Rachel S. Oeppen 1,2 Susan. Connolly 3 Jenny T. encardino 4 Diego Jaramillo 1 Received May 19, 2003; accepted after revision July 28, 2003. 1 Department of Pediatric Radiology, Massachusetts General Hospital, 32 Fruit St., oston, M 02114. ddress correspondence to D. Jaramillo. 2 Present address: Department of Radiology, Southampton General Hospital, Southampton, Hants SO16 6YD, England. 3 Department of Radiology, Children s Hospital, 300 Longwood ve., oston, M 02115. 4 Department of Radiology, Huntington Hospital, Huntington, NY 11743. JR 2004;182:111 117 0361 803X/04/1821 111 merican Roentgen Ray Society cute Injury of the rticular Cartilage and Subchondral one: Common but Unrecognized Lesion in the Immature Knee OJECTIVE. We studied the prevalence of injuries of the articular cartilage and subchondral bone after acute trauma in skeletally immature knees using high-resolution MRI. MTERILS ND METHODS. We reviewed knee MRIs of 126 young children and adolescents suspected to have internal knee derangement, including 82 with open physes and a control group of 44 who were skeletally mature. High-resolution proton density and T2-weighted pulse sequences were used in all patients. The prevalence of common injuries in the two groups was compared using chi-square analysis. Levels of interobserver agreement for evaluation of chondral lesions in the skeletally immature group were determined using the kappa statistic. RESULTS. In the skeletally immature group, chondral lesions were the most prevalent injuries (prevalence = 0.34, p = 0.009) followed by meniscal and anterior cruciate ligament injuries (prevalence = 0.23 and 0.24, respectively). No significant difference in the prevalence of chondral injury before and after physeal closure was seen (p = 0.45). There was no significant difference in the prevalence of anterior cruciate ligament injuries between the two groups, but meniscal injuries were more prevalent in the skeletally mature patients (prevalence = 0.41, p = 0.037). Interobserver agreement for chondral injuries in the group with open physes was good (weighted κ = 0.45 0.51). CONCLUSION. The most common injuries occurring as a result of acute trauma to the immature knee were chondral. In patients with open physes, chondral injuries were significantly more prevalent than anterior cruciate ligament and meniscal injuries. M RI of disorders of the knee in children is now widespread, and it is important to determine whether the lesions identified in adults are also present before skeletal maturity. Previous reports have established the diagnostic accuracy and relative frequency of meniscal and ligamentous injuries in children [1 5]. These studies have not addressed the frequency of cartilaginous injuries on MRI studies of pediatric patients. In recent years, interest in the advancement and development of new MRI techniques for the optimal evaluation of cartilage has been considerable [6 9]. There is also an increasing awareness that chondral injury may occur as a result of acute trauma. Potter et al. [10] described the use of multiplanar high-resolution fast spinecho proton density sequences in the accurate assessment of articular cartilage injury in the knees of adults, validating their findings with arthroscopy. To our knowledge, no reports relating to the use of this protocol in children and adolescents have been published, and little has been published regarding chondral injuries in children. t our institution, a high-resolution MRI protocol has been used routinely for the past 4 years. We performed a retrospective study to assess the prevalence of chondral injuries and other lesions after acute knee trauma in children and adolescents and to determine whether there are differences in prevalence of injury related to skeletal maturity. Materials and Methods The MRI studies of 126 children and adolescents investigated for possible internal derangement of the knee after trauma were evaluated retrospectively. Eighty-two patients were skeletally immature (42 girls, 40 boys; age range, 4 years 6 months 17 years 6 months, and 44 had closed physes (30 girls, 14 boys; age range, 14 years 8 months 20 years 1 month; median age, 17.4 years). ll of the studies were acquired using a standard high-resolution MRI protocol. Each study was reviewed, and the presence JR:182, January 2004 111
Oeppen et al. Fig. 1. Drawings show classification of osteochondral injuries. Grade 0 injury (not shown) is intact cartilage with normal signal and uniform thickness., Drawing shows grade 1 injury: thickening with abnormal signal., Drawing shows grade 2 injury: superficial ulceration or fissuring. C, Drawing shows grade 3 injury: deep ulceration or fissuring. D, Drawing shows grade 4 injury: full-thickness chondral injury with bruising of subchondral bone. E, Drawing shows grade 5 injury: osteochondral injury with separation of osteochondral fragment. and grade or type of injury or injuries were recorded. The 82 examinations of skeletally immature knees were each evaluated independently by three musculoskeletal radiologists, which enabled us to assess interobserver variability for the detection of chondral lesions. The 44 skeletally mature knees were reviewed once by any one of the three radiologists to allow comparison in frequency of injuries between the skeletally immature and mature patient groups. Patients Patients younger than 21 years old examined at Massachusetts General Hospital from January TLE 1 Kappa Statistic and Level of Interobserver greement κ Level of greement < 0 Poor 0.0 0.19 Slight 0.2 0.39 Fair 0.4 0.74 Good 0.75 1.00 Excellent 2000 to December 2001 for internal knee derangement were included in the study (142 individual examinations). Two patients with congenital deformities, two with joint disorders (juvenile rheumatoid arthritis and Lyme disease), three with tumors (chondroblastoma, osteochondroma, and lymphoma), one with a prior burn, one with a history of knee surgery, and one with osteomyelitis were excluded. Six patients were excluded because the examination was technically inadequate, which was defined as one that did not include the entire proximal tibial physis or that had motion artifacts. The remaining 126 examinations were considered both technically adequate and without significant underlying disease. Patients were classified as skeletally immature if at least one of the two major physes of the knee was open on MRIs (i.e., cartilaginous signal evident throughout the entire growth plate of the distal femoral or proximal tibial physes). MRI C D E The studies were performed on 1.5-T systems (Signa, General Electric Medical Systems, Milwaukee, WI). Each study included the following sequences: intermediate-weighted fast spin-echo localizing images (TR/TE, 1,000/10); axial fast spin-echo proton density weighted images (3,500/35; echo-train length, 8); sagittal fast spin-echo proton density weighted images (3,500/11; echo-train length, 4); fast spin-echo T2-weighted images (4,000/100; echo-train length, 8); coronal fat-suppressed fast spin-echo proton density weighted images (3,500/11; echo-train length, 4); and usually conventional T1-weighted spin-echo images (600/10). Imaging parameters for proton density weighted sequences were as follows: field of view, 16 cm; section thickness, 3.0 mm; interslice gap, 1 mm; matrix size, 256 320 512; and number of signals acquired, 2. The examinations were performed using an extremity linear receive transmit coil. Fat suppression was achieved through chemical shift selective saturation. Image nalysis Each study was reviewed by one of three radiologists experienced in musculoskeletal radiology without knowledge of a patient s specific history or mechanism of injury. ll three radiologists evaluated the skeletally immature group for evidence of chondral injuries, which allowed assessment of interobserver agreement for the detection of these lesions. Disagreements were settled by consensus. 112 JR:182, January 2004
MRI of the Immature Knee The tibial plateau, femoral condyles, and patella were examined for evidence of fracture, bone bruising, and chondral injury. The cruciate and collateral ligaments and the quadriceps, patellar, and popliteus tendons were inspected, and any sprain, tear, or avulsion injury was noted. The menisci were evaluated for the presence of abnormal intrasubstance signal and frank tears. Effusions and injuries to the patellar retinaculum, physes, and apophyses were also noted. Chondral injuries were assessed using a modified version of the Outerbridge classification [10, 11] (Fig. 1). We arbitrarily defined a full-thickness cartilaginous injury (grade 3) as an injury in which the entire signal intensity of the articular surface was lost. We called subchondral bone involvement (grade 4) only when clear signal intensity changes were present in the adjacent subchondral bone. Fifteen cartilaginous zones were assessed: three running from anterior to posterior on the medial and lateral femoral condyles and each tibial plateau; the medial and lateral facets; and apical surface of the patella. TLE 2 Frequency and Location of Chondral Injuries in Patients with Immature and Mature Skeletons No. of Patients Skeleton Prevalence Total Injured 82 44 28 18 Immature Mature p 0.34 0.41 0.45 Injuries (%) No. of Chondral Injuries Femoral Patellar 43 41 72 54 23 39 0.04 0.059 rthroscopy rthroscopy reports were available for 11 patients, five of whom were skeletally immature. The arthroscopy reports were correlated with MRIs with respect to the presence and location of articular injuries. Statistical nalysis Interobserver reliability for chondral lesions in the skeletally immature group was assessed using the kappa statistic. The ranges for kappa with respect to the level of agreement according to Landis and Koch [12] are shown in Table 1. The significance of differences in prevalence of injuries be- Fig. 2. Grade 1 lesion in 13-year-old girl with diffuse pain after trauma. Sagittal proton density weighted image shows area of focal thickening (arrow) of cartilage of medial femoral condyle adjacent to posterior horn of medial meniscus. Hemarthrosis is also present. Fig. 3. Grade 2 lesion in 16-year-old boy with history of pain after fall., Sagittal proton density weighted image shows subtle area of focal thinning (arrows) of cartilage of midportion of lateral femoral condyle., Coronal fat-suppressed proton density weighted image shows subtle partial defect (arrow) of corresponding area of lateral femoral condyle cartilage. JR:182, January 2004 113
Oeppen et al. Fig. 4. Grade 3 lesion in 14-year-old girl with history of knee trauma and lateral joint pain. Patient also had osteochondral injury in medial patella., Coronal fat-suppressed proton density image shows area of absent articular cartilage (arrow) in anterior aspect of lateral femoral condyle. Large effusion and diffuse edema of marrow are also present., Sagittal proton density image shows area of cartilage discontinuity (arrow), anterior to anterior horn of lateral meniscus. tween the groups was established using twosample tests of proportion. ssociations between injuries and skeletal maturity were evaluated using the chi-square test. Results Chondral Lesions The chondral injury data are summarized in Table 2. Forty-three chondral injuries were identified in 28 of the 82 skeletally immature patients (prevalence = 0.34). Grade 1 lesions (Fig. 2), seen in five patients, were characterized by a focal increase in thickness or by an alteration of the signal intensity of articular cartilage. grade 2 lesion (Fig. 3) was seen in one patient, in whom there was a focal partial-thickness defect of the cartilage. Grade 3 lesions (Fig. 4), seen in four patients, involved the entire thickness of the cartilage. Grade 4 lesions (Fig. 5), seen in 26 patients, involved the entire cartilage thickness and had definite signal intensity changes in the adjacent bone. Grade 5 lesions (Figs. 6 and 7) were seen in seven patients, in whom there was separation of an osteochondral fragment from the epiphysis; the fragment was visible in four of the seven. rticular chondral injuries were present in 18 of the 44 skeletally mature patients (prev- Fig. 5. Grade 4 lesion in 14-year-old girl with history of twisting injury. Other images (not shown) revealed anterior cruciate ligament tear and partial bucket-handle tear of medial meniscus. Sagittal proton density image shows discontinuity in cartilage in region of sulcus of lateral femoral condyle with adjacent edema (arrow). 114 alence = 0.41). The difference in the prevalence of chondral injury between the two groups was not statistically significant (p = 0.45). However, the distribution of chondral lesions in the two groups was significantly different. Thirty-one of the 43 chondral injuries in the skeletally immature group were femoral, compared with only 22 of 41 chondral lesions in the mature group (Table 2, p = 0.04). In contrast, only 10 of 43 lesions in the skeletally immature were patellar (Fig. 6) compared with 16 of 41 chondral lesions in the mature group (p = 0.059). Thirty-three osteochondral injuries involving the subchondral bone (grades 4 and 5) were seen in 24 skeletally immature patients (prevalence = 0.29). Six loose intraarticular bodies were present in four patients (three patellar [Fig. 6], two femoral, and one tibial in origin). Fifteen osteochondral injuries were identified in 10 skeletally mature patients (prevalence = 0.23), and no loose bodies were seen in this group. Interobserver agreement for chondral injuries in the skeletally immature group was good for both grade and location of the lesions with weighted kappa values for all observations ranging from 0.45 to 0.51. Table 3 shows the variation in weighted kappa values by location; agreement was better for the lateral femoral cartilage than for the patella. Kappa values for the few tibial lesions (Fig. 8) showed great variation. JR:182, January 2004
MRI of the Immature Knee Fig. 6. Grade 5 lesion in 14-year-old boy with history of lateral patellar dislocation after fall., xial proton density image shows discontinuity (thin arrow) of medial articular surface of patella at insertion of medial patellar retinaculum. Loose body (thick arrow) is adjacent to it., xial fat-suppressed T2-weighted image shows that subchondral edema is adjacent to area of cartilaginous disruption (arrow). rthroscopy rthroscopy was performed in five of the patients in the skeletally immature group. Chondral injuries were seen at arthroscopy in two of the five, and the cartilage was intact in the remaining three. On MRIs, only these two of the five patients had cartilaginous injuries. In the six of the patients in the skeletally mature group, arthroscopy was performed. Chondral injuries were reported in the arthroscopy reports of three. ll had MRI evidence of cartilaginous injuries that agreed in location and number with those seen arthroscopically. In the remaining three, the status of the cartilage was not mentioned. MRIs showed a grade 1 lesion in the patellar apex in one and a grade 2 lesion in the medial femoral condyle in another. patients had anterior cruciate ligament injuries (prevalence = 0.24), with a similar prevalence (11/44; prevalence = 0.25) noted in the skeletally mature patients. The difference in the prevalence of anterior cruciate ligament injuries in the two groups was not statistically significant (p = 0.94). Therefore, a significant excess of chondral injuries compared with meniscal and anterior cruciate ligament injuries was seen in the skeletally immature group (p = 0.009), but chondral and meniscal injuries were equal in prevalence in the skeletally mature patients. Other Injuries Four patients had Salter-Harris fractures that were not detected radiographically. Eight other patients had subtle physeal injuries (mi- Meniscal and Ligamentous Injuries Twenty-four meniscal injuries were present in 19 of 82 skeletally immature patients (prevalence = 0.23), eight involving the medial meniscus and 16 involving the lateral meniscus. Significantly more meniscal injuries eight medial and 13 lateral in 18 of the 44 patients (prevalence = 0.41) were detected in the skeletally mature group (p = 0.037). Twenty (12 boys, eight girls) of 82 skeletally immature JR:182, January 2004 Fig. 7. Grade 5 lesion in 12-yearold girl who experienced popping feeling in knee while playing basketball. Sagittal proton density weighted image shows focal loss of cartilage (arrow) and subchondral bone in anterior aspect of lateral femoral condyle. Osteochondral fragment was seen on another image (not shown). 115
Oeppen et al. TLE 3 Interobserver greement by Location Location Weighted κ 95% CI Lateral femur 0.66 0.42 0.91 Medial femur 0.52 0.29 0.74 Patella 0.51 0.29 0.73 Lateral tibia 0.65 0.27 1.02 Medial tibia 0.49 0.10 to 1.07 Note. CI = confidence interval. nor widening and alteration in physeal signal or paraphyseal bruising). No fractures were present in the skeletally mature group. Discussion This study shows that chondral injuries are the most common lesions in children examined on MRI for evaluation of internal derangement of the knee. Chondral lesions are significantly more prevalent than meniscal or ligamentous injuries in skeletally immature patients and are almost as prevalent in the skeletally mature. The distribution of chondral injuries differs with skeletal maturity: proportionally more femoral lesions are seen in skeletally immature patients (i.e., before physeal closure), and more patellar lesions are seen in skeletally mature patients. marked increase in injuries, particularly those related to sports, has been noted in older children [13] and adolescents. recent series documented a 54% increase in injuries from 1983 to 1999 [14]. Chondral injuries are significant because they may predispose a patient to premature osteoarthritis [15]. This tendency is especially true in patients with full-thickness cartilage injuries, which may be amenable to autologous chondrocyte transplantation [16]. Loose intraarticular bodies can cause locking and pain. Prior MRI reports have emphasized injuries to the ligaments and menisci. In fact, several of the largest series of MRI examinations of the knee in children did not comment on chondral injuries [2 5]. This omission may be because of the lack of sensitivity of conventional MRI sequences, particularly T1-weighted images, to injuries involving the articular cartilage. Chondral injuries are difficult to detect on MRI. Two techniques are used most commonly for this purpose. Fat-suppressed spoiled gradient-recalled echo imaging initially emerged as the best technique to show injuries to the surface of the joints [17]. Unfortunately, this technique does not depict the menisci and ligaments well. More recently, high-resolution fast spin-echo imaging with a longer TE has been shown to correlate well with arthroscopy for the depiction of chondral injuries [10]. This technique is, in turn, very sensitive to other lesions in the knee; therefore, it has become an integral part of many imaging protocols. With this sequence, the articular cartilage is of lower signal intensity than the adjacent joint fluid. n abnormality in the cartilage is seen as a discontinuity in the low-signal-intensity cartilage. Detection of subtle injuries (grades 1 and 2 in this series [Figs. 2 and 3]) continues Fig. 8. Grade 4 lesion in tibia of 16-year-old boy with history of knee swelling after fall. Coronal fat-suppressed proton density weighted image shows full-thickness lesion with accompanying slight subchondral edema in periphery of tibial plateau (arrow), adjacent to insertion of iliotibial band. to be challenging, and further improvements in MRI techniques are necessary to refine the diagnosis of these lesions. Subchondral bone marrow edema, best noted on fat-suppressed proton density and T2-weighted sequences, frequently accompanied cartilaginous injuries. Increased signal intensity of bone alerts the radiologist to the presence of a subchondral injury. Interobserver agreement was good overall and in the specific locations. Our data suggest that agreement is slightly better in the femur than the patella, but the overall range of agreement is the same for both areas. Differentiation of the deep layers of articular cartilage and subchondral bone using routine MRI is difficult [18 20]. The transition between cartilage and bone will appear to vary depending on factors such as TE and the orientation of the cartilage. For this reason, we decided to arbitrarily define a subchondral injury as being present only when a clear focus of abnormal signal intensity on the bone was present. Some of the lesions classified as grade 3 may indeed have had minimal involvement of the subchondral bone. The prevalence of anterior cruciate ligament injury in both skeletally immature and mature patient groups in our study (0.24 and 0.25, respectively) was similar to that reported previously [2 4]. The overall prevalence of meniscal injury was 0.23 and 0.41 before and after physeal closure, respectively. In previous studies, researchers have found prevalence of meniscal injury to be approximately 0.25 in skeletally immature patients [2] and as high as 0.59 in skeletally mature patients [4]; the findings from these prior reports agree with our findings. However, the medial meniscus was injured significantly more frequently than the lateral meniscus in these studies, whereas we found a significant excess of lateral meniscal tears both before and after physeal closure. We do not have an explanation for the high prevalence of lateral meniscal injuries in our group of patients. Only three lateral discoid menisci were noted in the immature group, and none was noted in the mature group; this difference is unlikely to account for the predominance of lateral meniscal injuries. The prevalence of physeal injuries and of other occult injuries was low in comparison with a previous report [21]. Whether immaturity protects against chondral injuries is uncertain. iomechanical differences related to the presence of residual epiphyseal cartilage under the 116 JR:182, January 2004
MRI of the Immature Knee articular cartilage would be expected. We compared patients before and after physeal closure to evaluate this difference. lthough the prevalence in younger patients was slightly lower, this difference was not significant. Most of the patients in the immature group were older than 10 years; at this stage of development, most of the epiphyseal cartilage has ossified. eyond biologic differences, it is possible that the patterns of injury differ before and after skeletal maturity because activities change between late childhood and puberty, early adolescence, and late adolescence [13, 22]. Our study does not include a systematic validation with arthroscopy. rthroscopy is performed less frequently in children than in adults [23]. Patients were referred from a number of sources: orthopedic surgeons, sports medicine physicians, and pediatricians in the community. Information about outcome or eventual arthroscopy and surgery could not be obtained systematically. In the five skeletally immature patients who underwent arthroscopy, there was complete agreement between MRI and arthroscopy. In the six skeletally mature patients who underwent arthroscopy, two lesions seen on MRIs were not reported arthroscopically. High-resolution MRI, however, has already been validated with arthroscopy; therefore, we believe that it is valid to discuss these injuries in the absence of arthroscopic correlation [10]. Our goal was to evaluate differences in distribution and prevalence of injuries, rather than to validate the technique. nother potential limitation of our study is its retrospective nature, given that our population reflects the local patterns of referral to MRI. The clinical presentations encompassed by internal derangement of the knee vary considerably. Inclusion of only patients presenting with specific clinical indications to a well-defined group of clinicians who had uniform criteria for referral to MRI would have been ideal. We have no reason to suspect, however, that there was a bias toward specific injuries. We arbitrarily selected the upper age limit as 21 years. We evaluated interobserver variability in the immature population only because this was the population of interest. One could argue, however, that the MRI studies of the immature patients underwent closer scrutiny than those of their mature counterparts because studies of the former were reviewed three times, whereas those of the latter were analyzed only once. In summary, our study shows that chondral injuries are the most common lesions after acute knee trauma in the skeletally immature population. The prevalence of these lesions does not change significantly with skeletal maturity. Chondral lesions are subtle but can be significant clinically and may easily remain undetected. We recommend that in skeletally immature patients MRI evaluation of internal knee derangement include sequences that are sensitive to articular cartilage injury. cknowledgments We thank Elkan F. Halpern for help with the statistical analysis; Linda anzi for help with illustrations; and Sherry rec, Noemi Chavez, and Elizabeth Olear for help with the preparation of this manuscript. References 1. l-otaibi L, Siegel MJ. The pediatric knee. Magn Reson Imaging Clin N m 1998;6:643 660 2. Lee K, Siegel MJ, Lau DM, Hildebolt CF, Matava MJ. nterior cruciate ligament tears: MR imaging-based diagnosis in a pediatric population. Radiology 1999;213:697 704 3. Kocher MS, DiCanzio J, Zurakowski D, Micheli LJ. Diagnostic performance of clinical examination and selective magnetic resonance imaging in the evaluation of intraarticular knee disorders in children and adolescents. m J Sports Med 2001;29:292 296 4. Zobel MS, orrello J, Siegel MJ, Stewart NR. Pediatric knee MR imaging: pattern of injuries in the immature skeleton. Radiology 1994;190:397 401 5. Major NM, eard LN Jr, Helms C. ccuracy of MR imaging of the knee in adolescents. JR 2003;180:17 19 6. Disler DG. Fat-suppressed three-dimensional spoiled gradient-recalled MR imaging: assessment of articular and physeal hyaline cartilage. JR 1997;169:1117 1123 7. Peterfy CG. Scratching the surface: articular cartilage disorders in the knee. Magn Reson Imaging Clin N m 2000;8:409 430 8. Recht M, obic V, urstein D, et al. Magnetic resonance imaging of articular cartilage. Clin Orthop 2001;391[suppl]:S379 S396 9. Gold GE, eaulieu CF. Future of MR imaging of articular cartilage. Semin Musculoskelet Radiol 2001;5:313 327 10. Potter HG, Linklater JM, llen, Hannafin J, Haas S. Magnetic resonance imaging of articular cartilage in the knee: an evaluation with use of fast-spin-echo imaging. J one Joint Surg m 1998;80:1276 1284 11. Noyes FR, Stabler CL. system for grading articular cartilage lesions at arthroscopy. m J Sports Med 1989;17:505 513 12. Landis JR, Koch GG. The measurement of observer agreement for categorical data. iometrics 1977;33:159 174 13. Kingma J, Ten Duis HJ. Injuries due to school sports accidents in 4 to 13-yr.-old children. Percept Mot Skills 2000;90:319 325 14. Jones SJ, Lyons R, Sibert J, Evans R, Palmer SR. Changes in sports injuries to children between 1983 and 1998: comparison of case series. J Public Health Med 2001;23:268 271 15. Conaghan PG. Update on osteoarthritis part 1: current concepts and the relation to exercise. r J Sports Med 2002;36:330 333 16. rittberg M, Lindahl, Nilsson, Ohlsson C, Isaksson O, Peterson L. Treatment of deep cartilage defects in the knee with autologous chondrocyte transplantation. N Engl J Med 1994;331:889 895 17. Disler DG, McCauley TR, Kelman CG, et al. Fatsuppressed three-dimensional spoiled gradientecho MR imaging of hyaline cartilage defects in the knee: comparison with standard MR imaging and arthroscopy. JR 1996;167:127 132 18. McGibbon C, encardino J, Palmer WE. Subchondral bone and cartilage thickness paragraph sign from MRI: effects of chemical-shift artifact. MGM 2003;16:1 9 19. McGibbon C, encardino J, Yeh ED, Palmer WE. ccuracy of cartilage and subchondral bone spatial thickness distribution from MRI. J Magn Reson Imaging 2003;17:703 715 20. Xia Y. Magic-angle effect in magnetic resonance imaging of articular cartilage: a review. Invest Radiol 2000;35:602 621 21. Mann DC, Rajmaira S. Distribution of physeal and nonphyseal fractures in 2,650 long-bone fractures in children aged 0-16 years. J Pediatr Orthop 1990;10:713 716 22. elechri M, Petridou E, Kedikoglou S, Trichopoulos D. Sports injuries among children in six European union countries. Eur J Epidemiol 2001;17:1005 1012 23. Maffulli N, Chan KM, undoc RC, Cheng JC. Knee arthroscopy in Chinese children and adolescents: an eight-year prospective study. rthroscopy 1997;13:18 23 JR:182, January 2004 117