The present study included the same population as our previous study1, in which a cohort of twenty-two girls with idiopathic scoliosis was followed through their growth spurt with use of serial spine radiographs, skeletal age radiographs, and a number of clinical and biochemical markers of maturation, all of which were assessed every six months. Patients underwent scoliosis brace treatment according to accepted criteria (initiation of bracing for curves of =25° or for curves of 20° to 24° with 5° of documented progression in patients with a Risser stage3 of =2). The key findings of the various skeletal maturity stages from the Tanner-Whitehouse-III RUS descriptions were developed into a radiographic skeletal maturity classification system based on when the individual bone changes occurred in relation to the curve acceleration phase in girls with idiopathic scoliosis. It was found that, with increasing age, the various bones underwent changes in an orderly sequence compared with each other. Specifically, the digits on the ulnar side of the hand (the fourth and fifth digits) are the last to have fully covered epiphyses (Tanner-Whitehouse-III stage F), the proximal epiphyses cap their metaphyses slightly before the distal epiphyses do (stage G), the distal phalangeal physes close before the proximal and middle phalangeal physes do (stages H and I), the digital physes close before the metacarpal physes do, and the distal radial physis closes last. As all of the digits are available on the skeletal age anteroposterior radiograph of the hand and wrists, there was no need to restrict the evaluation to the first, third, and fifth digits as is done in the Tanner-Whitehouse-III method. The modified staging system underwent several iterations to clarify the descriptions of the stages, and the specific Tanner-Whitehouse-III descriptions were then related to specific Greulich and Pyle atlas4 radiographs (Table I). All iterations underwent similar testing with use of different radiographs from the same subject cohort. A detailed description of the method relating the specific Tanner-Whitehouse-III descriptors and references to Greulich and Pyle atlas findings, a self-instructional PowerPoint presentation, and a self-assessment examination with detailed explanations were developed and provided to six reviewers. The reviewers included the originator of the current simplified method (J.O.S.), three additional practicing pediatric orthopaedic surgeons (J.G.K., S.K., J.F.M. III), one radiologist (J.A.B.), and one third-year orthopaedic surgery resident (K.D.A.).
Detailed Description of the Method
We divided the skeletal maturity stages in our classification system in a manner corresponding with typical maturity stages. The infantile rapid stage (stage 0) was not studied in this series. The juvenile slow stage (stage 1) is the stage before the beginning of the adolescent growth spurt. Secondary sexual characteristics are in Tanner stage 15. The preadolescent slow stage (stage 2) is the stage during which the adolescent growth spurt has started but before the peak height velocity has occurred. Secondary sexual characteristics are in Tanner stage 2. The adolescent rapid stage—early (stage 3) is the stage in which the scoliosis curves start the curve acceleration phase, corresponding with the time of the peak height velocity. Girls may be in Tanner stage 2 or 3. In the adolescent rapid stage—late (stage 4), scoliosis curves continue to increase rapidly. Girls are typically in Tanner stage 3 but are still usually in Risser stage 0, with an open triradiate cartilage of the acetabulum. In the adolescent steady progression stage—early (stage 5), girls are still typically in Risser stage 0, but at this stage the triradiate cartilage is closed. Menarche rarely occurs before this stage. During the adolescent steady progression stage—late (stage 6), the Risser sign is usually positive. Menarche usually has occurred. The early mature stage (stage 7) usually corresponds to Risser stage 4, and scoliosis progression can still occur. The mature stage (stage 8), which corresponds to Risser stage 5 and completion of growth, was not studied.
Radiographic Features of the Stages
Unlike the Tanner-Whitehouse-III method, which uses just the first, third, and fifth digits, this simplified method takes advantage of the natural concordance of all of the digits to provide every possible clue for accurate maturity staging. The Tanner-Whitehouse-III descriptors for the third and fifth digits should also be used for the second and fourth digits. Because skeletal maturity occurs as a continuum rather than as discrete stages, every effort has been made to describe the beginning and end of each stage, which is essential for the classification's reliability. If a patient has not fully reached the next radiographic stage, he or she is still considered to be in the less mature stage.
Individual Stages
Stage 1: Juvenile Slow Stage
The key feature of this stage is that all of the digital epiphyses are not covered (Figs. 1-A and 1-B). This stage lasts until all of the epiphyses are covered. The term "covered" means that the epiphysis is as wide as the metaphysis (Tanner-Whitehouse-III stage F) (see Appendix). Once the epiphyses are fully covered, the patient has entered the next phase.
The best place to look is often the middle phalanx of the fifth digit because the epiphysis often seems to be the last to be covered.
This stage can be compared with Greulich and Pyle atlas Female Standard 17 (skeletal age, eight years and ten months), which shows that the epiphyses of the middle and proximal phalanges of the fifth digit and those of all of the metacarpals are not covered. This is similar to Male Standard 22 (skeletal age, twelve years and six months).
Stage 2: Preadolescent Slow Stage
The key feature of this stage is that all digital epiphyses are covered (Figs. 2-A and 2-B). This stage lasts until all of the epiphyses cap their metaphyses (Tanner-Whitehouse-III stage G) (see Appendix).
The metacarpal dorsal and volar surfaces are clearly delineated; this finding corresponds with "covered" epiphyses.
This stage can be compared with Greulich and Pyle atlas Female Standard 18 (skeletal age, ten years), which shows all of the epiphyses covering their metaphyses. This is similar to a male skeletal age of thirteen years, at which there is early epiphyseal capping in some, but not in all, of the digits.
Stage 3: Adolescent Rapid Stage—Early
The key feature of this stage is that the preponderance of the epiphyses cap their metaphyses (Tanner-Whitehouse-III stage G) (see Appendix) (Figs. 3-A and 3-B). Epiphyses normally correspond with each other. If all but a couple of the epiphyses are capped, then the radiograph falls into this stage.
This is the crucial stage to identify, and it is important to familiarize oneself with all of the Tanner-Whitehouse-III stage-G criteria (see Appendix). Certain features should be seen. Specifically, in this stage, the middle and proximal phalangeal and thumb metacarpal epiphyses are capped. In addition, the heads of the second through fifth metacarpals are wider than their metaphyses and have clear dorsal and volar surfaces.
There are some important nuances. Specifically, it can be difficult to identify capping on the distal phalanges, particularly those of the thumb, which is usually rotated. If the distal phalanges cannot be seen well, they should not be used. However, if the distal phalanges are clearly capped, then the other phalanges probably are capped as well, and the patient is in this stage.
The Greulich and Pyle atlas Female Standards 19 and 20 (skeletal age, eleven and twelve years, respectively) and Male Standards 24 and 25 (skeletal age, thirteen years and six months and fourteen years, respectively) correspond with this stage.
Stage 4: Adolescent Rapid Stage—Late
The key feature of this stage is that one or more of the distal phalanges are beginning to demonstrate physeal closure (Figs. 4-A and 4-B). All of the features of the early adolescent rapid stage are present, except that one or more of the distal phalanges are clearly beginning to demonstrate physeal closure (Tanner-Whitehouse-III stage H). Stage H means that a portion but not the entire physis is clearly closed. A narrowed physis without partial closure is not in this stage (see Appendix).
If the distal phalanx of the thumb is clearly visible, it may be used. Often, it is too rotated to see clearly and should not be used as the sole criterion for this stage.
These features are best seen on Greulich and Pyle atlas Male Standard 15 (skeletal age, fifteen years) for the fourth and fifth digits. The Female Standard 21 (skeletal age, thirteen years) shows these features for the second, third, and fourth digits but overall is not in this stage because the thumb has a closed distal physis and the fifth digit is cut off from the radiograph.
Stage 5: Adolescent Steady Progression Stage—Early
The key feature of this stage is that all distal phalangeal physes are closed (Tanner-Whitehouse-III stage I) whereas the proximal phalangeal physes are open (Figs. 5-A and 5-B). Tanner-Whitehouse-III stage I indicates there is a white (not black) physeal scar (see Appendix). The first metacarpal physis closes at about the same time or a little later than the distal phalangeal physes do. This stage lasts until either the middle or proximal phalangeal physes are clearly closing (Tanner-Whitehouse-III stage H).
The Greulich and Pyle atlas Female Standard 22 (skeletal age, thirteen years and six months) and Male Standard 27 (skeletal age, fifteen years and six months) are in this stage.
Stage 6: Adolescent Steady Progression Stage—Late
The key feature of this stage is that the proximal and middle phalangeal physes are clearly closing (Tanner-Whitehouse-III stage H) (Figs. 6-A and 6-B). This stage lasts until the phalangeal and metacarpal epiphyses are closed. The finger metacarpals may have a prolonged physeal scar. Please note that a fully white physeal scar is closed (stage I), whereas any remaining black physis is not fully closed (stage II or earlier).
The Greulich and Pyle atlas Female Standard 23 (skeletal age, fourteen years) and Male Standard 28 (skeletal age, sixteen years) correspond with this stage.
Stage 7: Early Mature Stage
The key feature of this stage is that only the distal radial physis is open; the ulna is disregarded (Figs. 7-A and 7-B). All of the digital physes are closed (no black remains, corresponding with Tanner-Whitehouse-III stage I) (see Appendix).
The distal radial physis remains open (Tanner-Whitehouse-III stage G or H, i.e., some black remains).
The metacarpal white physeal scars may persist. If a black physis remains, the patient is still in stage 6.
Greulich and Pyle atlas Female Standard 24 (skeletal age, fifteen years) and Male Standard 29 (skeletal age, seventeen years) correspond with this stage.
Stage 8: Mature Stage
The key feature of this stage is that the distal radial physis is completely closed, i.e., no black remains (Tanner-Whitehouse-III stage I) (Figs. 8-A and 8-B) (see Appendix). The Greulich and Pyle atlas Female Standard 26 (skeletal age, seventeen years) and Male Standard 31 (skeletal age, nineteen years) correspond with this stage.
Reliability Testing
Thirty skeletal age radiographs, representing the maturity spectrum of the subjects followed through their growth spurt, were selected independently of the reviewers, digitized, and blinded. For the entire cohort, there were a total of 161 skeletal age radiographs, with an average of 7.3 radiographs (range, four to eleven radiographs) per subject. As the radiographs were randomly selected across the spectrum, the number of radiographs from a single subject varied for the different trials. No subject had more than two radiographs in any particular trial. The reviewers were instructed to view the radiographs on a high-resolution computer monitor (at least 1024 × 768 pixels) and to refer to the reference materials. Two batches of the radiographs, with separate numbering systems, were reviewed no sooner than twenty-four hours apart to determine interobserver and intraobserver reliability. An additional interobserver variability test was performed with a separate selection of radiographs and an atlas to identify whether use of a specific atlas improved interobserver reliability.
Validity Testing
All 161 skeletal age radiographs from our previous study1 were then classified according to the simplified skeletal maturity staging system and were compared with the timing of the curve acceleration phase. The magnitude of the main scoliotic curve at each skeletal maturity stage was then related to the final curve magnitude and the Lenke curve type1,6. Correlation of the stages with the curve acceleration phase was compared with the values from our previous study1.
Statistical Methods
All statistical analyses, including statistical analysis of the exact matches and determination of the weighted and unweighted kappa coefficients, were performed with use of SAS software (version 9.1; SAS Institute, Cary, North Carolina). Pearson correlation coefficients were computed with use of linear regression for the maturity scores compared with time relative to the curve acceleration phase. Curve-fitting was performed with use of TableCurve 2D (version 5.01; SPSS, Chicago, Illinois).
Logistic regression analysis of curve magnitude and maturity stage to predict the probability of the final curve being >50° was done for the most common curve types (Lenke types 1 and 3) and separately for all other curve types. To estimate the probability of the final curve being >50°, we first computed the logistic regression equation and then used that equation to calculate the probability for all combinations of curve magnitude, maturity stage, and Lenke curve type.
Reasons for the Method
Previous work identified the DSA scores from the Tanner-Whitehouse-III RUS method as the maturity indicator most closely correlated with curve behavior in a cohort of girls with idiopathic scoliosis. The DSA score, however, is cumbersome to obtain and is more appropriate for a research study than for clinical use unless reliable, rapid computerized measurements can be obtained, as suggested by several investigators8-11. Because of these limitations, we sought to develop a reliable but rapid skeletal maturity assessment system based on the findings used to determine the DSA score. Like any maturity indicator, skeletal maturity assessment can be problematic9. Skeletal maturity is a continuum and the changes over time can be subtle, yet clinical practice requires precise determination. In their description of the requirements of an orthopaedic classification system, Garbuz et al.7 indicated that a good classification system should be reliable as evidenced by strong kappa values for both intraobserver and interobserver reliability through testing that is balanced across the spectrum of the classification. Reliability testing demonstrated that the skeletal maturity system developed here is highly reliable as a clinical staging system, and our experience indicates that only a few seconds are required to classify the maturity once the system is learned in detail.
Learning the Method
Users should familiarize themselves with the system by reviewing the detailed description of the method, viewing the presentation, performing the self assessment, and referring to the table as needed. Whether or not to have standardized radiographs available for use during the grading will depend on the reader's experience. Both the figures in this article and the referenced plates in the Greulich and Pyle atlas demonstrate the findings at each stage. Readers of the radiographs were very consistent with themselves when just using the descriptions and the Greulich and Pyle atlas as needed, and a specific atlas for this new method did not improve the reliability of the readings. We believe that this was because the Greulich and Pyle atlas provides excellent figures illustrating the various stages, and a specific atlas does not improve on what can be found in the more readily available Greulich and Pyle atlas. Also, once the reader is comfortable with the method, the Tanner-Whitehouse-III descriptors are very clear and do not require access to any radiographic atlas. Similar descriptors are found in the section on the Greulich and Pyle atlas describing the maturity indicators of individual bones.
When possible, a patient's skeletal maturity radiographs should be viewed in sequence rather than as isolated measurements because, as Tanner et al. noted3, the changes are more obvious sequentially. While this was not possible for the test-retest, it is often possible in clinical practice and improves the assessments. We did find that there was a learning curve for the method. The orthopaedic resident in particular had substantial scoring improvements in the early iterations as compared with the other reviewers. While the kappa values were stronger for the more experienced reviewers, they remained good for the least experienced as well.
The Greulich and Pyle Atlas
While the Greulich and Pyle atlas provides excellent radiographs illustrating the various stages and is widely available, it must be used with caution. The atlas was developed from representative radiographs at various chronological ages in an era before timing relative to the peak height velocity, or even secondary sexual characteristics, was recognized as an important, quantifiable maturity indicator. Maturity near adolescence is better based on timing relative to the peak height velocity as normal adolescents have as much as a four-year variation in the timing of their growth spurt12. We concur with Tanner et al.3 that the concept of skeletal maturity is more appropriate than skeletal age, and we prefer a system that does not have a specific "age" attached to the radiograph. Bones do not have a different age than patients, but their maturities differ just as children's do. Because of this, the time-intervals of the standards in the Greulich and Pyle atlas are problematic during adolescence, as noted by DiMéglio et al.13. In the present series, the mode of stage-3 six-month visits was two (range, one to three) six-month visits, which is substantially shorter than the Greulich and Pyle atlas would suggest. We found that the maturity stages, which are separated by about a year in the Greulich and Pyle atlas, actually occur much more rapidly in the middle stages. In our simplified skeletal maturity system, stage 3 is the most important phase because of its tight correlation with the curve acceleration phase. This stage corresponds with the Greulich and Pyle female skeletal ages of both eleven and twelve years because of the atlas's over-reliance on the carpal bones. The simplified staging system does not use the carpal bones because our previous work demonstrated that the carpus matured before CAP 0. Another problem with the Greulich and Pyle atlas is that the descriptors were designed to highlight many subtleties rather than being designed for reliability. The Tanner-Whitehouse-III classification system avoids the subtleties in favor of clarity. Our modification is therefore based on simple but precise findings to enhance both usability and reliability.
Correlation of the Stages with Scoliosis Curve Behavior
In addition to reliability, a useful maturity staging system for scoliosis must correlate with curve behavior. Similar to the DSA scores from which they are derived, the stages of this system are highly correlated with the timing relative to the curve acceleration phase (Pearson r = 0.91). This correlation is stronger than those associated with either the Risser sign or the Greulich and Pyle atlas1. The correlation is even stronger using an S-shaped curve fit (a fourth-order polynomial [r = 0.94, r2 = 0.88]), reflecting the rapid skeletal maturity at the middle stages (stages 4 through 6) and the slower maturity changes at either end.
Determining Curve Prognosis
Despite the relatively small number of patients in the present study, the curve magnitude from stage 2 onward was highly prognostic of the eventual curve magnitude. This was particularly true for the Lenke type-1 curves (main thoracic curves) and Lenke type-3 curves (double curves with the thoracic curve being larger than the lumbar curve)6, for which the numbers were sufficiently large for analysis. In order to have estimates of the probability of the final curve being >50° within a 10% confidence interval, approximately thirty-six subjects with each curve pattern need to be followed through the stages to maturity to provide accurate prognosis for each curve pattern, maturity stage, and curve magnitude. From Table III, it is clear that those at greatest risk are patients with curves of =20° at stage 2 and of =30° at stage 3, for whom bracing is unlikely to succeed. The curve magnitude of >50° was chosen because it often represents a minimum threshold for surgery. The same procedure can be done for any threshold value, whether 45°, 60°, or ultimate curve size at maturity. Because the purpose of the present study was to identify proper maturity markers in patients with idiopathic scoliosis, and because commonly accepted bracing criteria were used, the present study should not be viewed as a natural history study but as a prospective study of maturity and curve progression typical of patients with similar maturity and curve sizes seen and treated in many North American clinical practices. While the design of the current study makes it possible to identify bracing failures, it cannot identify bracing successes. It remains important to further test these findings in both untreated patients and compliant patients who have been managed with bracing for each specific curve pattern and magnitude in a larger prospective study. As we do not know how our results were affected by bracing, specific recommendations regarding bracing must await the completion of a well-designed bracing study that is properly stratified, similar to the current study, according to curve magnitude, pattern, and skeletal maturity.
The simplified Tanner-Whitehouse-III system for the classification of skeletal maturity for patients with idiopathic scoliosis is rapid and reliable, and it correlates highly with curve behavior. It also has the advantages of the general familiarity with hand skeletal maturity, reference to the ubiquitous Greulich and Pyle atlas, and the avoidance of a complex scoring system. All of these qualities appear to make it useful as a clinical maturity indicator in patients with idiopathic scoliosis. In addition, the probability estimates of progression to surgery can assist clinicians in discussing the likely outcome of scoliosis management with the patient and parents.