewers including pediatricians, pediatric radiologists, and orthopedic surgeons, among other child health professionals with expertise in child protection. A third review was performed if there was disagreement among the initial reviewers. We included primary research ad- dressing the question of radiologically dating frac- tures in children younger than 17 years. Studies were excluded if they were review articles, consen- sus statements, or expert opinions; if details on chil- dren could not be extracted from mixed-age data; if the criteria for dating were not detailed; or if under- lying bone disease was present.

All included studies were analyzed using stan- dardized data extraction and critical appraisal forms [17]. Studies were graded for quality on the basis of study design, accurate documentation of the time of injury, and standardized criteria for ra- diologic dating.

F

AJR:184, April 2005

Radiologic Dating of Pediatric Fractures

D ow

nl oa

de d

fr om

w w

w .a

jr on

li ne

.o rg

b y

T ho

m as

J ef

fe rs

on U

ni v

on 0

9/ 23

/1 4

fr om

I P

a dd

re ss

1 47

.1 40

.2 33

.1 6.

C op

yr ig

ht A

R R

S . F

or p

er so

na l

us e

on ly

; al

l ri

gh ts

r es

er ve

d

Results Figure 1 summarizes the total number of

studies identified and reviewed. Three studies met the criteria for inclusion [18–20], reflect- ing data on 189 children, 56 of whom were younger than 5 years.

Two studies defined staging criteria (Table 1). Islam et al. [19] examined 707 radiographs of forearm fractures in 141 children randomly selected over a 4-year period; only 23 were younger than 5 years. All fractures were im- mobilized with casts. Fractures treated by surgical fixation were excluded. Patients un- derwent radiography at various times ranging from 0 to 100 days after injury. A pediatric ra- diologist who was unaware of the time inter- val after trauma assessed all radiographs. The study defined clear staging criteria that were based on data from the radiology and histol- ogy literature (Table 2).

Using their dating criteria, Islam found that periosteal reaction was not observed on any ra- diograph obtained before 2 weeks after the in- jury. However, only 22 patients (most with casts) underwent radiography between 7 and 14 days after the injury. The earliest radiographs

appear to have been obtained 7 days after the in- jury. Periosteal reaction was evident in all 33 pa- tients imaged 4 weeks after injury. Density increased at fracture margins at 2 weeks, with a peak at 4 to 6 weeks in 85% (128/150) of the fractures. No increase in fracture margin sclero- sis was seen after 11 weeks. Calcified callus (calcified periosteal reaction) was seen as early as 2 weeks after injury in 15% (18/117) of the fractures and at all fracture sites by 4 weeks. Af- ter 10 weeks, 90% (26/29) of the calluses had a density equal to or greater than that of the cor- tex. At 8 weeks, 50% of the fractures showed evidence of bridging. The earliest remodeling was seen at 4 weeks and was noted in 95% (91/ 96) from 8 weeks onward.

Yeo and Reed [20] also defined criteria with which to date fractures radiologically, looking only at callus formation. Patients with solitary closed nonpathologic fractures of the femoral shaft were included. All were treated by traction followed by the application of a hip spica cast. Radiographs were obtained as clinically indi- cated at varying time intervals (Table 1). Three stages of callus formation were defined (Table 2): stage 1, the earliest radiographically visible

calcification of callus; stage 2, callus com- pletely bridging the fracture; and stage 3, smooth, homogeneous mature callus in which the fracture line is still visible.

The third included study, conducted in 1979, assessed 23 newborns with fractured clavicles, humeri, and femurs sustained at birth. These were assessed solely for first ap- pearance of calcification at fracture site. The earliest appearance was 7 days after birth; peak calcification was seen 9–10 days after birth; and the latest appearance was 11 days after birth. The numbers included were again very small and differed for each fracture. No details were offered as to how many radio- graphs were acquired per child and at what time intervals.

Discussion Despite didactic statements in textbooks as

to the dating of fractures in children, there is a disappointing lack of primary evidence on which to base dating [21]. Given the high prevalence of abusive fractures in infants and toddlers, and to a lesser extent in preschool children [1, 2, 5, 22–24], it is particularly worrying that the two larger studies only in- cluded 33 children in this age group. Other limitations of the included studies are the variation of intervals between radiographs (especially at the early stages of healing) and the different numbers of radiographs per frac- ture (Table 1). The presence of casts, un- avoidably, impaired the detection of subtle radiographic signs. In addition, Yeo and Reed [20] and Islam et al. [19] chose different bones to study, femur and forearm, respectively, which may have different healing rates, but published evidence is lacking in this area.

Radiologists usually determine the age of fractures based on clinical experience and guidance offered in textbooks [21]. Unfortu- nately the terms describing the phases of heal- ing differ between the two included studies that offer criteria [19, 20], and these differ from the terminology in Kleinman’s textbook [21] (Table 3). The table in this often-quoted source is derived from the personal clinical experience of the authors and has not been further validated by any primary research (J. F. O’Connor, personal communication, June 2004). It is impossible to assess whether the three sets of criteria are in agreement as to the peak times at which phases of healing occur. A radiologist who regularly reports trauma radio- graphs, with a documented history for time of injury, can develop expertise in this area over time. However, because the criteria are not

MEDLINE CareData EMBASE SIGLE Social Sciences Citation CINAHL ASSIA ISI Proceedings Child Data TRIP database

1966–2004 1970–2004 1980–2004 1980–2004 1981–2004 1982–2004 1987–2004 1990–2004 1996–2004 1997–2004

Hand-search of text books Hand-search of all articles identified from other sources

Scanned total 1,556 titles and abstracts for duplicates and relevancy

Third Review 146

399 reviewed

Included in analysis 3

Translated 22

Fig. 1.—Chart displays our search strategy for articles on radiologic dating of fractures in children.

AJR:184, April 2005 1283

Prosser et al.

D ow

nl oa

de d

fr om

w w

w .a

jr on

li ne

.o rg

b y

T ho

m as

J ef

fe rs

on U

ni v

on 0

9/ 23

/1 4

fr om

I P

a dd

re ss

1 47

.1 40

.2 33

.1 6.

C op

yr ig

ht A

R R

S . F

or p

er so

na l

us e

on ly

; al

l ri

gh ts

r es

er ve

d

standardized or reproducible, less experienced radiologists have little primary evidence on which to base their practice.

Despite the conflicting conclusions of the in- cluded studies, there is agreement that hard cal- lus and early remodeling are seen at 8 weeks in most cases. Early callus was first noted 7 days

after injury and was present in 50% by 4 weeks. The variable interval between radiographs in the studies leaves gaps at the most crucial early stages of healing, and time frames may there- fore be inaccurate. There is universal agreement that the radiologic features noted are a contin- uum, with considerable overlap. Larger-scale

studies are needed to assess standardized criteria for dating fractures in children younger than 5 years.

The fractures in these studies were all im- mobilized, which limits its application to dat- ing fractures in child abuse. Many abusive fractures are occult [25, 26], and late presen- tation allows continued movement, further in- jury and repetitive fracture, further complicating the dating process. It is frequently stated that fractures heal faster in young chil- dren and especially in infants, but as yet, there is no published radiologic evidence to support this statement. It has been noted in adults that healing may be faster with coexistent severe head injury. Perkins and Skirving [27] found that the average femoral fracture healing time was 12.4 weeks in those with a head injury versus 15.7 weeks in control subjects (p < 0.00005). A study by Spencer [28] that in- cluded an age range from 4 to 67 years found almost identical changes: 12.4 weeks in the group with a head injury versus 15.2 weeks in the control subjects. Unfortunately, the data for the children were not separated from the data for adults, making it impossible to ana- lyze it for this review. This finding may be relevant in the context of nonaccidental head injury in which fractures coexist in as much as 50% of the cases [29].

Pergolizzi and Oestreich [30] highlighted the importance of familiarity with normal physiologic periosteal reaction in infants younger than 6 months. These infants may show symmetric diaphyseal periosteal reac- tion, although it may be more prominent on one side [31]. This should not be misinter- preted as a healing fracture.

In 1996, Kleinman et al. [32] mentioned that performing a repeat skeletal survey 2 weeks after the initial survey aided in the dat- ing of fractures in 18% (13/70) of children younger than 3 years. No details were given as to what specific features were used for dat-

TABLE 1 Key Features of Included Studies

Author (year) Study Type Total No. of

Children (no. < 5 yr)

Mean No. of Radiographs

per Child (range)

Age Range Association of Healing With Age/Sex Number and Site

of Fractures

Islam 2000 [19] Longitudinal 141 (23) 3.7 (2–8) 1–17 years (mean, 8 yr) No association (chi square) 131 fractured radii, 74 fractured ulnae

Yeo 1994 [20] Longitudinal 25 (10) 9 (6–17) Birth–14 yr No association (multiple regression analysis and Student's t test)

25 fractured femora

Cumming 1979 [18]

Longitudinal 23 (23) 1 Birth–11 days N/A 10 clavicles, 6 humeri, 7 femora

Note.—N/A = not applicable.

TABLE 2 Radiologic Features of Healing in Three Studies Included in Analysis

Radiologic Feature Islam 2000 [19]

Peak (range)

Yeo 1994 [20] Peak

(range)

Cumming 1979 [18] Peak

(range)

Fracture gap widening 4–6 wk, 56% (2–8)

Periosteal reaction (stage 1)

4–7 wk, 100% (2 wk onward)

1.6 wk (1–3 wk)

9–10 days (7–11 days)

Marginal sclerosis 4–6 wk, 85% (2–11)

1st callus 4–7 wk, 100% (2 wk onward)

Callus density > cortex density 13 wk, 90% ( 4 wk onward)

Bridging (stage 2)

13 wk, 50% (3 wk onward 10)

2.6 wk (1.5–3.7 wk)

Periosteal incorporation 14 wk (7 wk onward)

Remodeling (stage 3)

9 wk (4 wk onward)

8 wk (5–11 wk)

TABLE 3 Timetable of Radiologic Changes in Children’s Fractures

Category Early Peak Late

Resolution of soft tissues 2–5 days 4–10 days 10–21 days

SPNBF 4–10 days 10–14 days 14–21 days

Loss of fracture line definition 10–14 days 14–21 days

Soft callus 10–14 days 14–21 days

Hard callus 14–21 days 21–42 days 42–90 days

Remodeling 3 mo 1 yr 2 yr to physeal closure

Note.—Adapted from [21, 35] with permission. Repetitive injuries may prolong categories 1, 2, 5, and 6. SPNBF = subperiosteal new bone formation.

1284 AJR:184, April 2005

Radiologic Dating of Pediatric Fractures

D ow

nl oa

de d

fr om

w w

w .a

jr on

li ne

.o rg

b y

T ho

m as

J ef

fe rs

on U

ni v

on 0

9/ 23

/1 4

fr om

I P

a dd

re ss

1 47

.1 40

.2 33

.1 6.

C op

yr ig

ht A

R R

S . F

or p

er so

na l

us e

on ly

; al

l ri

gh ts

r es

er ve

d

ing in this study. Bone scans have no place in fracture dating because they show positive re- sults within 7 hr of injury [33] and can continue to show positive results for as long as 1 year.

Digital imaging is rapidly replacing stan- dard techniques in many centers. Although Kleinman et al. [34] found these digital tech- niques to be comparable to conventional imaging for identifying abusive fractures postmortem in the United States, no assess- ment of digital radiologic fracture dating has been performed. The direct digital radiogra- phy system used in the study by Kleinman et al. differs from the computed digital radiogra- phy system more widely used in the United Kingdom. Studies are urgently required to validate dating using both systems if this is to become standard practice.

In conclusion, our analysis showed that the evidence base for current methods of radio- logic dating is sparse. Dating of fractures in children is an inexact science. The radiologic features of bone healing are a continuum, with considerable overlap. Radiologic esti- mates of the time of injury are made in terms of weeks rather than days. It is vital for all in- vestigating agencies to be aware of these broad time frames. However, radiologists can clearly differentiate recent from old fractures. Such differentiation remains valuable in iden- tifying a child who has been subjected to re- peated abuse or whose injuries are thus shown to be inconsistent with the history offered.

Our findings have the following four im- plications for practice: the dating of fractures in children is an inexact science; clinicians must bear this fact in mind when offering time frames of injuries to investigating agencies or courts; periosteal reaction is seen as early as 4 days and is present in at least 50% of the cases by 2 weeks after the injury; and remod- eling peaks 8 weeks after injury.

Acknowledgments We thank our panel of expert reviewers,

the Welsh Child Protection Systematic Re- view Group: M. Barber, P. Barnes, M. Bhal, J. Bowen, R. Brooks, A. Butler, S. Datta, R. Frost, C. Graham, M. James-El- lison, N. John, A. Maddocks, S. Morris, A. Mott, A. Naughton, C. Norton, H. Payne,

L. Price, B. Ranton, P. Thomas, E. Webb, and C. Woolley.

References 1. Kogutt M, Swischuk L, Fagan C. Patterns of in-

jury and significance of uncommon fractures in the battered child syndrome. Am J Roentgenol Ra- dium Ther Nucl Med 1974;121:143–149

2. Loder R, Bookout C. Fracture patterns in battered children. J Orthop Trauma 1991;5:428–433

3. Worlock P, Stower M, Barbor P. Patterns of frac- tures in accidental and non-accidental injury in chil- dren: a comparative study. BMJ 1986;293:100–102

4. Duhaime A, Alario A, Lewander W, et al. Head injury in very young children: mechanisms, injury types, and ophthalmologic findings in 100 hospi- talized patients younger than 2 years of age. Pedi- atrics 1992;90:179–185

5. Leventhal J, Thomas S, Rosenfield N, Markowitz R. Fractures in young children. Distinguishing child abuse from unintentional injuries. Am J Dis Child 1993;147:87–92

6. Kleinman P, Blackbourne B, Marks S, Karellas A, Belanger P. Radiologic contributions to the inves- tigation and prosecution of cases of fatal infant abuse. N Engl J Med 1989;320:507–511

7. Applied Social Science Index and Abstracts (AS- SIA) [database online]. East Grinstead, West Sus- sex, England: Cambridge Scientific Abstracts. Updated February 23, 2004

8. CareData [database online]. London, England: Social Care Institute for Excellence. Updated February 23, 2004

9. MEDLINE [database online]. Bethesda, MD: Na- tional Library of Medicine, U.S. National Insti- tutes of Health. Updated February 23, 2004

10. National Children’s Bureau Database [database online]. Updated February 23, 2004

11. Cumulative Index to Nursing and Allied Health Lit- erature (CINAHL) [database online]. San Francisco, CA: Galen Digital Library of the University of Cal- ifornia–San Francisco. Updated February 23, 2004

12. EMBASE [database online]. Philadelphia, PA: Elsevier. Updated February 23, 2004

13. PsychINFO [database online]. Washington, DC: American Psychological Association. Updated February 23, 2004

14. System for Information on Grey Literature in Eu- rope (SIGLE) [database online]. The Hague. The Netherlands: European Association for Grey Lit- erature. Updated February 23, 2004

15. Social Science Citation Index [database online]. Philadelphia, PA: Thomson Scientific. Updated February 23, 2004

16. Turning Research into Practice (TRIP) Database Plus [database online]. London, England: TRIP Database Ltd. Updated February 23, 2004

17. National Health Service Centre for Reviews and Dis- semination (CRD). Undertaking systematic reviews of research on effectiveness: CRD’s guidance for those carrying out or commissioning reviews, 2nd ed. York, England: University of York, 2001. CRD report 4

18. Cumming W. Neonatal skeletal fractures: birth trauma or child abuse? J Can Assoc Radiol 1979;30:30–33

19. Islam O, Soboleski D, Symons S, Davidson L, Ashworth M, Babyn P. Development and dura- tion of radiographic signs of bone healing in chil- dren. AJR 2000;175:75–78

20. Yeo L, Reed M. Staging of healing of femoral frac- tures in children. Can Assoc Radiol J 1994; 45:16–19

21. Kleinman PK, ed. Diagnostic imaging of child abuse, 2nd ed. St Louis, MO: Mosby, 1998

22. Merten D, Kirks D, Ruderman R. Occult humeral epiphyseal fracture in battered infants. Pediatr Radiol 1981;10:151–154

23. Merten D, Radlowski M, Leonidas J. The abused child: a radiological reappraisal. Radiology 1983;146:377–381

24. McMahon P, Grossman W, Gaffney M, Stanitski C. Soft-tissue injury as an indication of child abuse. J Bone Joint Surg Am 1995;77:1179–1183

25. Smith F, Gilday D, Ash J, Green M. Unsuspected costo-vertebral fractures demonstrated by bone scanning in the child abuse syndrome. Pediatr Ra- diol 1980;10:103–106

26. Sty J, Starshak R. The role of bone scintigraphy in the evaluation of the suspected abused child. Ra- diology 1983;146:369–375

27. Perkins R, Skirving A. Callus formation and the rate of healing of femoral fractures in patients with head injuries. J Bone Joint Surg Br 1987; 69:521–524

28. Spencer R. The effect of head injury on fracture healing: a quantitative assessment. J Bone Joint Surg Br 1987;69:525–528

29. Kemp A, Stoodley N, Cobley C, Coles L, Kemp K. Apnoea and brain swelling in non-accidental head injury. Arch Dis Child 2003;88:472–476

30. Pergolizzi RJ, Oestreich A. Child abuse fracture through physiologic periosteal reaction. Pediatr Radiol 1995;25:566–567

31. Shopfner C. Periosteal bone growth in normal in- fants: a preliminary report. AJR 1966;97:154–163

32. Kleinman PK, Nimkin K, Spevak MR, et al. Fol- low-up skeletal surveys in suspected child abuse. AJR 1996;167:893–896

33. Rosenthall L, Hill R, Chuang S. Observation on the use of 99mTc-phosphate imaging in periph- eral bone trauma. Radiology 1976;119:637–641

34. Kleinman PK, O’Connor B, Nimkin K, et al. De- tection of rib fractures in an abused infant using digital radiography: a laboratory study. Pediatr Radiol 2002;32:896–901

35. O’Connor J, Cohen J. Dating fractures. In: Klein- man PK, ed. Diagnostic imaging of child abuse. St. Louis, MO: Mosby, 1998:168–177

AJR:184, April 2005 1285

Appendix 1 appears on next page

Prosser et al.

D ow

nl oa

de d

fr om

w w

w .a

jr on

li ne

.o rg

b y

T ho

m as

J ef

fe rs

on U

ni v

on 0

9/ 23

/1 4

fr om

I P

a dd

re ss

1 47

.1 40

.2 33

.1 6.

C op

yr ig

ht A

R R

S . F

or p

er so

na l

us e

on ly

; al

l ri

gh ts

r es

er ve

d

1. child abuse.mp. 2. child protection.mp. 3. (battered child or shaken baby or battered baby).mp. 4. 1 or 2 or 3 5. child:.mp. 6. non-accidental injur.mp. 7. non-accidental trauma.mp. 8. (non-accidental: and injur:).mp. 9. soft tissue injur:.mp.

10. physical abuse.mp. 11. (or/6-10) and 5 12. 4 or 11 13. fractur:.mp. 14. rib fractur:.mp. 15. skull fractur:.mp. 16. femoral fractur:.mp. 17. humeral fractur:.mp. 18. pelvic fractur:.mp. 19. spiral fractur:.mp. 20. metaphyseal fractur:.mp.

21. (corner fractur: or bucket handle fractur:).mp. 22. metaphyseal chip fractur:.mp. 23. classic metaphyseal lesion:.mp. 24. or/13-23 25. (investigat: adj3 fract:).mp. 26. (radiolog: adj3 fractur:).mp. 27. (roentgen: adj3 fract:).mp. 28. skeletal survey.mp. 29. bone scan:.mp. 30. Isotope Bone Scan:.mp. 31. Radionuclide.mp. 32. Scintigraphy.mp. 33. ((paediatric or pediatric) adj3 radiolog:).mp. 34. ((paediatric or pediatric) adj3 nuclear medicine).mp. 35. (ag: adj3 fractur:).mp. 36. ((dating or date) adj3 fractur:).mp. 37. (pattern: adj3 fractur:).mp. 38. (heal: adj3 fractur:).mp. 39. (timing adj3 healing).mp.

APPENDIX 1. Keywords and Phrases Used for the Fracture Dating Review

1286 AJR:184, April 2005