CHILDREN’S ORTHOPAEDICS
A comparison of the simplified olecranon and digital methods of assessment of skeletal maturity during the pubertal growth spurt F. Canavese, Y. P. Charles, A. Dimeglio, S. Schuller, M. Rousset, A. Samba, B. Pereira, J-P. Steib From CHU Clermont-Ferrand, Department of Pediatric Surgery, Clermont-Ferrand, France F. Canavese, MD, PhD, Professor, Consultant Paediatric Orthopaedic Surgeon, Professor of Paediatric Orthopaedic Surgery, Chief of Staff M. Rousset, MD, Consultant Paediatric Orthopaedic Surgeon A. Samba, MD, Consultant Paediatric Orthopaedic Surgeon CHU Clermont-Ferrand, Department of Pediatric Surgery, 1 Place Lucie et Raymond Aubrac, 63003 Clermont-Ferrand, France. Y. P. Charles, MD, PhD, Professor, Consultant Orthopaedic Surgeon, Professor of Orthopaedic Surgery S. Schuller, MD, Consultant Orthopaedic Surgeon J-P. Steib , MD, PhD, Professor, Consultant Orthopaedic Surgeon, Professor of Orthopaedic Surgery Hôpitaux Universitaires de Strasbourg, Department of Spine Surgery, 1 Place de l’hôpital, BP 426, 67091 Strasbourg Cedex, France. A. Dimeglio , MD, Professor, Professor of Paediatric Orthopaedic Surgery Université de Montpellier, Faculty of Medicine, 2 Rue de l’école de Médecine, 34060 Montpellier, France. B. Pereira, PhD, Biostatistician CHU Clermont-Ferrand, Biostatistics Unit (DRCI), 58 Rue Montalembert, 63000 ClermontFerrand, France. Correspondence should be sent to Prof Dr F. Canavese; e-mail: [email protected] ©2014 The British Editorial Society of Bone & Joint Surgery doi:10.1302/0301-620X.96B11. 33995 $2.00 Bone Joint J 2014;96-B:1556–60. Received 19 February 2014; Accepted after revision 11 August 2014
1556
Assessment of skeletal age is important in children’s orthopaedics. We compared two simplified methods used in the assessment of skeletal age. Both methods have been described previously with one based on the appearance of the epiphysis at the olecranon and the other on the digital epiphyses. We also investigated the influence of assessor experience on applying these two methods. Our investigation was based on the anteroposterior left hand and lateral elbow radiographs of 44 boys (mean: 14.4; 12.4 to 16.1 ) and 78 girls (mean: 13.0; 11.1 to14.9) obtained during the pubertal growth spurt. A total of nine observers examined the radiographs with the observers assigned to three groups based on their experience (experienced, intermediate and novice). These raters were required to determined skeletal ages twice at six-week intervals. The correlation between the two methods was determined per assessment and per observer groups. Interclass correlation coefficients (ICC) evaluated the reproducibility of the two methods. The overall correlation between the two methods was r = 0.83 for boys and r = 0.84 for girls. The correlation was equal between first and second assessment, and between the observer groups (r ≥ 0.82). There was an equally strong ICC for the assessment effect (ICC ≤ 0.4%) and observer effect (ICC ≤ 3%) for each method. There was no significant (p < 0.05) difference between the levels of experience. The two methods are equally reliable in assessing skeletal maturity. The olecranon method offers detailed information during the pubertal growth spurt, while the digital method is as accurate but less detailed, making it more useful after the pubertal growth spurt once the olecranon has ossified. Cite this article: Bone Joint J 2014;3:1556–60
Skeletal age represents a useful parameter for decision making in paediatric orthopaedics. Different methods are described using hand, elbow, foot, knee or pelvic radiographs for reference.1-7 The first two years of puberty are characterised by a significant growth spurt, followed by three years of gradual growth.8-10 The pubertal growth spurt occurs at around 11 to 13 years of skeletal age in girls and 13 to 15 years in boys.8-10 Sauvegrain, Nahm and Bronstein11 first developed a method to determine skeletal age from anteroposterior and lateral elbow radiographs in 1962. In 2005, Dimeglio et al12 simplified the original Sauvegrain method and demonstrated that morphology of the olecranon apophysis on lateral radiographs goes through five distinct and characteristic appearances, which are easy to distinguish at six-month intervals (Fig. 1).12 This sequence is clinically relevant because the olecranon exhibits a clear and regular morphological development during the pubertal growth spurt,
when the Risser sign is still 0.13 Closure of the epiphyses at the elbow indicates the end of the accelerated growth spurt, when the adolescent is entering the decelerating phase of pubertal growth.14 In 2008, Sanders et al15 presented a new method of digital skeletal age assessment derived from the Tanner-Whitehouse-III method,16 which is based on a radiological analysis of the metacarpals and fingers on anteroposterior hand radiographs. The simplified digital method provides several markers prior to the Risser 1 stage of skeletal maturity, it has been shown by various authors to add helpful information to the Risser stage 013 and covers the period from Risser 1 to Risser 5.14,15,17-20 Stages 1 and 2 correspond to the pre-pubertal period, stages 3 and 4 correspond to the pubertal growth spurt (Risser grade 0), and stages 5 to 8 cover the period from Risser grade 1 to 5, at which stage full skeletal maturity has occurred (Fig. 2). THE BONE & JOINT JOURNAL
A COMPARISON OF THE SIMPLIFIED OLECRANON AND DIGITAL METHODS OF ASSESSMENT OF SKELETAL MATURITY
1557
Fig. 1 Radiological sequence of skeletal age assessment based on the olecranon from 11 to 13 years in girls and 13 to 15 years in boys.
Fig. 2 Radiographs showing stages 3 and 4 (of the digital skeletal maturity assessment15 covering the same growth period as the olecranon sequence.12
Both Dimeglio et al’s12 and Sanders et al’s15 simplified methods of assessing skeletal age have been reported to have a modest learning curve and ease of clinical application. This study compares the accuracy of both methods during the pubertal growth spurt and investigates the accuracy in relation to the experience of the assessor. We also discuss the clinical relevance of each method.
Material and Methods Assessment of skeletal age was made on a set of anteroposterior left hand and lateral elbow radiographs using the simplified methods; Dimeglio et al12 for the elbow and Sanders et al15 for the simplified digital method. The consecutive radiographs had been taken during the pubertal growth period of 44 boys (mean: 14.4 years; 12.4 to 16.1) and 78 girls (mean: 13.0 years; 11.1 to 14.9) in the assessment of conditions such as idiopathic limb-length discrepancy, idiopathic scoliosis or sequelae of epiphyseal trauma of the lower limb. The decision of taking radiographs was made by the paediatric orthopaedic surgeon. Radiographs were taken during the follow-up of children whose standing height VOL. 96-B, No. 11, NOVEMBER 2014
increased by more than 4 cm within the last 6 months.14 The mean chronological ages were 14.4 years (12.4 to16.1; SD 1.3;) for boys and 13.0 years (11.1 to 14.9; SD 1.6) for girls. The children had no history of systemic illnesses or endocrinopathy. In clinical practice we usually further assess the pubertal stage with other factors such as sitting and standing height measurement and the presence of secondary sexual characteristics. In this study, these parameters were not taken into account. Institutional Review Board approval was obtained for this study (CPP 2014/CE09). We assigned nine observers into three groups of three depending on their experience. Group I (expert); more than ten years of experience in paediatric orthopaedic surgery, Group II (intermediate); orthopaedic surgeons with an essentially adult practice with < 20% of their work engaged in managing disorders of the paediatric spine, and group III (novice); orthopaedic residents engaged in paediatric training. All investigators had used the methods previously in paediatric orthopaedic clinics; no supplementary advice or training was given. When assessing the radiographs, all nine observers worked independently and no time limit was imposed. Each observer performed skeletal age determination twice with an interval of six weeks between both grading sessions. The radiographs were reordered between the analyses. Gender was not blinded but name and chronological age was concealed by an individual who did not participate in the review (BP). This data were transferred to a separate file which was not accessible to any of the observers during the analysis. We observed the extent of the correlation between the simplified olecranon and digital methods and further compared both methods according to the level of experience of the three observer groups. Statistical analysis. Statistical analysis was performed using Stata software version 12 (StataCorp, College Station, Texas). Quantitative variables were expressed as mean and standard deviation (SD). To measure the correlation between the two methods, Pearson correlation coefficients were calculated for all patients and for boys and girls separately, according to the observers’ experience. As the two methods had a different data spread, an analysis using Lin concordance coefficients was performed to compare
1558
F. CANAVESE, Y. P. CHARLES, A. DIMEGLIO, S. SCHULLER, M. ROUSSET, A. SAMBA, B. PEREIRA, J-P. STEIB
Table I. Mean olecranon skeletal age assessment (years) and digital score (1 to 8) with standard deviation (SD) grouped according to the experience of the observers
Olecranon simplified method Boys Girls Digital simplified method Boys Girls
Group I: experienced
Group II: intermediate
Group III: novice
Group I: experienced
Group II: intermediate
Group II: novice
14.45 (SD 0.69) 12.67 (SD 0.51)
14.30 (SD 0.74) 12.61 (SD 0.58)
14.35 (SD 0.70) 12.63 (SD 0.54)
12.67 (SD 0.51)
12.61 (SD 0.58)
12.63 (SD 0.54)
4.02 (SD 2.10) 4.48 (SD 1.93)
4.01 (SD 2.19) 4.47 (SD 2.09)
3.99 (SD 2.06) 4.55 (SD 1.97)
4.48 (SD 1.93)
4.47 (SD 2.09)
4.55 (SD 1.97)
Table II. Pearson correlation coefficients between simplified olecranon and digital methods: 1) first and second assessment separately for all observers together, and 2) overall assessment for each observer group with different levels of experience (p < 0.001 for all coefficients)
Assessment
Group
First Second Overall I – Experienced II – Intermediate III – Novice
Girls
Boys
0.83 0.83 0.83 0.82 0.83 0.83
0.84 0.85 0.84 0.83 0.84 0.85
Table III. Intra-class correlation coefficients (ICC) and 95% confidence interval for both olecranon and digital methods (ICC: part of the total variance that is due to variation between measured effect as assessment or observer
Girls Boys
Assessment effect: simplified olecranon method (%)
Assessment effect: simplified digital method (%)
Observer effect: simplified olecranon method (%)
Observer effect: simplified digital method (%)
0.1 (0 to 0.4) 0.4 (0 to 0.9)
0.2 (0 to 0.5) 0.1 (0 to 0.7)
2.9 (2.2 to 3.6) 3.0 (1.9 to 4.3)
1.1 (0.4 to 1.7) 1.2 (0.2 to 2.3)
both tools. Random-effect regression models were used to complete these usual analyses, taking into account the ‘within and between’ variabilities associated with subject, assessment and observer parameters. These parameters were considered as random-effects in these models contrary to gender, which can be studied as a fixed effect. Intra-class correlation coefficients (ICC) and 95% confidence intervals were considered to evaluate reproducibility of the two methods for each group and between the first and the second review. In this model, the level of observers’ experience was considered as a fixed effect. The tests were two-sided and the significance level was set at 5%. The number of observers was selected on the basis of our previous experience,12 feasibility and simulations of statistical power.20 We calculated that we required at least 20 measurements, with significant set at a p-value < 0.05, to obtain a coefficient correlation r > 0.8, which would indicate a good relationship.21-23
Results A total of 244 radiographs (122 hands and 122 elbows) were reviewed twice by each of the nine observers, for a total of 4392 ratings (2196 hands and 2196 elbows). The mean skeletal ages for the simplified olecranon method in years15 and
simplified digital method graded 1 to 8,12 grouped according to the observers’ experience are shown in Table I. Pearson correlation coefficients between chronological ages and mean skeletal ages (based on the olecranon) were r = 0.60 (p < 0.001) for boys and r = 0.67 (p < 0.001) for girls. When considering the determination of skeletal maturity based on the olecranon and digital methods, overall assessments demonstrated strong correlation between the two methods for both boys (r = 0.84, p < 0.001) and girls (r = 0.83, p < 0.001). An equally important correlation was found between the first and the second review for each gender, without a significant difference (p = 0.93 for boys and p = 0.98 for girls). Additionally, no major discrepancy between the correlation coefficients was found according to the degree of experience in each of the observer’s groups (r = 0.82) (Table II). Table III indicates the intra-class correlation obtained by gender with regard to assessment effect and observer effect for each method. There was an equally strong intra-class correlation (intra-class assessment variability and intra-class observer variability) for each method and for each gender (ICC observer ≤ 3% and ICC assessment ≤ 0.4%). When considering the investigator's experience no significant difference was found for method and gender (p = 0.96). THE BONE & JOINT JOURNAL
A COMPARISON OF THE SIMPLIFIED OLECRANON AND DIGITAL METHODS OF ASSESSMENT OF SKELETAL MATURITY
1559
Fig. 3 Pubertal growth diagram showing detailed use of the olecranon method during the growth spurt and complete use of the digital method during the entire pubertal growth period.
Discussion The Greulich and Pyle Atlas7 and Tanner-Whitehouse III score16 have been found to be reliable methods to assessed skeletal age. However, errors have been reported. Cundy et al24 showed that the interpretation of skeletal age by four radiologists according to the system of Greulich and Pyle,7 differed by more than two years in six of 60 cases. The variation in our study was much smaller for both simplified methods, which might reduce possible errors when assessing skeletal maturity in clinical practice. Our study demonstrates a good intra- and inter-observer reproducibility for each studied method. The correlation between simplified olecranon and digital methods was equally strong for boys and girls. The level of observers’ experience did not influence the accuracy of skeletal maturity assessment during the pubertal growth spurt. The ICC is commonly used to quantify the extent to which observers agree in an assessment. The ICC further evaluates consistency and reproducibility of quantitative measurements made by different observers measuring the same quantity.25 It seems relevant to have lower values of ICC for assessment and observer (random) effects. Indeed, low values indicated that variability of measures for each evaluated method was not explained by assessment and observer effects. It confirms that the part of variability explained by the observer was not important for the two simplified skeletal age methods. The reliability was good and was not influenced by the measure of assessment. It has been shown that the simplified olecranon method is useful for the follow-up of patients with lower limb discrepancies or idiopathic scoliosis during the pubertal growth spurt since it complements Risser stage 0.13 During this phase of accelerated growth, the information obtained from the olecranon epiphyseal lines complements the assessment of the triradiate cartilage, and it helps to idenVOL. 96-B, No. 11, NOVEMBER 2014
tify immature patients (i.e. patients with two ossification nuclei or a half moon-shape nucleus of the olecranon).12-14, As with other methods of assessment of skeletal age based on hand and wrist radiographs,1,4-7 the simplified digital method covers the pre-pubertal growth period and the decelerating growth phase of puberty. It is less detailed than the simplified olecranon method during the two year phase of accelerated growth although, with only stages 3 and 415 represented versus all five stages for the olecranon method.12 However, after the pubertal spurt the simplified olecranon method does not provide any information, whereas the digital method can continue to be used in combination with the Risser stage until the point of skeletal maturity (Fig. 3). The present study has demonstrated good intra- and inter- observer reproducibility for each studied method. The correlation between the simplified olecranon and digital methods was equally strong for boys and girls. The level of observers’ experience did not influence the accuracy of assessment of skeletal maturity. We feel a lateral elbow radiograph is the right choice when assessing skeletal maturity during the pubertal growth spurt. A hand radiograph might be preferable prior to puberty and from Risser grade 1. A combination of both methods adequately covers the gap between elbow fusion and Risser grade 1, as one complements the other. We wish to thank E. Polirsztok, MD, and A. Martucci, MD, for their help with this study. No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. This article was primary edited by E. Moulder and first proof edited by G. Scott.
References 1. Acheson RM. The Oxford method of assessing skeletal maturity. Clin Orthop Relat Res 1957;10:19–39. 2. Pyle SI, Hoerr NL. Radiographic atlas of skeletal development of the knee. A standard of reference. Springfield, USA: Charles C. Thomas, 1955.
1560
F. CANAVESE, Y. P. CHARLES, A. DIMEGLIO, S. SCHULLER, M. ROUSSET, A. SAMBA, B. PEREIRA, J-P. STEIB
3. Hoerr NL, Pyle SI, Francis CC. Radiographic atlas of skeletal development of the foot and ankle. Springfield, USA: Charles C. Thomas, 1962. 4. Tanner JM, Whitehouse RH, Marshall WA, Healy MJR, Goldstein H. Assessment of skeletal maturity and prediction of adult height (TW2-Method). London: Academic Press, 1975. 5. Sempé M, Pavia C. Atlas de la maturation squelettique. Paris: SIMEP, 1979. 6. De Roo T, Schröder HJ. Atlas van de skeletale leeftijd. Dordrecht: Intercontinental Graphics, 1977. 7. Greulich W, Pyle SI. Radiographic atlas of skeletal development of the hand and wrist. 2nd ed. Stanford, CA: Stanford University Press, 1959. 8. Fabry G, De Waele J. The timing of epiphysiodesis. In: De Pablos J, ed. Surgery of the growth plate. Madrid: Ediciones Ergon, 1998:223–228. 9. Little DG, Song KM, Katz D, Herring JA. Relationship of peak height velocity to other maturity indicators in idiopathic scoliosis in girls. J Bone Joint Surg [Am] 2000;82-A:685–693. 10. Duval-Beaupere G, Dubousset J, Queneau P, Grossiord A. A unique theory on the course of scoliosis. Presse Med 1970;78:1141–1146. [in French]. 11. Sauvegrain J, Nahm H, Bronstein N. Etude de la maturation osseuse du coude. Ann Radiol (Paris) 1962;5:542–550. 12. Diméglio A, Charles YP, Daures JP, de Rosa V, Kaboré B. Accuracy of the Sauvegrain method in determining skeletal age during puberty. J Bone Joint Surg [Am] 2005;87-A:1689–1696. 13. Risser JC. The Iliac apophysis; an invaluable sign in the management of scoliosis. Clin Orthop 1958;11:111–119. 14. Dimeglio A. Growth in pediatric orthopaedics. J Pediatr Orthop 2001;21:549–555.
15. Sanders JO, Khoury JG, Kishan S, et al. Predicting scoliosis progression from skeletal maturity: a simplified classification during adolescence. J Bone Joint Surg [Am] 2008;90-A:540–553. 16. Tanner JM, Whitehouse RH, Cameron N, et al. Assessment of skeletal maturity and prediction of adult height (TW3 Method). 3rd ed. London: W.B Saunders, 2001. 17. Charles YP, Diméglio A, Canavese F, Daures JP. Skeletal age assessment from the olecranon for idiopathic scoliosis at Risser grade 0. J Bone Joint Surg [Am] 2007;89-A:2737–2744. 18. Sanders JO, Browne RH, McConnell SJ, et al. Maturity assessment and curve progression in girls with idiopathic scoliosis. J Bone Joint Surg [Am] 2007;89-A:64– 73. 19. Sanders JO, Browne RH, Cooney TE, et al. Correlates of the peak height velocity in girls with idiopathic scoliosis. Spine (Phila Pa 1976) 2006;31:2289–2295. 20. Kottner J, Audigé L, Brorson S, et al. Guidelines for Reporting Reliability and Agreement Studies (GRRAS) were proposed. J Clin Epidemiol 2011;64:96–106. 21. Bonett DG. Sample size requirements for estimating intraclass correlations with desired precision. Stat Med 2002;21:1331–1335. 22. Giraudeau B, Mary JY. Planning a reproducibility study: how many subjects and how many replicates per subject for an expected width of the 95 per cent confidence interval of the intraclass correlation coefficient. Stat Med 2001;20:3205–3214. 23. Shoukri MM, Asyali MH, Donner A. Sample size requirements for the design of reliability study: review and new results. Stat Methods Med Res 2004;13:251–271. 24. Cundy P, Paterson D, Morris L, Foster B. Skeletal age estimation in leg length discrepancy. J Pediatr Orthop 1988;8:513–515. 25. Petrie A, Sabin C. Medical Statistics at a Glance. 3rd ed. Wiley-Blackwell. 2013:119.
THE BONE & JOINT JOURNAL
A comparison of the simplified olecranon and digital methods of assessment of skeletal maturity during the pubertal growth spurt F. Canavese, Y. P. Charles, A. Dimeglio, S. Schuller, M. Rousset, A. Samba, B. Pereira, J-P. Steib From CHU Clermont-Ferrand, Department of Pediatric Surgery, Clermont-Ferrand, France F. Canavese, MD, PhD, Professor, Consultant Paediatric Orthopaedic Surgeon, Professor of Paediatric Orthopaedic Surgery, Chief of Staff M. Rousset, MD, Consultant Paediatric Orthopaedic Surgeon A. Samba, MD, Consultant Paediatric Orthopaedic Surgeon CHU Clermont-Ferrand, Department of Pediatric Surgery, 1 Place Lucie et Raymond Aubrac, 63003 Clermont-Ferrand, France. Y. P. Charles, MD, PhD, Professor, Consultant Orthopaedic Surgeon, Professor of Orthopaedic Surgery S. Schuller, MD, Consultant Orthopaedic Surgeon J-P. Steib , MD, PhD, Professor, Consultant Orthopaedic Surgeon, Professor of Orthopaedic Surgery Hôpitaux Universitaires de Strasbourg, Department of Spine Surgery, 1 Place de l’hôpital, BP 426, 67091 Strasbourg Cedex, France. A. Dimeglio , MD, Professor, Professor of Paediatric Orthopaedic Surgery Université de Montpellier, Faculty of Medicine, 2 Rue de l’école de Médecine, 34060 Montpellier, France. B. Pereira, PhD, Biostatistician CHU Clermont-Ferrand, Biostatistics Unit (DRCI), 58 Rue Montalembert, 63000 ClermontFerrand, France. Correspondence should be sent to Prof Dr F. Canavese; e-mail: [email protected] ©2014 The British Editorial Society of Bone & Joint Surgery doi:10.1302/0301-620X.96B11. 33995 $2.00 Bone Joint J 2014;96-B:1556–60. Received 19 February 2014; Accepted after revision 11 August 2014
1556
Assessment of skeletal age is important in children’s orthopaedics. We compared two simplified methods used in the assessment of skeletal age. Both methods have been described previously with one based on the appearance of the epiphysis at the olecranon and the other on the digital epiphyses. We also investigated the influence of assessor experience on applying these two methods. Our investigation was based on the anteroposterior left hand and lateral elbow radiographs of 44 boys (mean: 14.4; 12.4 to 16.1 ) and 78 girls (mean: 13.0; 11.1 to14.9) obtained during the pubertal growth spurt. A total of nine observers examined the radiographs with the observers assigned to three groups based on their experience (experienced, intermediate and novice). These raters were required to determined skeletal ages twice at six-week intervals. The correlation between the two methods was determined per assessment and per observer groups. Interclass correlation coefficients (ICC) evaluated the reproducibility of the two methods. The overall correlation between the two methods was r = 0.83 for boys and r = 0.84 for girls. The correlation was equal between first and second assessment, and between the observer groups (r ≥ 0.82). There was an equally strong ICC for the assessment effect (ICC ≤ 0.4%) and observer effect (ICC ≤ 3%) for each method. There was no significant (p < 0.05) difference between the levels of experience. The two methods are equally reliable in assessing skeletal maturity. The olecranon method offers detailed information during the pubertal growth spurt, while the digital method is as accurate but less detailed, making it more useful after the pubertal growth spurt once the olecranon has ossified. Cite this article: Bone Joint J 2014;3:1556–60
Skeletal age represents a useful parameter for decision making in paediatric orthopaedics. Different methods are described using hand, elbow, foot, knee or pelvic radiographs for reference.1-7 The first two years of puberty are characterised by a significant growth spurt, followed by three years of gradual growth.8-10 The pubertal growth spurt occurs at around 11 to 13 years of skeletal age in girls and 13 to 15 years in boys.8-10 Sauvegrain, Nahm and Bronstein11 first developed a method to determine skeletal age from anteroposterior and lateral elbow radiographs in 1962. In 2005, Dimeglio et al12 simplified the original Sauvegrain method and demonstrated that morphology of the olecranon apophysis on lateral radiographs goes through five distinct and characteristic appearances, which are easy to distinguish at six-month intervals (Fig. 1).12 This sequence is clinically relevant because the olecranon exhibits a clear and regular morphological development during the pubertal growth spurt,
when the Risser sign is still 0.13 Closure of the epiphyses at the elbow indicates the end of the accelerated growth spurt, when the adolescent is entering the decelerating phase of pubertal growth.14 In 2008, Sanders et al15 presented a new method of digital skeletal age assessment derived from the Tanner-Whitehouse-III method,16 which is based on a radiological analysis of the metacarpals and fingers on anteroposterior hand radiographs. The simplified digital method provides several markers prior to the Risser 1 stage of skeletal maturity, it has been shown by various authors to add helpful information to the Risser stage 013 and covers the period from Risser 1 to Risser 5.14,15,17-20 Stages 1 and 2 correspond to the pre-pubertal period, stages 3 and 4 correspond to the pubertal growth spurt (Risser grade 0), and stages 5 to 8 cover the period from Risser grade 1 to 5, at which stage full skeletal maturity has occurred (Fig. 2). THE BONE & JOINT JOURNAL
A COMPARISON OF THE SIMPLIFIED OLECRANON AND DIGITAL METHODS OF ASSESSMENT OF SKELETAL MATURITY
1557
Fig. 1 Radiological sequence of skeletal age assessment based on the olecranon from 11 to 13 years in girls and 13 to 15 years in boys.
Fig. 2 Radiographs showing stages 3 and 4 (of the digital skeletal maturity assessment15 covering the same growth period as the olecranon sequence.12
Both Dimeglio et al’s12 and Sanders et al’s15 simplified methods of assessing skeletal age have been reported to have a modest learning curve and ease of clinical application. This study compares the accuracy of both methods during the pubertal growth spurt and investigates the accuracy in relation to the experience of the assessor. We also discuss the clinical relevance of each method.
Material and Methods Assessment of skeletal age was made on a set of anteroposterior left hand and lateral elbow radiographs using the simplified methods; Dimeglio et al12 for the elbow and Sanders et al15 for the simplified digital method. The consecutive radiographs had been taken during the pubertal growth period of 44 boys (mean: 14.4 years; 12.4 to 16.1) and 78 girls (mean: 13.0 years; 11.1 to 14.9) in the assessment of conditions such as idiopathic limb-length discrepancy, idiopathic scoliosis or sequelae of epiphyseal trauma of the lower limb. The decision of taking radiographs was made by the paediatric orthopaedic surgeon. Radiographs were taken during the follow-up of children whose standing height VOL. 96-B, No. 11, NOVEMBER 2014
increased by more than 4 cm within the last 6 months.14 The mean chronological ages were 14.4 years (12.4 to16.1; SD 1.3;) for boys and 13.0 years (11.1 to 14.9; SD 1.6) for girls. The children had no history of systemic illnesses or endocrinopathy. In clinical practice we usually further assess the pubertal stage with other factors such as sitting and standing height measurement and the presence of secondary sexual characteristics. In this study, these parameters were not taken into account. Institutional Review Board approval was obtained for this study (CPP 2014/CE09). We assigned nine observers into three groups of three depending on their experience. Group I (expert); more than ten years of experience in paediatric orthopaedic surgery, Group II (intermediate); orthopaedic surgeons with an essentially adult practice with < 20% of their work engaged in managing disorders of the paediatric spine, and group III (novice); orthopaedic residents engaged in paediatric training. All investigators had used the methods previously in paediatric orthopaedic clinics; no supplementary advice or training was given. When assessing the radiographs, all nine observers worked independently and no time limit was imposed. Each observer performed skeletal age determination twice with an interval of six weeks between both grading sessions. The radiographs were reordered between the analyses. Gender was not blinded but name and chronological age was concealed by an individual who did not participate in the review (BP). This data were transferred to a separate file which was not accessible to any of the observers during the analysis. We observed the extent of the correlation between the simplified olecranon and digital methods and further compared both methods according to the level of experience of the three observer groups. Statistical analysis. Statistical analysis was performed using Stata software version 12 (StataCorp, College Station, Texas). Quantitative variables were expressed as mean and standard deviation (SD). To measure the correlation between the two methods, Pearson correlation coefficients were calculated for all patients and for boys and girls separately, according to the observers’ experience. As the two methods had a different data spread, an analysis using Lin concordance coefficients was performed to compare
1558
F. CANAVESE, Y. P. CHARLES, A. DIMEGLIO, S. SCHULLER, M. ROUSSET, A. SAMBA, B. PEREIRA, J-P. STEIB
Table I. Mean olecranon skeletal age assessment (years) and digital score (1 to 8) with standard deviation (SD) grouped according to the experience of the observers
Olecranon simplified method Boys Girls Digital simplified method Boys Girls
Group I: experienced
Group II: intermediate
Group III: novice
Group I: experienced
Group II: intermediate
Group II: novice
14.45 (SD 0.69) 12.67 (SD 0.51)
14.30 (SD 0.74) 12.61 (SD 0.58)
14.35 (SD 0.70) 12.63 (SD 0.54)
12.67 (SD 0.51)
12.61 (SD 0.58)
12.63 (SD 0.54)
4.02 (SD 2.10) 4.48 (SD 1.93)
4.01 (SD 2.19) 4.47 (SD 2.09)
3.99 (SD 2.06) 4.55 (SD 1.97)
4.48 (SD 1.93)
4.47 (SD 2.09)
4.55 (SD 1.97)
Table II. Pearson correlation coefficients between simplified olecranon and digital methods: 1) first and second assessment separately for all observers together, and 2) overall assessment for each observer group with different levels of experience (p < 0.001 for all coefficients)
Assessment
Group
First Second Overall I – Experienced II – Intermediate III – Novice
Girls
Boys
0.83 0.83 0.83 0.82 0.83 0.83
0.84 0.85 0.84 0.83 0.84 0.85
Table III. Intra-class correlation coefficients (ICC) and 95% confidence interval for both olecranon and digital methods (ICC: part of the total variance that is due to variation between measured effect as assessment or observer
Girls Boys
Assessment effect: simplified olecranon method (%)
Assessment effect: simplified digital method (%)
Observer effect: simplified olecranon method (%)
Observer effect: simplified digital method (%)
0.1 (0 to 0.4) 0.4 (0 to 0.9)
0.2 (0 to 0.5) 0.1 (0 to 0.7)
2.9 (2.2 to 3.6) 3.0 (1.9 to 4.3)
1.1 (0.4 to 1.7) 1.2 (0.2 to 2.3)
both tools. Random-effect regression models were used to complete these usual analyses, taking into account the ‘within and between’ variabilities associated with subject, assessment and observer parameters. These parameters were considered as random-effects in these models contrary to gender, which can be studied as a fixed effect. Intra-class correlation coefficients (ICC) and 95% confidence intervals were considered to evaluate reproducibility of the two methods for each group and between the first and the second review. In this model, the level of observers’ experience was considered as a fixed effect. The tests were two-sided and the significance level was set at 5%. The number of observers was selected on the basis of our previous experience,12 feasibility and simulations of statistical power.20 We calculated that we required at least 20 measurements, with significant set at a p-value < 0.05, to obtain a coefficient correlation r > 0.8, which would indicate a good relationship.21-23
Results A total of 244 radiographs (122 hands and 122 elbows) were reviewed twice by each of the nine observers, for a total of 4392 ratings (2196 hands and 2196 elbows). The mean skeletal ages for the simplified olecranon method in years15 and
simplified digital method graded 1 to 8,12 grouped according to the observers’ experience are shown in Table I. Pearson correlation coefficients between chronological ages and mean skeletal ages (based on the olecranon) were r = 0.60 (p < 0.001) for boys and r = 0.67 (p < 0.001) for girls. When considering the determination of skeletal maturity based on the olecranon and digital methods, overall assessments demonstrated strong correlation between the two methods for both boys (r = 0.84, p < 0.001) and girls (r = 0.83, p < 0.001). An equally important correlation was found between the first and the second review for each gender, without a significant difference (p = 0.93 for boys and p = 0.98 for girls). Additionally, no major discrepancy between the correlation coefficients was found according to the degree of experience in each of the observer’s groups (r = 0.82) (Table II). Table III indicates the intra-class correlation obtained by gender with regard to assessment effect and observer effect for each method. There was an equally strong intra-class correlation (intra-class assessment variability and intra-class observer variability) for each method and for each gender (ICC observer ≤ 3% and ICC assessment ≤ 0.4%). When considering the investigator's experience no significant difference was found for method and gender (p = 0.96). THE BONE & JOINT JOURNAL
A COMPARISON OF THE SIMPLIFIED OLECRANON AND DIGITAL METHODS OF ASSESSMENT OF SKELETAL MATURITY
1559
Fig. 3 Pubertal growth diagram showing detailed use of the olecranon method during the growth spurt and complete use of the digital method during the entire pubertal growth period.
Discussion The Greulich and Pyle Atlas7 and Tanner-Whitehouse III score16 have been found to be reliable methods to assessed skeletal age. However, errors have been reported. Cundy et al24 showed that the interpretation of skeletal age by four radiologists according to the system of Greulich and Pyle,7 differed by more than two years in six of 60 cases. The variation in our study was much smaller for both simplified methods, which might reduce possible errors when assessing skeletal maturity in clinical practice. Our study demonstrates a good intra- and inter-observer reproducibility for each studied method. The correlation between simplified olecranon and digital methods was equally strong for boys and girls. The level of observers’ experience did not influence the accuracy of skeletal maturity assessment during the pubertal growth spurt. The ICC is commonly used to quantify the extent to which observers agree in an assessment. The ICC further evaluates consistency and reproducibility of quantitative measurements made by different observers measuring the same quantity.25 It seems relevant to have lower values of ICC for assessment and observer (random) effects. Indeed, low values indicated that variability of measures for each evaluated method was not explained by assessment and observer effects. It confirms that the part of variability explained by the observer was not important for the two simplified skeletal age methods. The reliability was good and was not influenced by the measure of assessment. It has been shown that the simplified olecranon method is useful for the follow-up of patients with lower limb discrepancies or idiopathic scoliosis during the pubertal growth spurt since it complements Risser stage 0.13 During this phase of accelerated growth, the information obtained from the olecranon epiphyseal lines complements the assessment of the triradiate cartilage, and it helps to idenVOL. 96-B, No. 11, NOVEMBER 2014
tify immature patients (i.e. patients with two ossification nuclei or a half moon-shape nucleus of the olecranon).12-14, As with other methods of assessment of skeletal age based on hand and wrist radiographs,1,4-7 the simplified digital method covers the pre-pubertal growth period and the decelerating growth phase of puberty. It is less detailed than the simplified olecranon method during the two year phase of accelerated growth although, with only stages 3 and 415 represented versus all five stages for the olecranon method.12 However, after the pubertal spurt the simplified olecranon method does not provide any information, whereas the digital method can continue to be used in combination with the Risser stage until the point of skeletal maturity (Fig. 3). The present study has demonstrated good intra- and inter- observer reproducibility for each studied method. The correlation between the simplified olecranon and digital methods was equally strong for boys and girls. The level of observers’ experience did not influence the accuracy of assessment of skeletal maturity. We feel a lateral elbow radiograph is the right choice when assessing skeletal maturity during the pubertal growth spurt. A hand radiograph might be preferable prior to puberty and from Risser grade 1. A combination of both methods adequately covers the gap between elbow fusion and Risser grade 1, as one complements the other. We wish to thank E. Polirsztok, MD, and A. Martucci, MD, for their help with this study. No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. This article was primary edited by E. Moulder and first proof edited by G. Scott.
References 1. Acheson RM. The Oxford method of assessing skeletal maturity. Clin Orthop Relat Res 1957;10:19–39. 2. Pyle SI, Hoerr NL. Radiographic atlas of skeletal development of the knee. A standard of reference. Springfield, USA: Charles C. Thomas, 1955.
1560
F. CANAVESE, Y. P. CHARLES, A. DIMEGLIO, S. SCHULLER, M. ROUSSET, A. SAMBA, B. PEREIRA, J-P. STEIB
3. Hoerr NL, Pyle SI, Francis CC. Radiographic atlas of skeletal development of the foot and ankle. Springfield, USA: Charles C. Thomas, 1962. 4. Tanner JM, Whitehouse RH, Marshall WA, Healy MJR, Goldstein H. Assessment of skeletal maturity and prediction of adult height (TW2-Method). London: Academic Press, 1975. 5. Sempé M, Pavia C. Atlas de la maturation squelettique. Paris: SIMEP, 1979. 6. De Roo T, Schröder HJ. Atlas van de skeletale leeftijd. Dordrecht: Intercontinental Graphics, 1977. 7. Greulich W, Pyle SI. Radiographic atlas of skeletal development of the hand and wrist. 2nd ed. Stanford, CA: Stanford University Press, 1959. 8. Fabry G, De Waele J. The timing of epiphysiodesis. In: De Pablos J, ed. Surgery of the growth plate. Madrid: Ediciones Ergon, 1998:223–228. 9. Little DG, Song KM, Katz D, Herring JA. Relationship of peak height velocity to other maturity indicators in idiopathic scoliosis in girls. J Bone Joint Surg [Am] 2000;82-A:685–693. 10. Duval-Beaupere G, Dubousset J, Queneau P, Grossiord A. A unique theory on the course of scoliosis. Presse Med 1970;78:1141–1146. [in French]. 11. Sauvegrain J, Nahm H, Bronstein N. Etude de la maturation osseuse du coude. Ann Radiol (Paris) 1962;5:542–550. 12. Diméglio A, Charles YP, Daures JP, de Rosa V, Kaboré B. Accuracy of the Sauvegrain method in determining skeletal age during puberty. J Bone Joint Surg [Am] 2005;87-A:1689–1696. 13. Risser JC. The Iliac apophysis; an invaluable sign in the management of scoliosis. Clin Orthop 1958;11:111–119. 14. Dimeglio A. Growth in pediatric orthopaedics. J Pediatr Orthop 2001;21:549–555.
15. Sanders JO, Khoury JG, Kishan S, et al. Predicting scoliosis progression from skeletal maturity: a simplified classification during adolescence. J Bone Joint Surg [Am] 2008;90-A:540–553. 16. Tanner JM, Whitehouse RH, Cameron N, et al. Assessment of skeletal maturity and prediction of adult height (TW3 Method). 3rd ed. London: W.B Saunders, 2001. 17. Charles YP, Diméglio A, Canavese F, Daures JP. Skeletal age assessment from the olecranon for idiopathic scoliosis at Risser grade 0. J Bone Joint Surg [Am] 2007;89-A:2737–2744. 18. Sanders JO, Browne RH, McConnell SJ, et al. Maturity assessment and curve progression in girls with idiopathic scoliosis. J Bone Joint Surg [Am] 2007;89-A:64– 73. 19. Sanders JO, Browne RH, Cooney TE, et al. Correlates of the peak height velocity in girls with idiopathic scoliosis. Spine (Phila Pa 1976) 2006;31:2289–2295. 20. Kottner J, Audigé L, Brorson S, et al. Guidelines for Reporting Reliability and Agreement Studies (GRRAS) were proposed. J Clin Epidemiol 2011;64:96–106. 21. Bonett DG. Sample size requirements for estimating intraclass correlations with desired precision. Stat Med 2002;21:1331–1335. 22. Giraudeau B, Mary JY. Planning a reproducibility study: how many subjects and how many replicates per subject for an expected width of the 95 per cent confidence interval of the intraclass correlation coefficient. Stat Med 2001;20:3205–3214. 23. Shoukri MM, Asyali MH, Donner A. Sample size requirements for the design of reliability study: review and new results. Stat Methods Med Res 2004;13:251–271. 24. Cundy P, Paterson D, Morris L, Foster B. Skeletal age estimation in leg length discrepancy. J Pediatr Orthop 1988;8:513–515. 25. Petrie A, Sabin C. Medical Statistics at a Glance. 3rd ed. Wiley-Blackwell. 2013:119.
THE BONE & JOINT JOURNAL
