Orthopedics

Feature Article 

An Abbreviated Scale for the Assessment of Skeletal Bone Age Using Radiographs of the Knee

Xin Tang, MD; Yubo Lu, MD; Mingfan Pang, MSPH; Derek T. Nhan, BS; Walter Klyce, BA; Jan Fritz, MD; R. Jay Lee, MD

Abstract

Hand and wrist radiographs are the most common means for estimating skeletal bone age. There is no widely used scale for estimating skeletal bone age using knee radiographs. Do skeletal bone age estimates from knee-maturity scales correlate sufficiently with both chronologic age and estimates from a hand-wrist scale to potentially substitute for estimates from the latter? The authors reviewed the records of 503 patients 6 to 19 years old who had hand and knee radiographs obtained within 30 days of each other. They analyzed radiographs using the O'Connor knee scale (based on 10 maturation markers) and a new, abbreviated version of the O'Connor scale (based on 7 markers). The authors also analyzed radiographs of the hands of boys 12.5 to 16 years old and girls 10 to 16 years old using the shorthand method. Multivariate linear regression was used for analysis. Inter- and intrarater reliabilities were assessed. Skeletal bone age derived from the O'Connor and abbreviated knee scales correlated with chronologic age (adjusted R2=0.88 and 0.90, respectively). Compared with estimates from the hand-wrist scale, estimates were lower by a mean of 0.91 years for boys and 0.38 years for girls when using the O'Connor scale and 0.96 years for boys and 0.52 years for girls when using the abbreviated scale. Inter- and intrarater reliabilities were very good (κ=0.82 and 0.90, respectively) and were substantial at each bony landmark measured. Knee radiographs can be used to estimate skeletal bone age using an abbreviated knee scale. [Orthopedics. 2018; 41(5):e676–e680.]

Abstract

Hand and wrist radiographs are the most common means for estimating skeletal bone age. There is no widely used scale for estimating skeletal bone age using knee radiographs. Do skeletal bone age estimates from knee-maturity scales correlate sufficiently with both chronologic age and estimates from a hand-wrist scale to potentially substitute for estimates from the latter? The authors reviewed the records of 503 patients 6 to 19 years old who had hand and knee radiographs obtained within 30 days of each other. They analyzed radiographs using the O'Connor knee scale (based on 10 maturation markers) and a new, abbreviated version of the O'Connor scale (based on 7 markers). The authors also analyzed radiographs of the hands of boys 12.5 to 16 years old and girls 10 to 16 years old using the shorthand method. Multivariate linear regression was used for analysis. Inter- and intrarater reliabilities were assessed. Skeletal bone age derived from the O'Connor and abbreviated knee scales correlated with chronologic age (adjusted R2=0.88 and 0.90, respectively). Compared with estimates from the hand-wrist scale, estimates were lower by a mean of 0.91 years for boys and 0.38 years for girls when using the O'Connor scale and 0.96 years for boys and 0.52 years for girls when using the abbreviated scale. Inter- and intrarater reliabilities were very good (κ=0.82 and 0.90, respectively) and were substantial at each bony landmark measured. Knee radiographs can be used to estimate skeletal bone age using an abbreviated knee scale. [Orthopedics. 2018; 41(5):e676–e680.]

Accurate estimates of remaining skeletal bone growth are helpful to guide surgical treatment around or involving the growth plates of skeletally immature patients. Chronologic age is not always an accurate predictor of remaining growth; instead, skeletal bone age can be estimated by assessing the morphological characteristics of developing bone. Many areas of the skeletal system mature according to reproducible patterns. One of the most common methods for estimating skeletal bone age is the hand-wrist method described by Greulich and Pyle.1 Although there is no commonly used method based on knee morphology, there are described patterns of maturation. The Pyle and Hoerr knee atlas shows standard knee morphology for different chronologic ages.2 Roche et al3 first examined 34 individual markers around the knee to determine skeletal bone age, and O'Connor et al4–6 derived formulas from a refined set of markers around the knee. No method for estimating skeletal bone age using the knee has been widely accepted.

If a simple method for accessing skeletal maturity using radiographs of the knee could be established, fewer lower-extremity patients would need to undergo radiography of the hand and wrist, which is commonly used for determining skeletal age.7 The original knee scales require the use of an atlas, while the O'Connor method includes markers that are difficult to grade. A more objective knee-based skeletal-maturity scale would help estimate the risk of growth disturbance posed when orthopedic procedures such as anterior cruciate ligament reconstruction require drilling through or around the growth plate.

Using the radiographic markers identified by previous studies,2–6 the authors sought to determine whether skeletal bone age estimates from an abbreviated knee-maturity scale and from a knee scale described by O'Connor et al correlate with both chronologic age and estimates from a hand-wrist scale, potentially eliminating the need for radiographs of the hand and wrist.

Materials and Methods

After institutional review board approval had been received, the authors retrospectively reviewed the radiographic records of all patients 6 to 19 years old who had radiographs of the hand and knee obtained within 30 days of each other at their tertiary hospital from January 2004 through January 2014. Of the 22,360 patients who had hand or knee radiographs, 361 boys and 319 girls had radiographs of the hand and knee. Patients with fractures, endocrine disorders, metabolic disorders, nutritional disorders, or genetic syndromes were excluded. Those without available radiographs of the anteroposterior and lateral knee were excluded. In this study group, indications for obtaining radiographs included determination of the cause of pain, evaluation for potential infection, and examination after injury. A total of 279 boys (77%) and 224 girls (70%) met the study criteria.

For initial analysis, the authors used the system described by O'Connor et al,4–6 grading 10 markers of skeletal maturity for each patient's radiographs (Table 1, Figure). Markers A to G grade the morphology of the femoral and tibial epiphyses and metaphyses, the tibial spine, the tibial tuberosity, and the fibular styloid. Markers H to J evaluate the union of the femoral, tibial, and fibular physes themselves. A numeric value was assigned to the grade of development of each marker as described by O'Connor et al.

Ten Markers Used in the O'Connor Method for Estimating Skeletal Bone Age Using Knee Radiographs

Table 1:

Ten Markers Used in the O'Connor Method for Estimating Skeletal Bone Age Using Knee Radiographs

Radiographic landmarks used for the abbreviated knee scale: A, lateral distal femoral epiphysis; B, lateral distal femoral metaphysis; D, lateral proximal tibial epiphysis; E, medial proximal tibial epiphysis; F, tibial tubercle; H, proximal tibial physis; and I, distal femoral physis.

Figure:

Radiographic landmarks used for the abbreviated knee scale: A, lateral distal femoral epiphysis; B, lateral distal femoral metaphysis; D, lateral proximal tibial epiphysis; E, medial proximal tibial epiphysis; F, tibial tubercle; H, proximal tibial physis; and I, distal femoral physis.

For subsequent analyses, the authors removed from the equation the following markers: development of the tibial spine, development of the fibular styloid, and union of the fibular physis. This simplified method consisted solely of markers of physeal, epiphyseal, and metaphyseal morphology of the femur and tibia.

For comparison with a hand-wrist skeletal-maturity scale, the authors used the shorthand method described by Heyworth et al,8 which is a scale for girls with skeletal bone ages of 10 to 16 years and boys with skeletal bone ages of 12.5 to 16 years. The subgroups of girls 10 to 14 years old and 15 to 16 years old were analyzed separately. Each set of radiographs was assessed independently by 2 orthopedic surgeons (X.T., R.J.L.) blinded to the patients' ages at the time of radiographic examinations.

Exploratory analyses suggested that the relationships between each maturity score and age were almost linear. A multivariate linear regression model was developed to predict skeletal bone age as follows:

age=ß0+ß1×sex+ß2×AG+ß3×HJ+ε.

ß1 is the coefficient of each covariate; ε is the error term that cannot be accounted for by the covariates. The AG stands for the sum of O'Connor markers A to G, and the HJ stands for the sum of markers H to J. Sex was treated as a binary variable, using 1 for boys and 0 for girls. The abbreviated model removes markers C, G, and J from the original model. Interaction terms between sex and maturity scores (eg, AG×sex, HJ×sex) were added to each model, but the nonsignificant P value of the coefficients led the authors to eliminate any interaction term from the models.

Predicted age was calculated for each patient according to both the O'Connor and the abbreviated regression equations. R2 was used to assess goodness of fit. The higher the R2 value, the greater the proportion of variation in age that can be predicted by the model. Assumptions of normality were examined using the Shapiro–Wilk test, and homoscedasticity of the residuals was tested using plots and the Breusch–Pagan test. Interrater reliability was assessed for each bony landmark using Cohen's κ given the use of nominal data. For the interpretation of κ, the degree of agreement was categorized as poor (κ<0), slight (κ=0.00–0.20), fair (κ=0.21–0.40), moderate (κ=0.41–0.60), good (κ=0.61–0.80), or very good (κ=0.81–1.00) according to generally accepted standards. To compare intrarater reliability, the same rater measured the same images twice, 1 week apart. Cohen's κ was then calculated using the same grading of agreements as for interrater reliability. All analyses were performed using Stata version 14 software (StataCorp LP, College Station, Texas) and SPSS version 22.0 software (SPSS, Inc, Chicago, Illinois).

Results

This study consisted of 503 patients (279 boys and 224 girls, with at least 15 boys and 15 girls of each age from 9 to 19 years). The individual markers of knee maturation of the O'Connor scale progressed on the grading scale with advancing chronologic age.

Estimating Chronologic Age

Age estimates from the O'Connor scale and the abbreviated scale correlated with chronologic age, with adjusted R2 values of 0.88 and 0.90, respectively (Table 2).

The O'Connor Scale and the Abbreviated Scale for Skeletal Bone Age Calculation and Its Correlation With Chronologic Age

Table 2:

The O'Connor Scale and the Abbreviated Scale for Skeletal Bone Age Calculation and Its Correlation With Chronologic Age

Estimating Skeletal Maturity

There were 123 boys 12 to 16 years old and 86 girls 10 to 16 years old evaluated using the shorthand method. For boys, skeletal bone ages predicted using both knee-based scales were lower than those predicted using the hand-wrist method, with mean differences of 0.91 years using the O'Connor scale and 0.96 years using the abbreviated scale. Similarly, for girls 10 to 14 years old, both knee-based scales produced skeletal bone age estimates that were lower than those predicted by the hand-wrist method, with mean differences of 0.38 years using the O'Connor scale and 0.52 years using the abbreviated scale. For girls 15 to 16 years old, according to the shorthand method, the knee scales predicted skeletal bone ages that were older, with mean differences of 0.4 years using the O'Connor scale and 0.46 years using the abbreviated scale.

Inter- and Intrarater Reliabilities

Interrater reliability of the measurements was 0.82 (95% confidence interval, 0.80–0.84). When analyzed at each section of the knee joint, reliability ranged from 0.78 (95% confidence interval, 0.72–0.84) at the epiphyseal union of the tibia to 0.86 (95% confidence interval, 0.79–0.92) at the medial corner of the tibial epiphysis. Intrarater reliability was 0.90 (95% confidence interval, 0.89–0.92). The range of intrarater reliability for each reviewer was from 0.76 (95% confidence interval, 0.68–0.84) at the styloid process of the fibula to 0.98 (95% confidence interval, 0.96–0.98) at the lateral corner of the proximal tibial epiphysis.

Discussion

Consideration of skeletal maturity is important when planning procedures around and involving the physes of the lower extremity. Often, radiographs of the hand and wrist are obtained to assess skeletal maturity of patients for whom knee radiographs are available. Grading of select markers around the knee with O'Connor's method and the authors' abbreviated method correlated well with chronologic age. The knee-based skeletal bone age estimates were lower than the skeletal bone age derived from the Heyworth method.8

This study was conducted in an urban tertiary referral center, and the sample was heterogeneous in terms of race. Previous studies have used homogeneous populations because the rate of skeletal maturity has been reported to vary across racial groups.1,2,4–6,9,10 However, the heterogeneity of the current study population allows the authors' findings to be generalized to the diverse patient populations at other tertiary referral centers. This study population was screened retrospectively. Although some coexisting systemic or syndromic conditions may have been missed, the authors' method allowed exclusion of patients who were diagnosed with conditions after the radiographs were obtained. Additionally, because patients with systemic and syndromic conditions were excluded from the study population, the authors' results may not apply to such patients. This is an area for future research.

Skeletal age estimates using the abbreviated knee scale correlated with chronologic age, as well as with skeletal age estimates derived from the O'Connor knee scale. The knee atlas of Pyle and Hoerr2 describes the morphologic changes that occur with skeletal maturity in approximately 12-month intervals for boys and girls. Although the atlas is a building block for evaluating skeletal maturity using the knee, it is difficult to use clinically, and their more homogeneous study population limits the atlas's generalizability to the more heterogeneous populations currently seen at major medical institutions.

Roche and French11 and O'Connor et al4–6 examined physeal closure patterns around the knee. The pattern of physeal closure itself, however, is gradual and is impractical for determining skeletal age.9 The Roche method focuses instead on morphology through 34 indicators of skeletal maturity.3 The estimates of this method have been found to correlate with estimates from the original Pyle method and with Tanner stages.10,12 The O'Connor method simplifies knee-maturity evaluation by using only 10 markers.4–6,9 The method grades structural changes around the physes, of the tibial spine, and of the fibular styloid to calculate age using a regression equation. However, the development of the tibial spine and fibular styloid and the physeal closure of the fibula are difficult to grade. Evaluation of the development of the tibial spine and fibular styloid is challenging because their morphology changes gradually. Although the amount of physeal closure in the distal femur and proximal tibia can be estimated, the narrow fibular physis is difficult to assess.

The O'Connor method involves a formula for calculating skeletal age. In creating an abbreviated knee scale, the authors attempted to further simplify grading by eliminating the markers that are difficult to interpret and focusing solely on the femur and tibia, which are the 2 major contributors to growth and deformity around the knee. Skeletal age estimates using the authors' abbreviated scale correlated well with chronologic age. The authors' results also validate a reliable method of measurement given the “very good” inter-and intrarater reliabilities with κ values of 0.82 and 0.90, respectively, using the standard grading scale for κ values. The measurements continued to be consistently reliable, at greater than 0.78, across each bony landmark. The good interrater agreement illustrates that the authors' findings are robust and unlikely to be influenced by user subjectivity. With similarly high intrarater reliability, a rater could likely reproduce the same outcome, confirming the applicability of the O'Connor method for skeletal age measurement of the knee using radiographs.

Skeletal age estimates from both knee-maturity scales correlated with those from the hand-wrist scale. However, there is a consistently younger estimated age from knee-maturity scales until the skeletal bone ages of 14 years for girls and 16 years for boys, after which, for girls 15 to 16 years old, the younger estimated age disappears. This younger age estimate from knee radiographs seems to be physiologic because it has been shown in other knee-hand comparative studies. Aicardi et al12 compared the Roche scale with the Pyle method and described a “laziness” in progression of maturation of the knee. In this study, O'Connor's simplified method and the abbreviated method estimated ages that were younger than that of the shorthand method until the upper end of skeletal bone age, which is 15 to 16 years for girls. The change at the upper end of skeletal bone age for girls 15 to 16 years old on the shorthand method is caused by functional closure of the physes around the knee. After 14 years of hand skeletal bone age for girls, the markers in the knee scales reach the end of the grading scale. Thus, although the knee skeletal bone age estimate does not change, the shorthand method differentiates between 15 and 16 years. This would likely be revealed after age 16 years for boys, although this was not analyzed because the shorthand scale stops at age 16 years for boys. This disagreement is clinically unimportant because the child's knee at functional closure of the physis can be managed like that of an adult. The physes are not producing significant growth; therefore, guided growth is no longer possible, and the physes can be violated during surgery without hindering growth potential. Furthermore, because differences in estimates from the hand and knee scales exist, direct estimates of skeletal bone age using knee-based scales would be more appropriate when planning procedures involving the femur or tibia, rather than indirectly estimating skeletal bone age through the hand.

Overall, knee-based scales are more difficult to use because of the need to differentiate among the subtle changes of 3 physes, whereas hand-based scales use the 21 physes in the hand and wrist along with the appearance of ossification centers. The shorthand method, based on the Greulich and Pyle method, covers boys 12 to 16 years old and girls 10 to 16 years old. It has been shown to have validity and reliability comparable or superior to those of other skeletal age assessment tools. This age range is the focus for the pediatric sports medicine subspecialist when deciding between performing an adult or a pediatric anterior cruciate ligament reconstruction. In the authors' experience, the shorthand method is popular among pediatric deformity and sports medicine subspecialists because of its ease of use and its inclusion in mobile computer applications already used for deformity calculation. Similarly, the development of the knee scale into an application will simplify its use.

Conclusion

Radiographs of the knee can be used to estimate skeletal bone age, eliminating the need for radiographs of the hand, which often estimate an older skeletal bone age compared with the knee method. This is applicable for conditions involving the lower extremity and for surgical planning around the knee, in which radiographs of the knee are often obtained as part of the initial evaluation. This spares the patient the cost and additional radiation exposure of obtaining a radiograph of the hand and wrist. Future studies should investigate new markers of femoral and tibial maturity using different imaging modalities and should validate the abbreviated knee scale through clinical application.

References

  1. Greulich WW, Pyle SI. Radiographic Atlas of Skeletal Development of the Hand and Wrist. Redwood City, CA: Stanford University Press; 1959.
  2. Pyle SI, Hoerr NL. Radiographic Atlas of Skeletal Development of the Knee: A Standard of Reference. Springfield, IL: Charles C. Thomas Publisher; 1955.
  3. Roche AF, Wainer H, Thissen D. Skeletal Maturity: The Knee Joint as a Biological Indicator. New York, NY: Plenum Medical Book Company; 1975.
  4. O'Connor JE, Bogue C, Spence LD, Last J. A method to establish the relationship between chronological age and stage of union from radiographic assessment of epiphyseal fusion at the knee: an Irish population study. J Anat. 2008; 212(2):198–209. doi:10.1111/j.1469-7580.2007.00847.x [CrossRef]
  5. O'Connor JE, Coyle J, Bogue C, Spence LD, Last J. Age prediction formulae from radiographic assessment of skeletal maturation at the knee in an Irish population. Forensic Sci Int. 2014; 234:188.
  6. O'Connor JE, Coyle J, Spence LD, Last J. Epiphyseal maturity indicators at the knee and their relationship to chronological age: results of an Irish population study. Clin Anat. 2013; 26(6):755–767. doi:10.1002/ca.22122 [CrossRef]
  7. Milewski MD, Nissen CW. Femoral physeal sparing/transphyseal tibial (hybrid) technique for ACL reconstruction in skeletally immature athletes. In: Parikh SN, ed. The Pediatric Anterior Cruciate Ligament: Evaluation and Management Strategies. Cham, Switzerland: Springer; 2018:147–155. doi:10.1007/978-3-319-64771-5_15 [CrossRef]
  8. Heyworth BE, Osei DA, Fabricant PD, et al. The shorthand bone age assessment: a simpler alternative to current methods. J Pediatr Orthop. 2013; 33(5):569–574. doi:10.1097/BPO.0b013e318293e5f2 [CrossRef]
  9. Ontell FK, Ivanovic M, Ablin DS, Barlow TW. Bone age in children of diverse ethnicity. AJR Am J Roentgenol. 1996; 167(6):1395–1398. doi:10.2214/ajr.167.6.8956565 [CrossRef]
  10. Vignolo M, Milani S, DiBattista E, Naselli A, Mostert M, Aicardi G. Modified Greulich-Pyle, Tanner-Whitehouse, and Roche-Wainer-Thissen (knee) methods for skeletal age assessment in a group of Italian children and adolescents. Eur J Pediatr. 1990; 149(5):314–317. doi:10.1007/BF02171555 [CrossRef]
  11. Roche AF, French NY. Differences in skeletal maturity levels between the knee and hand. Am J Roentgenol Radium Ther Nucl Med. 1970; 109(2):307–312. doi:10.2214/ajr.109.2.307 [CrossRef]
  12. Aicardi G, Vignolo M, Milani S, Naselli A, Magliano P, Garzia P. Assessment of skeletal maturity of the hand-wrist and knee: a comparison among methods. Am J Hum Biol. 2000; 12(5):610–615. doi:10.1002/1520-6300(200009/10)12:5<610::AID-AJHB5>3.0.CO;2-D [CrossRef]

Ten Markers Used in the O'Connor Method for Estimating Skeletal Bone Age Using Knee Radiographs

MarkerDescriptionGrade/Stage
AProximal projection of the lateral corner of the distal femoral epiphysis0/Projection absent 1/Projection present 2/Fused
BLateral capping of the metaphysis by the distal femoral epiphysis0/Capping absent 1/Capping incomplete 2/Capping complete 3/Fused
CDevelopment of the tibial spine0/Developing 1/Present
DDistal projection of the lateral corner of the proximal tibial epiphysis0/Projection absent 1/Projection present 2/Fused
EDistal projection of the medial corner of the proximal tibial epiphysis0/Projection absent 1/Projection present 2/Fused
FDevelopment of the tibial tuberosity (on lateral radiographs)0/Absent 1/Developing 2/Fusion incomplete distally 3/Fusion complete
GDevelopment of the styloid process of the fibula0/Absent 1/Early development 2/Present
H–JEpiphyseal union of the femur (H), tibia (I), and fibula (J)0/Physis open: continuous radiolucent gap between epiphysis and diaphysis 1/Partial closure: >50% of the growth plate is open 2/Active closure: ≤50% of the growth plate is open 3/Recent closure: epiphysis and diaphysis are fused, but there is discontinuity of trabeculae and notches at the peripheral margins of the bone 4/Complete closure: epiphysis and diaphysis in continuity

The O'Connor Scale and the Abbreviated Scale for Skeletal Bone Age Calculation and Its Correlation With Chronologic Age

ScaleAdjusted R2Formula to Determine Age in Yearsa
O'Connor knee scaleb0.881.70×sexc+0.54×AG+0.22×HJ+6.15
Abbreviated knee scaled0.901.85×sexc+0.70×AG'+0.25×HJ'+6.40
Authors

The authors are from the Department of Orthopaedic Surgery (XT, DTN, WK, RJL), The Johns Hopkins Hospital; the Russell H. Morgan Department of Radiology and Radiological Science (YL, JF), The Johns Hopkins University School of Medicine; and The Johns Hopkins University (MP), Bloomberg School of Public Health, Baltimore, Maryland.

Dr Tang, Dr Lu, Mr Pang, Mr Nhan, Mr Klyce, and Dr Lee have no relevant financial relationships to disclose. Dr Fritz has received material support from DePuy and Zimmer, has received grants from Siemens AG and BTG International, is on the speaker's bureau of Siemens AG, and holds a patent with Siemens AG.

Correspondence should be addressed to: R. Jay Lee, MD, Department of Orthopaedic Surgery, The Johns Hopkins Hospital, 1800 Orleans St, Baltimore, MD 21287 ( editorialservices@jhmi.edu).

Received: October 13, 2017
Accepted: April 23, 2018
Posted Online: July 27, 2018

10.3928/01477447-20180724-03

Sign up to receive

Journal E-contents