Acute Elbow Trauma in Children: Spectrum of Injury Revealed by MR Imaging Not Apparent on Radiographs

James F. Griffith 1 Derek J. Roebuck 1,2 Jack C. Y. Cheng 3 Yu Leung Chan 1 Timothy H. Rainer 4 Bobby K. W. Ng 3 Constantine Metreweli 1 Received December 14, 1999; accepted after revision June 8, 2000. 1 Department of Diagnostic Radiology and Organ Imaging, Chinese University of Hong Kong, Sha Tin, Hong Kong. Address correspondence to J. F. Griffith. 2 Present address: Department of Radiology, Great Ormond Street Hospital for Children, London WC1N 3JH, United Kingdom. 3 Department of Orthopaedics and Traumatology, Chinese University of Hong Kong, Sha Tin, Hong Kong. 4 Department of Accident and Emergency Medicine, Chinese University of Hong Kong, Sha Tin, Hong Kong. AJR 2001;176:53 60 0361 803X/01/1761 53 American Roentgen Ray Society Acute Elbow Trauma in Children: Spectrum of Injury Revealed by MR Imaging Not Apparent on Radiographs OBJECTIVE. The objective of this study is to evaluate the frequency and significance of unrecognized bone or soft-tissue injury in pediatric patients with elbow trauma assessed with radiographs alone. SUBJECTS AND METHODS. Fifty children (32 boys and 18 girls; mean age, 7.3 years; age range, 2 12 years) with acute elbow trauma were examined with radiography and MR imaging. Radiographs were categorized into those showing normal findings, an effusion, an equivocal fracture, or an unequivocal fracture. MR examinations were assessed for an effusion, fracture, transphyseal fracture extension, physeal injury, bone bruising, and ligament or muscle injury. Average clinical follow-up was 1.6 years (range, 6 28 months) after injury. RESULTS. Radiographs showed normal findings in seven children (14%), an effusion only in 17 children (34%), and an unequivocal or equivocal fracture in 26 children (52%). MR imaging showed an effusion in 48 children (96%); unequivocal fracture in 37 children (74%), including transphyseal fracture in seven children (14%) and other physeal injury in three children (6%); bone bruising in 45 children (90%); ligament injury in six children (14%); and muscle injury in 19 children (38%). A less severe spectrum of injury occurred in children with normal findings on radiographs than in those with an effusion or fracture seen on radiography. Follow-up radiographs did not help in the detection of radiographically occult fractures. MR findings had no appreciable effect on patient treatment and no value in predicting duration of convalescence or clinical outcome at an average of 1.6 years after injury. CONCLUSION. In children with elbow trauma, MR imaging reveals a broad spectrum of bone and soft-tissue injury beyond that recognizable radiographically. However, the additional information afforded by MR imaging has little bearing on treatment or clinical outcome. I n children, elbow trauma may lead to bony, cartilaginous, or soft-tissue injury. Radiographs, the principal means of imaging elbow trauma, reveal joint effusions or fractures and allow recognition of injuries that require surgical intervention. Radiographs do not show bone bruising, cartilaginous, or soft-tissue injury and may underestimate physeal injury. In their study of eight children with radiographically visible fractures around the elbow, Beltran et al. [1] reported that two of these children, both of whom had humeral condylar fractures, had unsuspected transphyseal fracture extension through unossified epiphyseal cartilage shown by MR imaging. Carey et al. [2] reported that in 14 suspected physeal injuries (of the elbow, wrist, knee, and ankle), MR imaging changed the radiographic diagnosis in 50% of the cases by showing either radiographically occult fractures (five cases) or unsuspected transphyseal extension (two cases) [2]. MR findings resulted in a change of treatment in 36% of the cases. In contrast, in a study of 29 children with acute fractures of the distal tibia, Petit et al. [3] showed that the radiographic diagnosis was changed as a result of MR findings in only one case and that treatment was not changed in any of the cases [3]. The extent and significance of unrecognized injury when pediatric patients with elbow trauma are assessed with radiographs alone are not known. We examined 50 children with elbow trauma that did not require immediate surgery but resulted in an orthopedic referral for radiography and MR imaging to determine the frequency and significance of radiographically occult bone, physeal, and soft-tissue injury. Subjects and Methods From September 1997 to July 1999, 50 children (32 boys and 18 girls; mean age, 7.3 years; age range, 2 12 years) with elbow trauma, who had been referred to the AJR:176, January 2001 53

Griffith et al. Prince of Wales Hospital Orthopedic Fracture Clinic, were examined. Children with a full range of movement and no radiographic abnormality that is, those who had been discharged from the emergency department without referral to the orthopedic clinic were not included. Also not included were children with displaced fractures that required early surgery (closed reduction or internal fixation). The study cohort, therefore, comprised children with undisplaced fractures, children with a radiographic effusion in the absence of a visible fracture, and children with limited elbow movement and normal findings on radiographs. The time from injury to presentation was less than 24 hr in all cases. Injuries resulted from a fall (n = 44), bicycle accident (n = 3), games (n = 2), or sports (n = 1). Conventional anteroposterior and lateral radiographs were obtained on a standard imaging system in all children at presentation. The interval between presentation and MR imaging ranged from less than 1 day to 16 days, with a median of 6 days. Radiographic Criteria Two radiologists interpreted the elbow radiographs by consensus, categorizing them as showing normal findings (Fig. 1A), a joint effusion in the absence of a fracture (Fig. 2A), one or more A A equivocal fractures, or one or more unequivocal fractures (Fig. 3A). An effusion was diagnosed on the basis of displacement of the anterior fat pad with or without visualization of the posterior fat pad (Fig. 2A). An unequivocal fracture was diagnosed on the basis of an unequivocal disruption, often minimal, of the bone cortex and adjacent trabecula (Fig. 3A). If doubt existed as to the presence of a fracture, an equivocal fracture was diagnosed. The location of any equivocal or unequivocal fractures was recorded. All children were treated with immobilization in a right-angled elbow cast or back slab regardless of radiographic findings. B B Fig. 1. 10-year-old boy with acute elbow trauma sustained during fall. A, Lateral radiograph of elbow shows undisplaced anterior fat pad (arrow) and no posterior fat pad. No fracture is visible. B, Sagittal T2-weighted short tau inversion-recovery MR image of elbow shows undisplaced fracture of proximal radial metaphysis (arrow). Bone bruising of proximal radius adjacent to fracture can be seen. Note absence of significant joint effusion. Fig. 2. 6-year-old boy with acute elbow trauma sustained during fall. A, Radiograph of elbow shows joint effusion, but it does not reveal fracture. B, Sagittal T2-weighted short tau inversion-recovery MR image of elbow shows widening of physis as hyperintense band (arrow) extending between metaphysis and hypointense unossified cartilage. 54 AJR:176, January 2001

MR Imaging of Acute Elbow Trauma in Children MR Imaging Technique MR imaging examinations of the immobilized elbow were performed on a 1.5-T machine (Gyroscan; Philips, Best, The Netherlands) with a standard surface coil. Pulse MR sequences comprised a T1-weighted spin-echo sequence (TR/TE, 500/18), a T2-weighted short tau inversion-recovery (STIR) sequence (1728/70; inversion time, 150 msec) in both sagittal and coronal planes (relative to the long axis of the humerus), and a gradient-recalled echo sequence (500/15; flip angle, 30 ) in the coronal plane. Because the elbow was examined at right angles, coronal imaging of the arm equated to axial imaging of the forearm. Sedation was needed for two children (ages, 2 and 5 years). One child, originally scheduled for MR imaging, was excluded from the study because of unsuccessful attempts at oral sedation. MR Imaging Criteria A Fig. 3. 10-year-old boy with acute elbow trauma sustained during fall. A, Anteroposterior radiograph of elbow shows fracture (arrow) of lateral humeral condyle. Note normal appearance of capitulum. B, Gradient-recalled echo coronal MR image of humerus shows fracture (arrow) of lateral condyle. C, Gradient-recalled echo coronal MR image obtained more anterior to B shows fracture (arrow) extending through capitulum and unossified cartilage. Radiographs were available for review at the time of MR imaging reporting. MR studies were interpreted by two radiologists who noted the presence of bone fracture, transphyseal fracture extension, physeal injury, bone bruising, annular ligament injury, collateral ligament injury, and muscle injury. In each case, a final diagnosis was reached by consensus. An effusion was diagnosed when fluid within the elbow joint displaced the anterior fat pad or posterior fat pad. A fracture was diagnosed when there was unequivocal disruption of cortical bone in continuity with a medullary hypointense line on T1-weighted imaging or a hyperintense line on T2-weighted imaging (Fig. 2B). If doubt existed as to the presence of a fracture on MR imaging, no fracture was diagnosed. Transphyseal fracture was diagnosed when a hypointense line on T1-weighted imaging or a hyperintense line on T2-weighted imaging was seen traversing the physis and epiphyseal cartilage (Fig. 3). Any extension to the articular margin or separation of the articular fragments was noted. Physeal injury was diagnosed on the basis of visualization of a linear hyperintensity on T2-weighted imaging extending along the physis (Fig. 2B). Bone bruising (trabecular microfracture) [4] was diagnosed when diffuse hyperintensity of medullary signal on T2- weighted imaging and hypointensity of medullary signal on T1-weighted imaging in the absence of a fracture at the same location were seen (Fig. 4B). Annular ligament injury was diagnosed on the basis of diffuse hyperintensity of the annular ligament on T2-weighted imaging (Fig. 5). In this respect, differentiation between a distended fluid-filled inferior recess of the elbow joint and an annular ligament injury was made by assessing images obtained in both sagittal and axial planes. Collateral ligament injury was diagnosed when discontinuity, disruption, or hyperintensity of either the medial or lateral collateral ligament was present (Fig. 6). Muscle injury was diagnosed on the basis of diffuse hyperintensity present within muscle on T2-weighted imaging (Fig. 7); the injured muscle group was noted. For each study, the MR imaging sequence that provided the most diagnostic information was recorded. Followup radiographs, when available, in children with fractures shown on MR imaging but not visible radiographically were reviewed. B Early clinical follow-up. After injury, children were examined in the outpatient clinic approximately every 2 weeks. At each visit, the cast was removed and the elbow joint was examined. If passive elbow movement was painful or if there was marked local tenderness, the cast was reapplied. Late clinical follow-up. Seven months after the completion of the study, all case notes were reviewed and all children were recalled for clinical reassessment. Any change of treatment on the basis of additional MR findings was noted as was the duration of immobilization, the duration of convalescence, and the level of functional recovery. For both the affected elbow and the unaffected elbow, the carrying angle and the range of flexion extension were measured with a goniometer, and pronation supination was assessed subjectively. Varus or valgus deformity was defined as an increase in side-to-side variation in carrying angle of greater than 10. Limited movement was defined as a limitation of either flexion and extension or pronation and supination of greater than 5 when compared with the contralateral unaffected elbow. Analysis of results. First, we reviewed the MR imaging findings in children with normal findings on radiographs, a radiographic effusion, or a radiographic fracture. Second, we studied the frequency of specific injuries on radiographs and MR images (i.e., fracture, transphyseal extension, physeal injury, bone bruising, ligament injury, and muscle injury). Finally, we studied the effect and the relationship of the additional MR imaging findings on treatment, convalescence, and longterm clinical outcome. C AJR:176, January 2001 55

Griffith et al. Fig. 4. 10-year-old boy with acute elbow trauma sustained during fall from bicycle. A, Lateral radiograph of elbow shows large joint effusion. No fracture is visible. B, Sagittal T2-weighted short tau inversion-recovery (STIR) MR image of elbow shows large hemarthrosis with muscle edema of triceps (white arrow), bone bruising of ulna (black arrow), and subcutaneous edema. C, Axial T2-weighted STIR MR image of proximal forearm shows fracture (arrow) of radial head. Statistical analysis. The Bonferroni test for multiple comparisons between means of different subgroups was used. A p value of less than 0.05 was regarded as statistically significant. Results MR Findings in Similar Radiographic Subgroups A B At presentation, seven (14%) of 50 children examined had normal findings on radiographs, 17 (34%) had an effusion only, and the remaining 26 (52%) had an unequivocal or equivocal fracture. None of the children had normal findings on the MR imaging examination. The spectrum of injuries shown by MR imaging in the different radiographic subgroups is shown in Table 1. Among radiographically similar subgroups, a broad spectrum of injuries was found. The spectrum of radiographically occult injuries was most severe in patients with a radiographic fracture, less severe in those with an effusion, and least severe in those with normal findings on the radiographs. Analysis of Specific Injuries Fractures. Radiographs showed a total of 27 fractures (21 unequivocal, 6 equivocal) in 26 children (52%) (Table 2), whereas MR imaging showed a total of 38 fractures (all unequivocal) in 37 children (74%) (Fig. 2A). Eleven children (22%) had fractures shown by MR imaging, despite having radiographs with normal findings (n = 2) (Fig. 1) or radiographs showing an effusion only (n = 9) (Fig. 4). These radiographs were reviewed after identification of fractures on MR imaging, and no fractures were visible in retrospect. The location of these radiographically occult fractures is shown in Table 2. Four of the six radiographically equivocal fractures had unequivocal fractures in a corresponding location when examined on MR imaging. The remaining two radiographically equivocal fractures had no fracture in a corresponding location on MR imaging. Transphyseal fracture extension. Transphyseal fracture extension to the epiphyseal cartilage was seen on MR imaging in seven (14%) of the 50 children examined or, alternatively, in seven (18%) of the 38 fractures shown by MR imaging (Fig. 3). Transphyseal fracture extension was not suspected radiographically in any case. All C 56 AJR:176, January 2001

MR Imaging of Acute Elbow Trauma in Children TABLE 1 TABLE 2 Summary of Bone and Soft-Tissue Injury Revealed by MR Imaging in 50 Children with Radiographs Showing Normal Findings, Radiographic Effusion, or Equivocal or Unequivocal Fracture Type of Injury Revealed by MR Imaging Type of Fracture Revealed by MR Imaging Normal (n = 7) Radiographic Findings in 37 Children with 38 Unequivocal Fractures on MR Imaging Fracture Correctly Identified Effusion Only (n = 17) Fracture Misclassified or Equivocal Effusion Only Fracture (n = 26) Normal Total Transphyseal fracture 0 3 4 7 Fracture 2 4 22 28 Growth plate injury 0 2 1 3 Bone bruising 5 17 23 45 Ligament injury 1 2 3 6 Muscle injury 1 6 12 19 Supracondylar 10 1 1 0 12 Lateral condylar 4 3 4 0 11 Medial condylar 2 0 1 1 4 Radial head or neck 0 1 1 1 3 Olecranon 3 1 1 0 5 Coronoid process 1 0 0 0 1 Proximal ulna 1 0 1 0 2 Total 21 6 9 2 38 Total transphyseal fractures occurred in continuity with an adjacent metaphyseal fracture. In no case was an isolated fracture of the cartilage seen in the absence of a metaphyseal fracture. Identification of transphyseal fracture extension lead to reclassification of these fractures as type IV injuries with the Salter- Harris classification system. Transphyseal extension was seen in four (36%) of 11 lateral humeral condyle fractures, one (25%) of four medial condyle humeral fractures, one (8%) of 12 supracondylar fractures, and one (20%) of five olecranon fractures. Four (57%) of the seven transphyseal fractures were seen to extend to the articular margin, and one (14%) was associated with a 1-mm separation of the articular fragments. Physeal injury. Three (6%) of the 50 children examined had fractures that extended along the physis. In the first patient, a displaced medial humeral epicondyle was visible on radiographs and on MR images (Salter- Harris I). In the second patient, radiographically occult physeal plate widening was revealed by MR imaging (Salter-Harris I) (Fig. 2). In the third patient, radiographically occult physeal injury was seen on MR imaging in association with a metaphyseal fracture of the distal humerus (Salter-Harris II) (Fig. 8). Fig. 5. 5-year-old boy with acute elbow trauma sustained during fall. Axial T2-weighted short tau inversion-recovery image through proximal forearm shows thin rim of hyperintensity around proximal radius (R) that is consistent with annular ligament injury (arrows). Additional sagittal images allowed differentiation of annular ligament injury from fluid surrounding proximal radius. U = ulna. Fig. 6. 11-year-old boy with prior elbow dislocation. Coronal T2- weighted short tau inversion-recovery MR image of elbow shows discontinuity (arrow) of lateral collateral ligament. AJR:176, January 2001 57

Griffith et al. Fig. 7. 10-year-old boy with elbow injury sustained during fall. Sagittal T2-weighted short tau inversion-recovery MR image of elbow shows edema of brachialis and triceps muscles. Bone bruising of ulna and hemarthrosis are also present. The sensitivity and specificity of radiographs in depicting the various types of bone fracture depicted by MR imaging are shown in Table 3. With respect to overall fracture detection, the sensitivity and specificity of radiographs was 71% and 56%, respectively. Bone bruising. Forty-five (90%) of the 50 children examined had bone bruising separate from an accompanying fracture on MR imaging (Figs. 4B and 4C). Bone bruising involved a single bone in 33 (73%) of 45 cases, two bones in 10 (22%) of 45 cases, and three bones in two (4%) of 45 cases. Bone bruising was located in the olecranon process of the ulna (n = 27), the distal humerus (n = 15), the proximal radius (n = 13), and the coronoid process of the ulna (n = 4). Ligament injury. Six (12%) of the 50 children examined had a ligament injury evident on MR imaging. Annular ligament injury occurred in three children (6%) in conjunction with a fracture of the radial neck (n = 1), bone bruising of the olecranon and proximal radius (n = 1), and no other injury (n = 1) (Fig. 5). Collateral ligament injury was seen in three children (6%) in conjunction with a fracture of the olecranon (n = 1), a fracture of the lateral humeral condyle (n = 1), and a dislocated elbow (n = 1) (Fig. 6). Muscle injury. Nineteen (38%) of the 50 children had muscle injury on MR imaging. Muscle injury involved a single muscle in eight (42%) of 19 cases, two muscles in nine (47%) of 19 cases, and three muscles in two (11%) of 19 cases. Muscle injury involved the brachialis 7 (n = 15), triceps brachii (n = 4), supinator (n = 4), brachioradialis (n = 3), flexor digitorum muscles (n = 2), forearm extensors (n = 2), pronator teres (n =1), and biceps brachii (n = 1). Muscle injury was accompanied by bone fracture in all cases except three of the four cases in which supinator muscle injury was seen but a bone fracture was not present (Fig. 9). Most useful MR imaging sequence. MR imaging sequences that provided the most diagnostic information were the T2-weighted STIR 8 Fig. 8. 9-year-old boy with acute elbow trauma sustained during fall. Sagittal T2-weighted MR image shows metaphyseal fracture (arrow) with widening of adjacent growth plate. Fig. 9. 7-year-old girl with acute elbow injury sustained during fall. Axial T2-weighted short tau inversion-recovery MR image of proximal forearm shows edema of supinator muscle (arrows). R = radius, U = ulna. sequences in both the sagittal plane (n = 28) and the coronal plane (n = 10), gradient-recalled echo sequences in the coronal plane (n = 8), and T1- weighted spin-echo sequences in both the sagittal plane (n = 3) and the coronal plane (n = 1). Follow-up radiographs. Follow-up radiographs were available for review in eight of the 11 children who were shown to have a fracture on MR imaging; at presentation, these children had either normal findings on radiographs or a radiographic effusion. The mean interval be- 58 AJR:176, January 2001

MR Imaging of Acute Elbow Trauma in Children tween the initial radiographs and the follow-up radiographs was 15 days (range, 11 17 days). In no case was a healing fracture or periosteal new bone formation apparent. Effect of MR Imaging Findings on Treatment and Outcome Treatment modification. All children in this study were treated with immobilization in a right-angled elbow cast or back slab regardless of the radiographic findings at presentation (Table 4). Additional findings shown by MR imaging did not lead to treatment modification in any child. Short-term clinical outcome. There was a tendency for children with a radiographically fracture or a transphyseal fracture visible on MR imaging to have a marginally longer convalescent period (Table 4), although this difference was not statistically significant (p > 0.05). We did not detect a clear correlation between the extent of bony and soft-tissue injury and the duration of convalescence (Table 4). Medium-term outcome. Forty-seven (94%) of 50 patients returned for clinical reassessment at an average of 1.6 years after injury (range, 0.6 2.4 years) (Table 4). All children made a complete functional recovery (Table 4). No carrying-angle deformity was present in any child. None of the children with a fracture visible solely on MR imaging had limited elbow movement (Table 4). TABLE 4 Group Discussion In this study of selected children with moderately severe elbow trauma, MR imaging showed radiographically occult fractures occurred in 22% of the children, radiographically occult transphyseal extension in 14% of the children, and radiographically occult physeal injury in 4% of the children. In a retrospective study of 54 children with elbow trauma and radiographic evidence of effusion but no fracture, Donnelly et al. [5] found that only 17% had a fracture visible on follow-up radiographs and concluded that the presence of effusion should not be considered evidence of an occult fracture [5]. Hall [6] disagreed with this conclusion, pointing out that many elbow fractures are intraarticular and may not be evident on follow-up radiographs because they do not produce periosteal new bone. When the differences in patient selection and imaging protocols are considered, the present study favors the latter interpretation because 22% of the children with normal findings on radiographs (Fig. 1) and 53% of the children with a radiographic TABLE 3 Note. MR imaging was used as the gold standard. effusion only (Fig. 2) were shown to have a fracture when examined by MR imaging. MR findings show that many childhood elbow fractures, both intra- and extraarticular, are not identifiable on radiography. Followup radiographs did not allow the identification of these fractures. MR imaging helped to clarify radiographically equivocal fractures. Four of six equivocal fractures on radiographs were confirmed as fractures when examined on MR images. No equivocal fractures were diagnosed with MR findings, which reflects the diagnostic criteria applied. Muscle edema (39%), especially of the brachialis, was a common finding and may have resulted from attempts to brace the elbow joint during injury. Muscle injury was accompanied by a bone fracture in all cases, with the exception of three of four cases of supinator injury that were not accompanied by a fracture (Fig. 9). This finding suggests that the mechanism Sensitivity, Specificity, Negative Predictive Value, and Positive Predictive Value of Radiographs Obtained at Presentation in Children with Moderately Severe Elbow Trauma Type of Injury Seen on Radiographs at Presentation Sensitivity (%) Specificity (%) Predictive Value (%) Negative Positive Fracture 71 56 93 57 Transphyseal fracture extension (Salter-Harris IV) 0 100 0 86 Growth plate injury with or without metaphyseal fracture 33 100 100 96 Type and Duration of Immobilization, Duration of Convalescence, and Level of Recovery on Clinical Review After Completion of Study Treatment Duration of Immobilization (wk) Duration of Total Duration of Symptoms After Symptoms (wk) Immobilization (wk) Follow-Up Period (yr) Symptomatic Recovery Range of Movement a Cast Slab Mean Range Mean Range Mean Range Mean Range Complete Incomplete Full Limited Radiographic fracture 25 1 4.2 2 6 5.3 0 10 9.5 3 14 1.6 0.7 2.4 25 b 0 21 b 4 (n = 26) Radiographic effusion 16 1 3.4 1 6 3.6 0 7 7 4 13 1.5 0.7 2.4 16 b 0 15 b 1 (n = 17) Normal radiographic 7 0 3.5 2 4 4.3 2 4 7.8 4 10 2.1 1.9 2.2 6 b 0 5 b 1 findings (n = 7) Fracture seen only on MR 10 1 3.6 1 6 3.7 1 6 7.4 4 13 1.5 0.7 2.4 11 0 11 0 imaging (n = 11) Transphyseal fracture 6 1 4.5 1 6 5.5 2 8 10 4 14 1.3 0.7 1.5 7 0 7 0 seen only on MR imaging (n = 7) a For both the affected elbow and the unaffected elbow, the carrying angle and the range of flexion extension were measured with a goniometer, and pronation supination was assessed subjectively. Varus or valgus deformity was defined as an increase in side-to-side variation in carrying angle of greater than 10. Limited movement was defined as a limitation of either flexion and extension or pronation and supination of greater than 5 when compared with the contralateral unaffected elbow. b One patient in each of these groups was not available for clinical examination at 7 months. AJR:176, January 2001 59

Griffith et al. of supinator muscle injury may differ from that of other muscle injury. This study helps broaden our understanding of acute elbow trauma in children in four respects. First, this study shows that a spectrum of both bone and soft-tissue injuries occurs in children during elbow trauma, much of which is not recognizable radiographically (Tables 1 and 2). In radiographically comparable subgroups, a spectrum of injures is shown by MR imaging. This spectrum of injury may, in part, explain the variable recovery seen in children with radiographically comparable injuries. However, in this study we were not able to show a clear correlation in this respect. For example, some children with transphyseal bone fracture, bone bruising, and muscle injury recovered fully in 4 weeks, whereas others with apparently similar injuries took 10 weeks to fully recover. As expected, the range of unrecognized injuries was least severe in those with normal findings on radiographs than in those with a radiographic effusion or fracture (Table 1). Second, this study shows that radiographically unrecognized transphyseal or physeal injury is not an uncommon finding, occurring in 20% of children with moderately severe injury. Third, this study confirms that most children with moderately severe acute elbow trauma sustain an appreciable injury, even those with normal findings on radiographs (Table 1). This finding helps justify the active treatment, be it a short period of immobilization or not, of all children with moderately severe elbow trauma. Fourth, this study shows that in this subgroup of injury the additional findings revealed by MR imaging have no appreciable bearing on patient treatment or outcome in either the short or the long term. The duration of cast immobilization in children with a fracture visible solely on MR imaging was not different from that of other subgroups. Similarly, on longer term follow-up, the outcome of children with a fracture visible solely on MR imaging was not different from that of other subgroups (Table 4). One limitation of this study was that a direct comparison between the accuracy in fracture detection of radiographs and that of MR images was not possible given the delay (mean, 6 days; range, 1 16 days) between radiography and MR imaging. A further limitation was the use of only conventional frontal and lateral radiographs. The addition of oblique views and a high-detail imaging system may have revealed some of the occult fractures shown only by MR imaging. Our current policy is to operatively reduce and internally fixate radiographically displaced elbow fractures in pediatric patients. All other patients with significant elbow trauma, regardless of radiographic findings, are treated with a variable period of immobilization. Therefore, according to this policy, radiographs serve primarily to identify patients who would benefit from surgical intervention. Most institutions may find currently performing MR imaging on all pediatric patients with moderately severe elbow trauma impractical, and on the basis of this study, we do not recommend this approach. Although lateral condylar fractures are the most likely to be associated with transphyseal fracture extension, this feature can accompany other fractures as can physeal injuries. In this study, no particular subgroup of elbow injury in pediatric patients was identified that would especially warrant examination on MR imaging. Although MR images reveal more injury than radiographs, this additional information appears to have little bearing on treatment or clinical outcome. References 1. Beltran J, Rosenberg ZS, Kawelblum M, Montes L, Bergman AG, Strongwater A. Pediatric elbow fractures: MRI evaluation. Skeletal Radiol 1994;23: 277 281 2. Carey J, Spence L, Blickman H, Eustace S. MRI of pediatric growth plate injury: correlation with plain film radiographs and clinical outcome. Skeletal Radiol 1998;27:250 255 3. Petit P, Panuel M, Faure F, et al. Acute fracture of the distal tibial physis: the role of gradient-echo MR imaging versus plain film examination. AJR 1996;166:1203 1206 4. Rangger C, Kathrein A, Freund MC, Klestil T, Krecy A. Bone bruise of the knee: histology and cryosections in five cases. Acta Orthop Scand 1998;69:291 294 5. Donnelly LF, Klostermeier TT, Klosterman LA. Traumatic elbow effusions in pediatric patients: are occult fractures the rule? AJR 1998;171:243 245 6. Hall FM. Traumatic elbow effusions. (letter) AJR 1999;172:550 551 7. Feldman F, Staron R, Zwass A, Rubin A, Haramati N. MR imaging: its role in detecting occult fractures. Skeletal Radiol 1994;23:439 444 60 AJR:176, January 2001