- Email: [email protected]
Defining Ulnar Variance in the Adolescent Wrist: Measurement Technique and Interobserver Reliability
SCIENTIFIC ARTICLE
Defining Ulnar Variance in the Adolescent Wrist: Measurement Technique and Interobserver Reliability Charles A. Goldfarb, MD, Nicole L. Strauss, MD, Lindley B. Wall, MD, Ryan P. Calfee, MD
Purpose The measurement technique for ulnar variance in the adolescent population has not been well established. The purpose of this study was to assess the reliability of a standard ulnar variance assessment in the adolescent population. Methods Four orthopedic surgeons measured 138 adolescent wrist radiographs for ulnar variance using a standard technique. There were 62 male and 76 female radiographs obtained in a standardized fashion for subjects aged 12 to 18 years. Skeletal age was used for analysis. We determined mean variance and assessed for differences related to age and gender. We also determined the interrater reliability. Results The mean variance was – 0.7 mm for boys and – 0.4 mm for girls; there was no significant difference between the 2 groups overall. When subdivided by age and gender, the younger group (ⱕ15 y of age) was significantly less negative for girls (boys, – 0.8 mm and girls, – 0.3 mm, p⬍.05). There was no significant difference between boys and girls in the older group. The greatest difference between any 2 raters was 1 mm; exact agreement was obtained in 72 subjects. Correlations between raters were high (rp 0.87– 0.97 in boys and 0.82– 0.96 for girls). Interrater reliability was excellent (Cronbach’s alpha, 0.97– 0.98). Conclusions Standard assessment techniques for ulnar variance are reliable in the adolescent population. Open growth plates did not interfere with this assessment. Young adolescent boys demonstrated a greater degree of negative ulnar variance compared with young adolescent girls. (J Hand Surg 2011;36A:272–277. Copyright © 2011 by the American Society for Surgery of the Hand. All rights reserved.) Key words Ulnar variance, measurement, reliability, impaction, adolescent.
U
LNAR VARIANCE IS the relative length of the ulna
in relation to the radius at the level of the wrist. In adults, several different measurement techniques are described to quantify the proximal-distal distance between the most ulnar aspect of the distal radius and the distal extent of the ulna (excluding the From the Shriners Hospital for Children and the Department of Orthopaedic Surgery, St. Louis Children’s Hospital at Washington University School of Medicine, St. Louis, MO. The authors thank Jennifer A. Steffen for assistance with data preparation. Received for publication April 19, 2010; accepted in revised form November 3, 2010. No benefits in any form have been received or will be received related directly or indirectly to the subject of this article. Correspondingauthor:CharlesA.Goldfarb,MD,DepartmentofOrthopaedicSurgery,Washington University School of Medicine, 660 South Euclid Avenue, Campus Box 8233, St. Louis, MO 63110; e-mail: [email protected]. 0363-5023/11/36A02-0011$36.00/0 doi:10.1016/j.jhsa.2010.11.008
272 䉬 © ASSH 䉬 Published by Elsevier, Inc. All rights reserved.
ulnar styloid).1,2 In a review article, Graham3 noted that ulnar variance “can be calculated in a variety of ways, all of which yield similar results.” Steyers and Blair4 detailed 3 different measurement techniques for ulnar variance—the line technique, the concentric circle technique,1 and the method of perpendiculars5— and reported similar results and a high reliability among these techniques. The method of perpendiculars was noted to have the highest interobserver and intraobserver reliability. Although multiple articles evaluate ulnar variance in adults, few data are available for the younger subject.6 It is important to assess ulnar variance in the adolescent given the different pathological states in which ulnar variance has a major role: Kienböck’s disease, multiple osteochondromatosis, growth arrest, ulnocarpal impaction, type I and II radial longitudinal deficiency,7 and
273
ADOLESCENT ULNAR VARIANCE
Madelung’s deformity. A single measurement technique to determine ulnar variance in children has been described6,8,9 addressing subjects ages 1.5 to 15 years without full ossification of the epiphysis. In a description of the measurement technique, Hafner et al.6 measured the distance between the ulnar and radial metaphysis (either proximal metaphysis to proximal metaphysis or distal metaphysis to distal metaphysis) to quantify variance. Variance was relatively constant across the age spectrum, although it became slightly more ulnar negative with increasing age. There was no noted relationship between variance and gender. In this investigation, we evaluated ulnar variance in a large number of adolescent subjects with the standard measurement technique currently used in adults. We hypothesized that adolescent ulnar variance may be reliably assessed with this technique rather than the less familiar technique of Hafner et al.6 In addition, we hypothesized that mean variance measurements would approximate previously reported adult values. MATERIALS AND METHODS We obtained institutional review board approval to assess ulnar variance for 138 adolescent wrists. All films had been obtained at our pediatric orthopedic hospital for a recent study on skeletal age using standardized radiographic techniques10; dedicated musculoskeletal radiology technicians obtained all films with strict adherence to our protocol. A single posterior-anterior radiograph of the left wrist was obtained with the forearm in neutral rotation for each subject. These films were standardized with size markers to allow a true length measurement. All films were of satisfactory quality, and we excluded no films from analysis. The radiographs were de-identified of personal health information, and each participant’s hand radiograph was coded for chronological and skeletal age. One attending orthopedic surgeon, who regularly establishes skeletal age using the Greulich and Pyle Atlas,11 determined skeletal age in all subjects. There were 76 girls and 62 boys, 12 to 18 years of age. Table 1 lists the distribution of subjects by skeletal age. Subjects enrolled were predominately white (n ⫽ 116), with a minority of African American (n ⫽ 10), Hispanic (n ⫽ 2), Asian (n ⫽ 1), and other (n ⫽ 9) subjects. Potential subjects were excluded if they had either a pre-existing diagnosis of, or were being evaluated for, any genetic syndrome, growth abnormality, congenital abnormality of the upper extremity, or history of any upper-extremity fracture or pain. Owing to the age range of subjects, 12 boys and 20 girls had closed physes of the distal ulna.
TABLE 1. Distribution of Participants by Skeletal Age Skeletal Age
Girls
Boys
11
3
2
12
8
4
13
7
13
14
13
7
15
11
9
16
14
4
17
11
12
18
9
8
19
0
3
76
62
Total
Four orthopedic surgeons (2 attending hand surgeons, 1 fellow in hand surgery, and 1 senior level resident) independently measured ulnar variance for each radiograph. The measurements were performed using the method of perpendiculars as assessed by Steyers and Blair.4 First, the sclerotic line representing the volar distal radius was identified.1 This line, and its most ulnar point, was chosen because of the consistency of its appearance and the fact that it has been consistently used as the reference on the radius when determining ulnar variance across measurement techniques in adults and adolescents.1,2,4,8 We then used a template of perpendicular lines with distances measured in 1-mm increments to determine the proximal- distal distance between this point and a line parallel to the most distal, flat articular surface of the distal ulna, excluding consideration of the ulnar styloid (Fig. 1). The longitudinal lines of the template were placed parallel to the long axis of the ulna. A positive measure indicates that the ulna is longer than the radius, and a negative measure, that the ulna is shorter. Data analysis We performed data analysis using SPSS statistical software (version 17.0.1; SPSS, Inc., Chicago, IL). Skeletal age was used for all assessments. We first determined mean ulnar variance for the entire cohort. Unpaired Student’s t-test compared the mean variance between girls and boys. We divided the subjects into 2 groups based on skeletal age: a younger group (ⱕ15 y) and an older group (ⱖ16 y). There were 35 boys and 42 girls in the younger group and 27 boys and 34 girls in the older group. We believed that the older group represented a skeletally mature group and should be separated from the younger group. Unpaired Student’s t-
JHS 䉬 Vol A, February
274
ADOLESCENT ULNAR VARIANCE
sented the maximum number available from another investigation of skeletal maturation in the adolescent.10 We performed a post hoc assessment of power to evaluate our ability to detect a difference in mean variance of 1.0 mm between any 2 groups analyzed in this study (⫽0.08, ␣⫽0.05). We chose 1 mm as the smallest clinically significant difference for analysis. Provided a standard deviation of approximately 1.1 mm in variance measures, 20 subjects in each group (e.g., females/ males or older females/younger females) were required.
FIGURE 1: Left wrist radiograph of a 12-year-old girl with template superimposed. The template is parallel to the longitudinal axis of the ulna. The ulnar variance in this subject was ⫹2 mm, as agreed on by all 4 raters.
TABLE 2. Age
Mean Variance for Cohort by Skeletal
Skeletal Age
Mean Variance
Cohort
95% Confidence Interval
11
⫺0.5
⫺1.6 to 0.6
12
⫺0.2
⫺0.7 to 0.3
13
⫺0.5
⫺1.0 to 0.0
14
⫺0.7
⫺1.1 to ⫺0.2
15
⫺0.6
⫺1.1 to ⫺0.1
16
⫺0.7
⫺1.5 to 0.1
17
⫺0.6
⫺1.1 to ⫺0.3
18
⫺0.3
⫺0.9 to 0.4
19
0.7
⫺2.5 to 3.9
tests determined the relationship of mean variance between younger and older age groups both within and between genders. We used one-way analysis of variance to compare the mean ulnar variance between each skeletal age for girls and boys. We determined interrater reliability for this explicitly defined measure of ulnar variance. Greatest difference and percentage of exact agreement among the 4 raters were determined. Pearson correlations and Cronbach’s alpha were calculated to compare performance between the raters and intraclass correlation coefficients were calculated using 2-way random effects modeling. The sample size in this study was one of convenience. The 138 standardized wrist radiographs repre-
RESULTS Entire cohort The mean variance for the 138 subjects was – 0.5 mm (SD, 1.2 mm). We detected no significant difference in ulnar variance across skeletal ages in this cohort (p⫽.74). Table 2 presents the mean variance by skeletal age for the entire cohort. Gender The mean variance for the 62 male subjects was – 0.7 mm (SD, 1.2 mm) and the mean variance for the 76 female subjects was – 0.4 mm (SD, 1.1). Table 3 provides a summary of mean variance per skeletal age interval; skeletal age ranges from 11 to 19 years owing to variability between skeletal and chronological ages. The mean skeletal ages for girls and boys was similar (girls, 15.0 y; boys, 15.3 y). Figures 2 and 3 further quantify the relationship among variance, age, and gender. There was no significant difference between the boys and girls (p⫽.16). We also divided the entire cohort into 2 groups, a younger group (ⱕ15 y) and an older group (ⱖ16 y). There were 35 boys and 42 girls in the younger group. The mean variance was – 0.8 mm for the younger boys (SD, 1.0 mm) and – 0.3 mm (SD, 1.0 mm) for the younger girls. The boys’ variance was significantly more ulnar negative than the girls’ (p⫽.03) in this younger group. The older group included 27 boys and 34 girls. The mean variance was – 0.5 mm (SD, 1.5 mm) for the boys and – 0.5 mm (SD, 1.3 mm) for the girls. There was no significant difference between older boys and girls (p⫽.98). Age We evaluated the effect of skeletal age on ulnar variance (Table 3). There was no significant difference in mean variance across the age spectrum for either boys (p⫽.09) or girls (p⫽.94). Similarly, when comparing the younger and older age groups within each gender, there were no significant differences among the boys (p⫽.30) or girls (p⫽.54).
JHS 䉬 Vol A, February
275
ADOLESCENT ULNAR VARIANCE
TABLE 3.
Mean Variance for Boys and Girls, by Skeletal Age
Skeletal Age
Mean Variance Girls
95% Confidence Interval
11
⫺0.8
⫺3.6 to 2.1
Mean Variance Boys 0.0
95% Confidence Interval *
12
0.0
⫺0.6 to 0.6
⫺0.6
⫺2.1 to 1.0
13
⫺1.1
⫺0.8 to 0.6
⫺0.8
⫺1.5 to ⫺0.0
14
⫺0.4
⫺1.0 to 0.2
⫺1.1
⫺2.0 to ⫺0.1
15
⫺0.4
⫺1.3 to 0.5
⫺0.9
⫺1.4 to ⫺0.3
16
⫺0.3
⫺1.2 to 0.6
⫺2.1
⫺4.1 to ⫺0.1
17
⫺0.5
⫺1.2 to 0.2
⫺0.6
⫺1.6 to 0.3
18
⫺0.6
⫺1.6 to 0.3
0.2
⫺0.8 to 1.1
19
NA
0.7
⫺2.5 to 3.9
*Not provided because all 8 (4 raters/2 subjects) observations at this age were 0.00.
FIGURE 2: Box plot of ulnar variance in boys, by skeletal age.
Reliability The greatest difference between any 2 raters was 1 mm. All 4 raters demonstrated exact agreement in 72 cases (52%). Correlations among all 4 raters were high, ranging from rp 0.87 to 0.97 for boys and 0.82 to 0.96 for girls. Measures from each rater were significantly correlated to the other 3 raters (p⬍.001) for each gender. In
addition, interrater reliability judged by Cronbach’s alpha was excellent, at 0.98 for boys and 0.97 for girls. The intraclass correlation for single measures was 0.92 for boys and 0.89 for girls. Discussion Two previously published reports assess ulnar variance in a pediatric or adolescent population.8,9 In both of
JHS 䉬 Vol A, February
276
ADOLESCENT ULNAR VARIANCE
FIGURE 3: Box plot of ulnar variance in girls, by skeletal age.
those articles, the authors used the technique of Hafner et al.6 in gymnasts at risk for early physeal closure due to repetitive trauma. De Smet et al.8 evaluated 156 elite gymnasts with open physes and a mean chronological age of 15.9 years, and DiFiori et al.9 evaluated 59 gymnasts of differing skill levels of a mean chronological age of 9.3 years. Both reported mean ulnar variance values that were significantly greater than the average nonathlete values provided by Hafner et al.6 (data from 535 radiographs of children of all ages). Neither of these studies used skeletal age to assess variance. We believe that the technique of Hafner et al.6 may be ideal for children without ossification of the epiphysis, but it has limitations in the adolescent population. The measurement technique is not familiar to most hand surgeons, and the measurement values are not comparable to those commonly reported in adult subjects. The technique uses the physis location to define the relationship between radius and ulna. Although it provides a constant point for comparison, the technique does not truly compare radius and ulna length. Instead, these comparison points allow assessment in a single patient over time or allow comparison with previously established normal values (as provided in the report by
Hafner et al.). In addition, it is unclear whether the radiographs in the article were standardized for either forearm position or size parameters. Reports on average ulnar variance in the normal adult population, utilize varied measurement techniques. Schuind et al.2 measured ulnar variance, among other radiographic parameters, in 120 adults using the Palmar concentric circle technique, and reported an average variance of – 0.9 mm (range, – 4.2 to 2.3 mm). Nakamura et al.,12 in a Japanese population of 325 subjects and using the Palmer template technique, reported an overall variance mean of 0.20 mm (SD, 1.39 mm), with male subjects averaging – 0.14 mm (SD, 1.34 mm), significantly less than the female subjects’ average of 0.77 mm (SD, 1.27 mm). The authors reported that age correlated with variance; as age increased, variance increased or became more positive. We found an average ulnar variance of – 0.7 for boys and – 0.4 for girls in the adolescent age group, values between the reported average values from Nakamura et al. and Schuind et al. Ulnar variance did not change significantly over the adolescent age spectrum, and the only statistically significant finding was that variance in the younger boys were more negative than variance in the younger girls.
JHS 䉬 Vol A, February
ADOLESCENT ULNAR VARIANCE
There are several limitations to our study. Although we enrolled 138 subjects, additional subjects would lend greater validity to the average values. We chose to use skeletal age as opposed to chronologic age because we believe this to be a more reliable measure of maturation, although it resulted in unequal age distributions within our cohort. In addition, with a preponderance of white subjects, we are unable to comment on any differences attributable to race. Also, we screened subjects for study inclusion by verbal report from parents for any history or current issue related to growth or trauma of the upper extremity. It is possible that a subtle growth abnormality due to previous trauma or other causes may have existed in our subject population, although none of the radiographs suggested an abnormality. Furthermore, we acknowledge that the measurements performed in this study are not possible before ossification of the distal radial and ulnar epiphysis. Finally, we excluded the nonossified epiphysis of both the radius and the ulna from consideration; magnetic resonance imaging would be required to evaluate the contribution of any such cartilage “cap.” As such, we assessed the relative lengths of the ossified epiphyses of the radius and ulna. The traditional technique for ulnar variance assessment, as currently employed in adult subjects, is more familiar to hand surgeons than the technique of Hafner et al.,6 considering both the measurement technique and the variance distance interpretation. The results of this investigation confirm that this more traditional assessment of ulnar variance may be used reliably and accurately in the adolescent subject. The open physis of the radius and ulna in subjects with a well-developed epiphysis did not interfere with the measurement of
277
ulnar variance. The high interrater reliability speaks to the ease of this method for determining ulnar variance in the adolescent population. Employing an explicitly defined measurement technique, raters were consistent across a range of experience and levels of training. REFERENCES 1. Palmer AK, Glisson RR, Werner FW. Ulnar variance determination. J Hand Surg 1982;7:376 –379. 2. Schuind FA, Linscheid RL, An KN, Chao EY. A normal data base of posteroanterior roentgenographic measurements of the wrist. J Bone Joint Surg 1992;74A:1418 –1429. 3. Graham TJ. Surgical correction of malunited fractures of the distal radius. J Am Acad Orthop Surg 1997;5:270 –281. 4. Steyers CM, Blair WF. Measuring ulnar variance: a comparison of techniques. J Hand Surg 1989;14A:607– 612. 5. Coleman DA, Blair WF, Shurr D. Resection of the radial head for fracture of the radial head. Long-term follow-up of seventeen cases. J Bone Joint Surg 1987;69A:385–392. 6. Hafner R, Poznanski AK, Donovan JM. Ulnar variance in children— standard measurements for evaluation of ulnar shortening in juvenile rheumatoid arthritis, hereditary multiple exostosis and other bone or joint disorders in childhood. Skeletal Radiol 1989;18:513–516. 7. Bayne LG, Klug MS. Long-term review of the surgical treatment of radial deficiencies. J Hand Surg 1987;12A:169 –179. 8. De Smet L, Claessens A, Lefevre J, Beunen G. Gymnast wrist: an epidemiologic survey of ulnar variance and stress changes of the radial physis in elite female gymnasts. Am J Sports Med 1994;22: 846 – 850. 9. DiFiori JP, Puffer JC, Aish B, Dorey F. Wrist pain, distal radial physeal injury, and ulnar variance in young gymnasts: does a relationship exist? Am J Sports Med 2002;30:879 – 885. 10. Calfee R, Sutter M, Steffen J, Goldfarb C. Skeletal and chronological ages in American adolescents: current findings in skeletal maturation. J Child Orthop 2010;4:467– 470. 11. Greulich W, Pyle S. Radiographic atlas of skeletal of the hand and wrist. 2nd ed. Stanford, CA: Stanford University Press, 1959. 12. Nakamura R, Tanaka Y, Imaeda T, Miura T. The influence of age and sex on ulnar variance. J Hand Surg 1991;16B:84 – 88.
JHS 䉬 Vol A, February
Defining Ulnar Variance in the Adolescent Wrist: Measurement Technique and Interobserver Reliability Charles A. Goldfarb, MD, Nicole L. Strauss, MD, Lindley B. Wall, MD, Ryan P. Calfee, MD
Purpose The measurement technique for ulnar variance in the adolescent population has not been well established. The purpose of this study was to assess the reliability of a standard ulnar variance assessment in the adolescent population. Methods Four orthopedic surgeons measured 138 adolescent wrist radiographs for ulnar variance using a standard technique. There were 62 male and 76 female radiographs obtained in a standardized fashion for subjects aged 12 to 18 years. Skeletal age was used for analysis. We determined mean variance and assessed for differences related to age and gender. We also determined the interrater reliability. Results The mean variance was – 0.7 mm for boys and – 0.4 mm for girls; there was no significant difference between the 2 groups overall. When subdivided by age and gender, the younger group (ⱕ15 y of age) was significantly less negative for girls (boys, – 0.8 mm and girls, – 0.3 mm, p⬍.05). There was no significant difference between boys and girls in the older group. The greatest difference between any 2 raters was 1 mm; exact agreement was obtained in 72 subjects. Correlations between raters were high (rp 0.87– 0.97 in boys and 0.82– 0.96 for girls). Interrater reliability was excellent (Cronbach’s alpha, 0.97– 0.98). Conclusions Standard assessment techniques for ulnar variance are reliable in the adolescent population. Open growth plates did not interfere with this assessment. Young adolescent boys demonstrated a greater degree of negative ulnar variance compared with young adolescent girls. (J Hand Surg 2011;36A:272–277. Copyright © 2011 by the American Society for Surgery of the Hand. All rights reserved.) Key words Ulnar variance, measurement, reliability, impaction, adolescent.
U
LNAR VARIANCE IS the relative length of the ulna
in relation to the radius at the level of the wrist. In adults, several different measurement techniques are described to quantify the proximal-distal distance between the most ulnar aspect of the distal radius and the distal extent of the ulna (excluding the From the Shriners Hospital for Children and the Department of Orthopaedic Surgery, St. Louis Children’s Hospital at Washington University School of Medicine, St. Louis, MO. The authors thank Jennifer A. Steffen for assistance with data preparation. Received for publication April 19, 2010; accepted in revised form November 3, 2010. No benefits in any form have been received or will be received related directly or indirectly to the subject of this article. Correspondingauthor:CharlesA.Goldfarb,MD,DepartmentofOrthopaedicSurgery,Washington University School of Medicine, 660 South Euclid Avenue, Campus Box 8233, St. Louis, MO 63110; e-mail: [email protected]. 0363-5023/11/36A02-0011$36.00/0 doi:10.1016/j.jhsa.2010.11.008
272 䉬 © ASSH 䉬 Published by Elsevier, Inc. All rights reserved.
ulnar styloid).1,2 In a review article, Graham3 noted that ulnar variance “can be calculated in a variety of ways, all of which yield similar results.” Steyers and Blair4 detailed 3 different measurement techniques for ulnar variance—the line technique, the concentric circle technique,1 and the method of perpendiculars5— and reported similar results and a high reliability among these techniques. The method of perpendiculars was noted to have the highest interobserver and intraobserver reliability. Although multiple articles evaluate ulnar variance in adults, few data are available for the younger subject.6 It is important to assess ulnar variance in the adolescent given the different pathological states in which ulnar variance has a major role: Kienböck’s disease, multiple osteochondromatosis, growth arrest, ulnocarpal impaction, type I and II radial longitudinal deficiency,7 and
273
ADOLESCENT ULNAR VARIANCE
Madelung’s deformity. A single measurement technique to determine ulnar variance in children has been described6,8,9 addressing subjects ages 1.5 to 15 years without full ossification of the epiphysis. In a description of the measurement technique, Hafner et al.6 measured the distance between the ulnar and radial metaphysis (either proximal metaphysis to proximal metaphysis or distal metaphysis to distal metaphysis) to quantify variance. Variance was relatively constant across the age spectrum, although it became slightly more ulnar negative with increasing age. There was no noted relationship between variance and gender. In this investigation, we evaluated ulnar variance in a large number of adolescent subjects with the standard measurement technique currently used in adults. We hypothesized that adolescent ulnar variance may be reliably assessed with this technique rather than the less familiar technique of Hafner et al.6 In addition, we hypothesized that mean variance measurements would approximate previously reported adult values. MATERIALS AND METHODS We obtained institutional review board approval to assess ulnar variance for 138 adolescent wrists. All films had been obtained at our pediatric orthopedic hospital for a recent study on skeletal age using standardized radiographic techniques10; dedicated musculoskeletal radiology technicians obtained all films with strict adherence to our protocol. A single posterior-anterior radiograph of the left wrist was obtained with the forearm in neutral rotation for each subject. These films were standardized with size markers to allow a true length measurement. All films were of satisfactory quality, and we excluded no films from analysis. The radiographs were de-identified of personal health information, and each participant’s hand radiograph was coded for chronological and skeletal age. One attending orthopedic surgeon, who regularly establishes skeletal age using the Greulich and Pyle Atlas,11 determined skeletal age in all subjects. There were 76 girls and 62 boys, 12 to 18 years of age. Table 1 lists the distribution of subjects by skeletal age. Subjects enrolled were predominately white (n ⫽ 116), with a minority of African American (n ⫽ 10), Hispanic (n ⫽ 2), Asian (n ⫽ 1), and other (n ⫽ 9) subjects. Potential subjects were excluded if they had either a pre-existing diagnosis of, or were being evaluated for, any genetic syndrome, growth abnormality, congenital abnormality of the upper extremity, or history of any upper-extremity fracture or pain. Owing to the age range of subjects, 12 boys and 20 girls had closed physes of the distal ulna.
TABLE 1. Distribution of Participants by Skeletal Age Skeletal Age
Girls
Boys
11
3
2
12
8
4
13
7
13
14
13
7
15
11
9
16
14
4
17
11
12
18
9
8
19
0
3
76
62
Total
Four orthopedic surgeons (2 attending hand surgeons, 1 fellow in hand surgery, and 1 senior level resident) independently measured ulnar variance for each radiograph. The measurements were performed using the method of perpendiculars as assessed by Steyers and Blair.4 First, the sclerotic line representing the volar distal radius was identified.1 This line, and its most ulnar point, was chosen because of the consistency of its appearance and the fact that it has been consistently used as the reference on the radius when determining ulnar variance across measurement techniques in adults and adolescents.1,2,4,8 We then used a template of perpendicular lines with distances measured in 1-mm increments to determine the proximal- distal distance between this point and a line parallel to the most distal, flat articular surface of the distal ulna, excluding consideration of the ulnar styloid (Fig. 1). The longitudinal lines of the template were placed parallel to the long axis of the ulna. A positive measure indicates that the ulna is longer than the radius, and a negative measure, that the ulna is shorter. Data analysis We performed data analysis using SPSS statistical software (version 17.0.1; SPSS, Inc., Chicago, IL). Skeletal age was used for all assessments. We first determined mean ulnar variance for the entire cohort. Unpaired Student’s t-test compared the mean variance between girls and boys. We divided the subjects into 2 groups based on skeletal age: a younger group (ⱕ15 y) and an older group (ⱖ16 y). There were 35 boys and 42 girls in the younger group and 27 boys and 34 girls in the older group. We believed that the older group represented a skeletally mature group and should be separated from the younger group. Unpaired Student’s t-
JHS 䉬 Vol A, February
274
ADOLESCENT ULNAR VARIANCE
sented the maximum number available from another investigation of skeletal maturation in the adolescent.10 We performed a post hoc assessment of power to evaluate our ability to detect a difference in mean variance of 1.0 mm between any 2 groups analyzed in this study (⫽0.08, ␣⫽0.05). We chose 1 mm as the smallest clinically significant difference for analysis. Provided a standard deviation of approximately 1.1 mm in variance measures, 20 subjects in each group (e.g., females/ males or older females/younger females) were required.
FIGURE 1: Left wrist radiograph of a 12-year-old girl with template superimposed. The template is parallel to the longitudinal axis of the ulna. The ulnar variance in this subject was ⫹2 mm, as agreed on by all 4 raters.
TABLE 2. Age
Mean Variance for Cohort by Skeletal
Skeletal Age
Mean Variance
Cohort
95% Confidence Interval
11
⫺0.5
⫺1.6 to 0.6
12
⫺0.2
⫺0.7 to 0.3
13
⫺0.5
⫺1.0 to 0.0
14
⫺0.7
⫺1.1 to ⫺0.2
15
⫺0.6
⫺1.1 to ⫺0.1
16
⫺0.7
⫺1.5 to 0.1
17
⫺0.6
⫺1.1 to ⫺0.3
18
⫺0.3
⫺0.9 to 0.4
19
0.7
⫺2.5 to 3.9
tests determined the relationship of mean variance between younger and older age groups both within and between genders. We used one-way analysis of variance to compare the mean ulnar variance between each skeletal age for girls and boys. We determined interrater reliability for this explicitly defined measure of ulnar variance. Greatest difference and percentage of exact agreement among the 4 raters were determined. Pearson correlations and Cronbach’s alpha were calculated to compare performance between the raters and intraclass correlation coefficients were calculated using 2-way random effects modeling. The sample size in this study was one of convenience. The 138 standardized wrist radiographs repre-
RESULTS Entire cohort The mean variance for the 138 subjects was – 0.5 mm (SD, 1.2 mm). We detected no significant difference in ulnar variance across skeletal ages in this cohort (p⫽.74). Table 2 presents the mean variance by skeletal age for the entire cohort. Gender The mean variance for the 62 male subjects was – 0.7 mm (SD, 1.2 mm) and the mean variance for the 76 female subjects was – 0.4 mm (SD, 1.1). Table 3 provides a summary of mean variance per skeletal age interval; skeletal age ranges from 11 to 19 years owing to variability between skeletal and chronological ages. The mean skeletal ages for girls and boys was similar (girls, 15.0 y; boys, 15.3 y). Figures 2 and 3 further quantify the relationship among variance, age, and gender. There was no significant difference between the boys and girls (p⫽.16). We also divided the entire cohort into 2 groups, a younger group (ⱕ15 y) and an older group (ⱖ16 y). There were 35 boys and 42 girls in the younger group. The mean variance was – 0.8 mm for the younger boys (SD, 1.0 mm) and – 0.3 mm (SD, 1.0 mm) for the younger girls. The boys’ variance was significantly more ulnar negative than the girls’ (p⫽.03) in this younger group. The older group included 27 boys and 34 girls. The mean variance was – 0.5 mm (SD, 1.5 mm) for the boys and – 0.5 mm (SD, 1.3 mm) for the girls. There was no significant difference between older boys and girls (p⫽.98). Age We evaluated the effect of skeletal age on ulnar variance (Table 3). There was no significant difference in mean variance across the age spectrum for either boys (p⫽.09) or girls (p⫽.94). Similarly, when comparing the younger and older age groups within each gender, there were no significant differences among the boys (p⫽.30) or girls (p⫽.54).
JHS 䉬 Vol A, February
275
ADOLESCENT ULNAR VARIANCE
TABLE 3.
Mean Variance for Boys and Girls, by Skeletal Age
Skeletal Age
Mean Variance Girls
95% Confidence Interval
11
⫺0.8
⫺3.6 to 2.1
Mean Variance Boys 0.0
95% Confidence Interval *
12
0.0
⫺0.6 to 0.6
⫺0.6
⫺2.1 to 1.0
13
⫺1.1
⫺0.8 to 0.6
⫺0.8
⫺1.5 to ⫺0.0
14
⫺0.4
⫺1.0 to 0.2
⫺1.1
⫺2.0 to ⫺0.1
15
⫺0.4
⫺1.3 to 0.5
⫺0.9
⫺1.4 to ⫺0.3
16
⫺0.3
⫺1.2 to 0.6
⫺2.1
⫺4.1 to ⫺0.1
17
⫺0.5
⫺1.2 to 0.2
⫺0.6
⫺1.6 to 0.3
18
⫺0.6
⫺1.6 to 0.3
0.2
⫺0.8 to 1.1
19
NA
0.7
⫺2.5 to 3.9
*Not provided because all 8 (4 raters/2 subjects) observations at this age were 0.00.
FIGURE 2: Box plot of ulnar variance in boys, by skeletal age.
Reliability The greatest difference between any 2 raters was 1 mm. All 4 raters demonstrated exact agreement in 72 cases (52%). Correlations among all 4 raters were high, ranging from rp 0.87 to 0.97 for boys and 0.82 to 0.96 for girls. Measures from each rater were significantly correlated to the other 3 raters (p⬍.001) for each gender. In
addition, interrater reliability judged by Cronbach’s alpha was excellent, at 0.98 for boys and 0.97 for girls. The intraclass correlation for single measures was 0.92 for boys and 0.89 for girls. Discussion Two previously published reports assess ulnar variance in a pediatric or adolescent population.8,9 In both of
JHS 䉬 Vol A, February
276
ADOLESCENT ULNAR VARIANCE
FIGURE 3: Box plot of ulnar variance in girls, by skeletal age.
those articles, the authors used the technique of Hafner et al.6 in gymnasts at risk for early physeal closure due to repetitive trauma. De Smet et al.8 evaluated 156 elite gymnasts with open physes and a mean chronological age of 15.9 years, and DiFiori et al.9 evaluated 59 gymnasts of differing skill levels of a mean chronological age of 9.3 years. Both reported mean ulnar variance values that were significantly greater than the average nonathlete values provided by Hafner et al.6 (data from 535 radiographs of children of all ages). Neither of these studies used skeletal age to assess variance. We believe that the technique of Hafner et al.6 may be ideal for children without ossification of the epiphysis, but it has limitations in the adolescent population. The measurement technique is not familiar to most hand surgeons, and the measurement values are not comparable to those commonly reported in adult subjects. The technique uses the physis location to define the relationship between radius and ulna. Although it provides a constant point for comparison, the technique does not truly compare radius and ulna length. Instead, these comparison points allow assessment in a single patient over time or allow comparison with previously established normal values (as provided in the report by
Hafner et al.). In addition, it is unclear whether the radiographs in the article were standardized for either forearm position or size parameters. Reports on average ulnar variance in the normal adult population, utilize varied measurement techniques. Schuind et al.2 measured ulnar variance, among other radiographic parameters, in 120 adults using the Palmar concentric circle technique, and reported an average variance of – 0.9 mm (range, – 4.2 to 2.3 mm). Nakamura et al.,12 in a Japanese population of 325 subjects and using the Palmer template technique, reported an overall variance mean of 0.20 mm (SD, 1.39 mm), with male subjects averaging – 0.14 mm (SD, 1.34 mm), significantly less than the female subjects’ average of 0.77 mm (SD, 1.27 mm). The authors reported that age correlated with variance; as age increased, variance increased or became more positive. We found an average ulnar variance of – 0.7 for boys and – 0.4 for girls in the adolescent age group, values between the reported average values from Nakamura et al. and Schuind et al. Ulnar variance did not change significantly over the adolescent age spectrum, and the only statistically significant finding was that variance in the younger boys were more negative than variance in the younger girls.
JHS 䉬 Vol A, February
ADOLESCENT ULNAR VARIANCE
There are several limitations to our study. Although we enrolled 138 subjects, additional subjects would lend greater validity to the average values. We chose to use skeletal age as opposed to chronologic age because we believe this to be a more reliable measure of maturation, although it resulted in unequal age distributions within our cohort. In addition, with a preponderance of white subjects, we are unable to comment on any differences attributable to race. Also, we screened subjects for study inclusion by verbal report from parents for any history or current issue related to growth or trauma of the upper extremity. It is possible that a subtle growth abnormality due to previous trauma or other causes may have existed in our subject population, although none of the radiographs suggested an abnormality. Furthermore, we acknowledge that the measurements performed in this study are not possible before ossification of the distal radial and ulnar epiphysis. Finally, we excluded the nonossified epiphysis of both the radius and the ulna from consideration; magnetic resonance imaging would be required to evaluate the contribution of any such cartilage “cap.” As such, we assessed the relative lengths of the ossified epiphyses of the radius and ulna. The traditional technique for ulnar variance assessment, as currently employed in adult subjects, is more familiar to hand surgeons than the technique of Hafner et al.,6 considering both the measurement technique and the variance distance interpretation. The results of this investigation confirm that this more traditional assessment of ulnar variance may be used reliably and accurately in the adolescent subject. The open physis of the radius and ulna in subjects with a well-developed epiphysis did not interfere with the measurement of
277
ulnar variance. The high interrater reliability speaks to the ease of this method for determining ulnar variance in the adolescent population. Employing an explicitly defined measurement technique, raters were consistent across a range of experience and levels of training. REFERENCES 1. Palmer AK, Glisson RR, Werner FW. Ulnar variance determination. J Hand Surg 1982;7:376 –379. 2. Schuind FA, Linscheid RL, An KN, Chao EY. A normal data base of posteroanterior roentgenographic measurements of the wrist. J Bone Joint Surg 1992;74A:1418 –1429. 3. Graham TJ. Surgical correction of malunited fractures of the distal radius. J Am Acad Orthop Surg 1997;5:270 –281. 4. Steyers CM, Blair WF. Measuring ulnar variance: a comparison of techniques. J Hand Surg 1989;14A:607– 612. 5. Coleman DA, Blair WF, Shurr D. Resection of the radial head for fracture of the radial head. Long-term follow-up of seventeen cases. J Bone Joint Surg 1987;69A:385–392. 6. Hafner R, Poznanski AK, Donovan JM. Ulnar variance in children— standard measurements for evaluation of ulnar shortening in juvenile rheumatoid arthritis, hereditary multiple exostosis and other bone or joint disorders in childhood. Skeletal Radiol 1989;18:513–516. 7. Bayne LG, Klug MS. Long-term review of the surgical treatment of radial deficiencies. J Hand Surg 1987;12A:169 –179. 8. De Smet L, Claessens A, Lefevre J, Beunen G. Gymnast wrist: an epidemiologic survey of ulnar variance and stress changes of the radial physis in elite female gymnasts. Am J Sports Med 1994;22: 846 – 850. 9. DiFiori JP, Puffer JC, Aish B, Dorey F. Wrist pain, distal radial physeal injury, and ulnar variance in young gymnasts: does a relationship exist? Am J Sports Med 2002;30:879 – 885. 10. Calfee R, Sutter M, Steffen J, Goldfarb C. Skeletal and chronological ages in American adolescents: current findings in skeletal maturation. J Child Orthop 2010;4:467– 470. 11. Greulich W, Pyle S. Radiographic atlas of skeletal of the hand and wrist. 2nd ed. Stanford, CA: Stanford University Press, 1959. 12. Nakamura R, Tanaka Y, Imaeda T, Miura T. The influence of age and sex on ulnar variance. J Hand Surg 1991;16B:84 – 88.
JHS 䉬 Vol A, February