The aim of treatment of Legg-Calvé-Perthes disease is to prevent secondary degenerative arthritis of the hip by trying to ensure that the femoral head does not become distorted during the course of the disease. The outcome of treatment of Legg-Calvé-Perthes disease traditionally has been evaluated by defining the shape of the femoral head1-3. As alterations in the size and shape of the femoral head in early childhood can influence the growth of the acetabulum, some authors have also taken into account the configuration of the acetabulum as an outcome measure4. Currently, the most widely used system for the classification of outcome of Legg-Calvé-Perthes disease is that of Stulberg et al., who allocated the affected hip into one of five classes4. The classification system has merit in that it includes the shape and size of the femoral head, the shape and dimensions of the acetabulum, and the congruency between the femoral head and the acetabulum, all of which can influence the onset of secondary degenerative arthritis. Those authors showed that their classification system is valid as it is predictive of the likelihood of late degenerative arthritis4. However, the classification system is not entirely satisfactory as it is not highly reproducible5, and it has the major limitation of being a categorical qualitative measure of outcome rather than a continuous measure. The reproducibility of the classification system improves if Class I and II are pooled and Classes IV to V are pooled to give a three-group classification6, but this is likely to make the classification less discriminative.
From the early part of the last century, attempts have been made to quantitatively measure the extent of deformation of the femoral head, both during the active stage of the disease and after healing3,7-12, and indices such as the epiphyseal index7 and the epiphyseal quotient8 were among the first to be described. Several authors only measured the deformation on anteroposterior radiographs and did not pay attention to the findings on lateral radiographs. The inherent limitation of this approach is that the configuration of the femoral head can be grossly different on the two views, as shown in the radiographs in the Appendix.
The present study was undertaken with the aim of devising and evaluating the efficacy of a method of quantifying the shape and size of the femoral head and the femoral-acetabular relationship while taking into consideration both anteroposterior and lateral radiographic views of the hip of skeletally mature individuals with healed Legg-Calvé-Perthes disease.
Engineers have quantified the lack of roundness of objects and expressed the degree of deviation from circularity as roundness error13. We explored the possibility of using these principles to achieve our aim.
Computer images of anteroposterior and lateral radiographs of both hips of 121 skeletally mature individuals with healed unilateral Legg-Calvé-Perthes disease that had been treated in our center were identified. The sample included some children who had been managed surgically and some who had been managed symptomatically.
Qualitative Outcome Assessment
The outcome for each hip was classified according to the system of Stulberg et al.4 independently on two separate occasions by five investigators before any of the radiographic measurements were made (see Appendix). The weighted kappa value was calculated to estimate intraobserver and interobserver agreement.
To test associations between radiographic measurements and the Stulberg classification, each hip was designated as a specific Stulberg class if at least three of the five investigators concurred and the other investigators differed by one class only. One hundred and twelve of the 121 hips could be allocated to a Stulberg class on the basis of this criterion.
Radiographic Measurements
Measurements were made for the 121 affected hips and the contralateral, normal hips by four investigators independently with use of Digimizer image-analysis software (version 4.1.1.0; MedCalc Software, Mariakerke, Belgium) as outlined below. The interobserver reproducibility of measurement was assessed by calculating the intraclass correlation coefficient (ICC). A subset of thirty radiographs was measured twice by each investigator, with an interval of two weeks between measurements, and the ICC was calculated to assess intraobserver agreement.
Measurement of Sphericity of the Femoral Head
On the anteroposterior radiograph, two reference points were marked on the medial and lateral margins of the femoral head (Fig. 1-A) and a circle touching these reference points was drawn to match the size of the femoral head. If the circle perfectly fitted the contour of the articular margin, the radius of the circle (r ap) was noted. If the articular margin did not conform to the arc of the circle, the size of the circle was adjusted so that it just touched the reference points and the articular margin without extending outside the femoral head (maximum inscribed circle [MIC]) (Fig. 1-B). Another concentric circle was drawn to just touch the outer limits of the articular margin without extending inside the femoral head (minimum circumscribed circle [MCC]) (Fig. 1-C). The radii of these circles (r MIC-ap and r MCC-ap) were noted (Fig. 1-D). The same measurements were performed on the Lauenstein lateral radiograph. The roundness errors, which are the differences in the radii of MIC and MCC in both views, were expressed as ratios as follows:
Roundness error (RE) on the anteroposterior radiograph:
Roundness error on the lateral radiograph:
Ellipsoid deformation of the head (ED) was considered to be present when the radius of the femoral head on the anteroposterior radiograph differed from the radius of the femoral head on the lateral radiograph. ED was computed with use of the following formulae:
Ellipsoid deformation of the femoral head:
If r ap was greater than r lat, the formula was:
The extent to which the shape of the femoral head deviated from sphericity was the sum of the roundness errors on the anteroposterior and lateral views and the ellipsoid deformation (Sphericity Deviation Score [SDS] = RE AP + RE Lat + ED).
Measurement of Congruency of the Femoral Head and the Acetabulum
The arc of the femoral articular surface that was parallel to the weight-bearing region of the acetabular roof (extending from the lateral extremity of the condensed subchondral bone of the sourcil to the junction of the roof and acetabular fossa14) was measured on both the anteroposterior and lateral views of the normal and affected hips (Figs. 2-A and 2-B). The extent of congruency was expressed as the sum of the arcs on the anteroposterior and lateral views and was designated as the composite femoral congruency arc (CFCA).
Measurement of Extent of Femoral Head Enlargement (Coxa Magna)
The extent of enlargement of the femoral head was computed by comparing the radius of the femoral head of the diseased hip with that of the contralateral, normal femoral head (r normal) with use of the following formula:
Extent of femoral head enlargement (FHE):
Measure of Femoral Neck Length and Trochanteric Growth
The articulotrochanteric distance (ATD)2 was measured on the anteroposterior radiograph and was compared with the ATD of the normal hip. The degree of growth aberration of the femoral neck was expressed as a ratio of the difference in the ATD values:
Femoral neck growth inhibition (FNGI): 
Measurement of Acetabular Configuration
The Sharp angle15 was measured on both sides on the anteroposterior radiograph.
Quantitative Outcome Assessment Applied to Stulberg Classification
The values of the sphericity deviation score, extent of femoral head enlargement, composite femoral congruency arc, femoral neck growth inhibition, and Sharp angle for each Stulberg class were calculated.
Statistical Methods
Statistical analysis was performed with the Statistical Package for Social Sciences (SPSS, version 16; IBM, Chicago, Illinois). For the purposes of analysis, the mean values for the four investigators were computed for all measurements for each hip, and these mean values were taken as the dimensions of the hips. The mean, standard deviation, and 95% confidence interval (CI) for each measurement were calculated for hips that fell into each of the five Stulberg classes. Pearson correlation coefficients were calculated, and regression lines were plotted to test associations between the various outcome measurements. Analysis of variance and post hoc tests were used to determine the association of these measurements with the Stulberg classes of outcome. The level of significance was set at p < 0.05 for all analyses.
Source of Funding
No external funding was obtained for this study. Grants to support one of the authors (S.T.) on a fellowship were not utilized for the study.
At the outset, it is necessary to justify the need for making measurements on radiographs to determine the quality of the outcome of treatment of Legg-Calvé-Perthes disease. As the primary aim of treatment of Legg-Calvé-Perthes disease is to retain the sphericity of the femoral head, it is imperative that the shape of the femoral head is defined accurately when evaluating the final outcome. A recent study demonstrated that the reliability of visual assessment of femoral head sphericity is not satisfactory, with interobserver kappa values of 0.46 and 0.44 for anteroposterior and frog-leg lateral radiographs, respectively16. This finding highlights the need for a more reproducible method of assessment of sphericity. The second and more important need for making these measurements is to devise reliable and reproducible continuous outcome measures for Legg-Calvé-Perthes disease.
The optimum outcome of treatment of Legg-Calvé-Perthes disease is a femoral head that is spherical, is the same size as that in the uninvolved hip (or minimally enlarged), and is congruent with the weight-bearing part of the acetabulum through a sufficiently large arc of motion in different planes. The quantitative outcome measures described in the present study can reliably identify hips with these attributes as the sphericity deviation score, extent of femoral head enlargement, and the composite femoral congruency arc measure these variables.
The criterion validity of these measurements was confirmed by the finding that each of these measurements was clearly different for hips in the different Stulberg classes. The sphericity deviation score can very accurately differentiate hips that are spherical from those that are not, making it possible to distinguish hips that are Stulberg Class I and II from hips that are Class III, IV, or V. This is of practical relevance as good outcomes in Legg-Calvé-Perthes disease only include Stulberg Classes I and II, whereas all other classes are poor outcomes.
One of the criteria used by Stulberg et al. to distinguish Class-IV from Class-V hips was to assess whether the femoral head and acetabulum are congruent4. If the femoral head was aspherical but was congruent with the acetabulum, they classified the hip as Class IV, but if the hip was incongruent, they classified it as Class V. Rather than classifying the hips as being congruent or incongruent, we attempted to quantify the degree of congruency and demonstrated that it is possible to do so. The measure of congruency (the composite femoral congruency arc) was highest for Stulberg Class-I hips and lowest for Class-V hips, and there was a significant inverse correlation between the sphericity deviation score and the composite femoral congruency arc, indicating that the more aspherical the femoral head, the less congruous the hip. It also may have been useful to assess congruity on weight-bearing radiographs to get an impression of the true functional relationship between the femur and the acetabulum; however, we did not have access to weight-bearing radiographs.
All of the outcome scores described in the study are ratios rather than absolute measurements, which overcomes the problem of variations in magnification. However, because the values of the normal side are used for some of the scores, they can only be applied for unilateral cases. However, the most important score, the sphericity deviation score, can be used for both unilateral and bilateral cases as it is computed without having to take dimensions of the contralateral hip into consideration.
The engineering principles of measuring deviation from circularity or roundness error have been applied in the past to quantify deformation of the femoral head following developmental dysplasia of the hip17 and Legg-Calvé-Perthes disease3. Those authors used the best-fit circle method, which is not easy to apply. Furthermore, they had to digitize radiographic images by tracing them on a digitizing tablet, which is laborious and is prone to some degree of error due to inaccuracies of tracing. These methods of measuring femoral head deformity have not become popular.
With radiographic images being more widely stored and displayed on computer-based Picture Archiving and Communication Systems (PACS), the use of additional image-analysis software for quantitative measurements would not be difficult. Any one of several commercially available image-analysis software programs may be used for these measurements. Image-analysis programs dedicated to orthopaedics can easily incorporate measurements such as the ones used in the present study.
The rationale of having done several different measurements in the present study is that the sequelae of Legg-Calvé-Perthes disease include femoral head deformation, femoral head enlargement, coxa brevis with trochanteric overgrowth, acetabular dysplasia, and joint incongruity; we included separate measurements to quantify each of these sequelae.
The utility of these measurements for predicting prognosis in a given patient needs to be defined. On the basis of the 95% confidence intervals in Table II, it appears that a hip with a sphericity deviation score of <5, a femoral head enlargement value of <9%, and a composite femoral congruency arc of >375° would in all probability be graded as Stulberg Class I or II. It also would be possible to make a composite score that incorporates each of the measurements, but as there is a very strong correlation between each of these measurements (Table III), it may be unnecessary to compute such a composite score. We believe that the two most crucial of all of the sequelae of Legg-Calvé-Perthes disease are femoral head deformation and loss of joint congruity, and hence it needs to be seen if either of these two measures alone would be sufficient to accurately separate poor and good outcomes. As there is such a good correlation between sphericity deviation score and the composite femoral congruency arc, we suggest that measuring sphericity deviation score alone may be sufficient to distinguish between good and poor outcomes.
It may be argued that sufficient information cannot be obtained from radiographs alone and that computed tomographic scanning or magnetic resonance imaging may be needed to confirm subtle alterations in shape and phenomena like femoroacetabular impingement. However, radiography is likely to remain the method of evaluation for some time to come.
Finally, the study confirms that quantitative measures of the outcome of Legg-Calvé-Perthes disease are reliable and simple to apply, which likely will have important implications when planning future studies. Continuous outcome measures described in the present study such as sphericity deviation score should facilitate studies on the results of interventions for Legg-Calvé-Perthes disease with more practicable sample sizes than those that are required with discrete outcome measures like the Stulberg classification.
Note: The clinical fellowship of Dr. Stéphane Tercier was supported by Swiss grants (Fonds du département médico-chirurgical de Pédiatrie et de Perfectionnement du CHUV, Fondation SICPA, Fondation de l’Hôpital Orthopédique de Lausanne et Fonds de la Société Suisse d’Orthopédie et Traumatologie).
Disclosure: None of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of any aspect of this work. None of the authors, or their institution(s), have had any financial relationship, in the thirty-six months prior to submission of this work, with any entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. Also, no author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.