Ultrasonographic Measurements of the Ulnar Nerve at the Elbow

Transcription

1 Article Ultrasonographic Measurements of the Ulnar Nerve at the Elbow Role of Confounders Kerry Thoirs, PhD, MMed (Rad), DMU, Marie A. Williams, PhD, Grad Cert (Physio), BAppSci (Physio), Maureen Phillips, MHealth, DCR(R), DMU Objective. The purpose of this study was to identify factors confounding high-resolution ultrasonographic measurements of the ulnar nerve to test their influence when discriminating between limbs affected and unaffected by ulnar nerve entrapment (UNE) at the elbow. Methods. High-resolution ultrasonographic measurements of ulnar nerve dimensions at the elbow were compared between 2 groups of subjects: symptomatic and asymptomatic for UNE. Rank analysis of covariance regression tests were performed to determine whether significant differences existed between the 2 groups. The changing coefficient method (using rank analysis of covariance tests) was used to test for potential confounding effects of age, weight, height, body mass index, sex, limb sidedness, limb handedness, and nerve mobility. These tests were repeated for each measurement while controlling for the identified confounders. Exact 2-tailed Wilcoxon signed rank tests were performed to test for significant differences between measurements of the diameter of the ulnar nerve with the elbow in full extension and full flexion. Results. Age, weight, body mass index, sex, and elbow position were shown to have confounding influences on high-resolution ultrasonographic measurements of the ulnar nerve. No confounding effect was apparent for limb sidedness or dominance. Cross-sectional area and long-axis diameter measurements demonstrated significant differences between nerves with and without UNE after controlling for confounders. Conclusions. Two cross-sectional measurements (area and maximum cross-sectional diameter) of the ulnar nerve, made at the level of the medial epicondyle, were found to be robust discriminators between nerves with and without UNE. In the absence of normative reference values of the ulnar nerve, the contralateral limb may be used as the comparative control. Key words: confounder; measurements; ulnar nerve; ulnar nerve entrapment; ultrasonography. Abbreviations ANCOVA, analysis of covariance; BMI, body mass index; CI, confidence interval; HRUS, high-resolution ultrasonography; UNE, ulnar nerve entrapment Received November 2, 2007, from the School of Health Sciences, University of South Australia, Adelaide, South Australia, Australia. Revision requested December 19, Revised manuscript accepted for publication January 10, We thank Adrian Estermann, PhD, who provided statistical assistance for this study, and the staff of the Department of Neurophysiology, Royal Adelaide Hospital, and Gregory Bain, MBBS, FRACS, FAOrthA, for assistance with recruiting subjects for the study. Address correspondence to Kerry Thoirs, PhD, MMed (Rad), DMU, School of Health Sciences, University of South Australia, City East Campus, North Terrace, Adelaide SA 5000, Australia. kerry.thoirs@unisa.edu.au High-resolution ultrasonography (HRUS) can be used to clarify a suspected diagnosis of ulnar nerve entrapment (UNE) neuropathy at the elbow using assessments of nerve shape, echogenicity, and the appearances of the surrounding anatomy. A small number of studies have provided evidence that HRUS measurements of the ulnar nerve diameter and cross-sectional area can discriminate between individuals with and without UNE. 1 4 Therefore, HRUS measurements offer potential quantitative tools for the diagnosis of UNE. There is a lack of consistency for measurements of the ulnar nerve for nerves both affected and unaffected by UNE in these studies. 5,6 These inconsistencies confuse interpretations of how measurements derived from 2008 by the American Institute of Ultrasound in Medicine J Ultrasound Med 2008; 27: /08/$3.50

2 Ultrasonographic Measurements of the Ulnar Nerve at the Elbow research studies should be applied in clinical practice. The differences in reported values may have arisen because of methodological differences in the research designs, differences in the techniques used to produce the measurements, measurement errors, or differences between the characteristics of the sample populations. The degree to which these factors influence HRUS measurements of the ulnar nerve is unknown. If these factors are not identified or their influence not appropriately corrected when determining reference data, then the reference data may not accurately reflect normal and pathologic values. Previous studies have suggested that factors such as age, sex, and a subluxing or dislocating nerve may have a confounding effect on HRUS measurements of the ulnar nerve at the elbow. 7 9 This study sought to identify factors confounding HRUS measurements of the ulnar nerve and to directly test their impact on measurements likely to discriminate between nerves affected and unaffected by UNE. Materials and Methods This study used an experimental comparative cohort approach in which 2 adult sample populations underwent identical scanning and measurement protocols. Ethics approval was granted from the Human Research Ethics Committee of the University of South Australia and the Research Ethics Committee of the Royal Adelaide Hospital (Adelaide, South Australia, Australia). Informed consent was obtained from all subjects. People with and without symptoms of UNE were recruited over 16 months. Subjects were recruited from the general population, including from the offices of specialist and general medical practitioners, physiotherapists, occupational therapists, and the Department of Neurophysiology, Royal Adelaide Hospital. Rather than allocation of subjects to symptomatic and asymptomatic groups, the right and left upper limbs of each subject were considered independently and either included or excluded from the study. It was possible for a subject to have both limbs allocated to the same group, each limb allocated to a different group, or one limb excluded and the other allocated to a group. Limbs were included in the asymptomatic group if the participant reported no clinical signs of UNE and had no prior diagnosis of UNE in that limb. The chance of including a subclinical case in the asymptomatic group was reduced by excluding people who had been exposed to any events or conditions that were predisposing to ulnar nerve disease (hereditary, metabolic, immune, systematic, and traumatic disorders associated with ulnar neuropathy). Limbs were included in the symptomatic group if the subject had a diagnosis of UNE for that limb by a medical practitioner (clinical and electrophysiologic testing) or if the subject had typical motor and sensory clinical signs and symptoms of UNE, with no prior history predisposing to UNE at a site other than the elbow (eg, shoulder or wrist pain, arthritis, and diabetes). Limbs were excluded from both the asymptomatic and symptomatic groups if the ulnar nerve had prior surgical treatment. All HRUS images were collected at the University of South Australia by a stringent scanning and measurement protocol previously determined to have intra- and inter-rater reliability. 10,11 All HRUS images and measurements of the ulnar nerve were obtained with a 5- to 13-MHz linear array transducer and a Sonoline Antares ultrasound machine (Siemens Medical Solutions, Issaquah, WA). One experienced and accredited sonographer produced 3 images for each limb with the subject in a supine position on an examination couch with the head and neck aligned with the long axis of the body and the upper limb abducted to 90 from the body. Two images were produced with the elbow in a fully extended position, and 1 image was produced with the elbow in a fully flexed position. The nerve was also assessed by the sonographer in terms of its mobility in association with elbow flexion. The nerve was categorized as stable if it remained posterior to the medial epicondyle, subluxing if it moved to the level of the medial epicondyle, or dislocating if it moved anterior to the medial epicondyle. The images were stored on the hard drive of the ultrasound machine and accessed at a later date by a second experienced accredited sonographer who was blinded to the symptomatic status of the limb. The second sonographer made and recorded diameter (short axis, long axis, full extension, and full flexion) and cross-sectional area measurements of the nerve from the 3 images. The 738 J Ultrasound Med 2008; 27:

3 Thoirs et al echogenic rim surrounding the nerve was excluded from all measurements as described in previous studies. 2,3,8,9 The HRUS images and measurements are summarized in Figure 1. Normalization of the data was performed by rank transformation before statistical analysis to meet the assumption of a normal distribution of data required for multiple regression. A crude regression model was produced by rank analysis of covariance (ANCOVA) regression tests to determine whether significant differences existed for HRUS measurements of the ulnar nerve between the asymptomatic and symptomatic limbs (P <.05). Potential confounders (age, weight, height, body mass index [BMI], sex, limb sidedness, limb handedness, and nerve mobility) to the HRUS measurements were then determined by the change-in-estimate method. 12 This method used stepped linear regression (rank ANCOVA) to identify a factor as a confounder if it affected a change of greater than 15% on the crude regression coefficient of the ranked dependent variable. A final regression model, which included all identified confounders, was then produced by rank ANCOVA regression tests to determine whether significant differences existed between the asymptomatic limbs and the symptomatic limbs (P <.05) while controlling for identified confounders. Exact 2-tailed Wilcoxon signed rank tests were performed to determine whether significant differences existed between 2 identical measurements of the diameter of the ulnar nerve made with the elbow in full extension and full flexion. A B Figure 1. Summary of HRUS images and measurements of the ulnar nerve. A, Technique for short-axis measurement. The shortaxis measurement was made from an image that was obtained by scanning in a plane traversing the olecranon process (OL) and the medial epicondyle (ME) with the elbow in full extension. The image clearly showed the osseous floor of the cubital tunnel, the medial epicondyle, the olecranon process, the retroepicondylar groove (RE GR), the joint between the trochlea and olecranon process (JNT), and the ulnar nerve in a cross section. The measurement was made at the widest dimension of the short axis of the nerve. B, Technique for long-axis measurement. The long-axis measurement was made from an image that was obtained by scanning in a plane traversing the olecranon process and the medial epicondyle with the elbow in full extension. The image clearly showed the osseous floor of the cubital tunnel, the medial epicondyle, the olecranon process, the retroepicondylar groove, the joint between the trochlea and olecranon process, and the ulnar nerve in a cross section. The measurement was made at the widest dimension of the long axis of the nerve. The cross-sectional area of the nerve was calculated as a product of the short- and long-axis measurements. C, Technique for full extension and full flexion measurements. Diameter measurements were made from each of 2 longitudinal images of the nerve, 1 obtained with the elbow extended and 1 obtained with the elbow flexed. The nerve diameter measurement was obtained in an identical fashion from each of these images (as shown). The measurement was made at the level of the trochlear-coronoid articulation (asterisk) and did not include the echogenic borders of the nerve. C J Ultrasound Med 2008; 27:

4 Ultrasonographic Measurements of the Ulnar Nerve at the Elbow Results Demographic characteristics of the asymptomatic (No UNE; n = 108 limbs) and symptomatic (UNE; n = 22 limbs) groups are presented in Table 1. In the symptomatic group, sensory symptoms were present in all assessed limbs, and of the 17 limbs tested with nerve conduction studies, 16 had positive findings for UNE. Motor weakness was present in less than 50% of the symptomatic limbs. Descriptive statistics for HRUS measurements are presented in Table 2. Significant differences between all measurements of the ulnar nerve were shown between the 2 subject groups (P <.05, rank ANCOVA crude regression model; Table 3). Age, weight, BMI, and sex were found to have a confounding influence on measurements (Table 4). Weight consistently had the largest predicted confounding effect across all measurements (short axis, 22%; long axis, 19.21%; area, 20.89%; full extension, 42.35%; and full flexion, 64.03%). Limb handedness, limb sidedness, and nerve mobility were found to have no confounding influence on any of the measurements. After controlling for the identified confounders, significant differences were shown between the 2 subject groups (P <.05, rank ANCOVA final regression model; Table 3) for only 2 measurements (long axis and area). Table 1. Demographic Characteristics of Subjects No UNE UNE Characteristic (n = 108 Limbs) (n = 22 Limbs) Height, m 1.69 ± ± 0.1 Weight, kg ± ± BMI, kg/m ± ± 7.83 Age, y ± ± 13.4 Sex Male 33 (25.4) 14 (10.8) Female 75 (57.7) 8 (0.6) Limb Right 54 (41.5) 13 (10) Left 54 (41.5) 9 (6.9) Handedness Dominant 54 (41.5) 13 (10) Nondominant 54 (41.5) 9 (6.9) Mobility Stable 99 (76.2) 19 (14.6) Subluxing 8 (6.2) 3 (2.3) Dislocating 1 (0.7) 0 (0) Values are mean ± SD or number (percent). To investigate arm position as a possible confounder for nerve dimension, exact 2-tailed Wilcoxon signed rank tests were performed to determine whether significant differences existed between ulnar nerve dimensions measured from an image produced with the elbow fully flexed and an image with the elbow fully extended. Significant differences existed between the measurements within the symptomatic group (P =.001; 95% confidence interval [CI], ) and within the asymptomatic group (P <.001; 95% CI, <.001 <.001). This finding shows that elbow position significantly affects the measurement and suggests that elbow position is a confounding factor for the measurement. Discussion A number of studies have shown that HRUS measurements of the ulnar nerve diameter and area at the elbow can differentiate between nerves affected and unaffected by UNE. 1 4 The results of this study concur with those previous findings, but perhaps more importantly, our findings show that HRUS measurements of the ulnar nerve are susceptible to confounding influences arising from physical characteristics of the individual (age, weight, BMI, and sex) within a sample population and the position of the elbow during scanning procedures. Only 2 of the 5 measurements tested could be confidently accepted to discriminate between asymptomatic and symptomatic limbs. The influence of symptoms on the dimensions of the ulnar nerve (long axis and area) was greater than the combined influences of confounders and therefore robust enough to be used to discriminate between symptomatic and asymptomatic nerves without threat from the physical characteristics of the patient. Specifically, these measurements represented the maximum cross-sectional diameter measurement and cross-sectional area measurement of the ulnar nerve taken from an image produced at the level of the medial epicondyle. Conversely, the susceptibility of the other measurements to the effect of confounding variables reduces the dependability of these measurements as discriminators between symptomatic and asymptomatic nerves. 740 J Ultrasound Med 2008; 27:

5 Thoirs et al Table 2. Descriptive Statistics for Ulnar Nerve Measurements No UNE UNE Measurement Mean ± SD n 95% CI Mean ± SD n 95% CI Short axis, cm 0.21 ± ± Long axis, cm 0.38 ± ± Area, cm ± ± Full extension, cm 0.23 ± ± Full flexion, cm 0.19 ± ± The specific position of the elbow during scanning procedures was found to influence ulnar nerve measurements. With 2 identical scanning and measurement protocols, except for elbow position, the diameter measurement of the nerve taken at a site inferior to the medial epicondyle was significantly larger when made in elbow extension than when made in elbow flexion. The difference may be explained by the alteration in the length of the path of the nerve that occurs with elbow flexion 13 and subsequent changes in nerve shape 9 that occur to accommodate the altered nerve path. These findings support previous studies that reported statistically significant differences of measurements of ulnar nerve size at the elbow between male and female subjects 8,9 and between different age groups. 8,9 However, our findings are inconsistent with a previous report that suggested that subject height and subluxing or dislocating ulnar nerves affected HRUS measurements of nerve size. 9 Although due care was taken throughout this experiment to reduce the impact of methodological errors, 2 obvious limitations require consideration. The symptomatic sample included subjects who either had medically confirmed UNE or had symptoms but had not sought medical advice. In the absence of a recognized reference standard test for UNE, it is possible that a proportion of the symptomatic participants did not have UNE but a condition that mimicked Table 3. Crude and Final Regression Models for Rank ANCOVA Tests Crude Regression Model for Rank ANCOVA Tests Final Regression Model for Rank ANCOVA Tests 95% CI for B 95% CI for B Lower Upper Lower Upper Measurement B Limit Limit P B Limit Limit P Short axis * Long axis * * Area <.001* * Full extension * Full flexion * B indicates regression coefficient. *Statistically significant (P <.05). Table 4. Identification of Confounders to Ulnar Nerve Measurements by Change in Estimate Method Potential Confounding Factors Determined by Change From the Crude Regression Coefficient After Linear Regression, % Limb Limb Nerve Measurement Age Height Weight BMI Sex Handedness Sidedness Mobility Short axis 29.35* * 20.12* 16.52* Long axis * Area 18.12* * Full extension * 31.11* 47* Full flexion 38.3* * 45.39* 37.6* *Confounding factors (>15% change). J Ultrasound Med 2008; 27:

6 Ultrasonographic Measurements of the Ulnar Nerve at the Elbow UNE. The inclusion of blinded assessment and stringent reliable measurement protocols with clearly defined selection of imaging planes and placement of measurement calipers should have resulted in a negligible systematic error. The chance of a type 2 error for nonparametric statistical testing needs to be considered for the results of this study, especially because the sample of symptomatic limbs was small (n = 22). The method of analysis (ANCOVA) is likely to have minimized the chances of a type 2 error because this approach and permutation testing (Monte Carlo) within the Wilcoxon signed ranked analysis were used to increase statistical power. 14,15 The cross-sectional area measurement of the ulnar nerve appears to be a good choice as an objective measurement to differentiate between nerves affected and not affected by UNE. The findings of this study suggest a cutoff value for the cross-sectional area of the asymptomatic ulnar nerve of 0.09 cm 2, derived as the upper limit of the 95% CI of the mean, subject to sensitivity and specificity testing. By comparison, the reported and derived values of the upper limit of the 95% CI of the mean from previous studies ranged from 0.06 to 0.1 cm 2. 2,4,8,9 The variability of the reported values for this measurement is confusing when attempting to apply these results as reference values. In view of the results of this study that the cross-sectional area measurement of the ulnar nerve was immune to the confounding influences of patient characteristics, the differences in the data were more likely due to differences in the measurement protocols than differences in the sample populations. Elbow position was found to affect measurements of the ulnar nerve. Not all previous reports provided clear descriptions of subject positioning or the specific measurement protocol used to derive cross-sectional measurements. The cross-sectional area measurement was derived as a product of the short and long axes for this study and 3 previous studies 2,3,9 and by using the trace tool for 2 studies. 4,8 In addition, 1 study 4 included the echogenic rim in the measurement in comparison with the other studies, which did not include this echogenic tissue in the measurement. All of these factors potentially affect the variability of the measurement. In lieu of using reference values, the ultrasonographic technique of using the asymptomatic contralateral limb (nerve) as a comparative control when investigating UNE may be a more realistic option for objective assessment of nerve size. This practice is supported by the findings of this study because limb sidedness and limb dominance did not have a significant confounding effect on any of the measurements. These findings are consistent with other reports comparing HRUS measurements between limbs within individuals with a range of musculoskeletal structures, 16 the ulnar nerve at the wrist, 17 and the ulnar nerve at the elbow. 8,9 Using the contralateral limb as a control is advantageous also because it eliminates the confounding characteristics of sex, age, weight, and BMI and therefore may be applied to a wider range of measurements rather than being limited to measurements at the level of the medial epicondyle only. To further validate this practice, a larger sample of individuals with unilateral UNE needs to be investigated to determine the magnitude of the difference of the size of the ulnar nerve between limbs that would be indicative of UNE. In summary, 2 cross-sectional measurements of the ulnar nerve made at the level of the medial epicondyle, representing the cross-sectional area and maximum width, were found to be robust discriminators for people with and without UNE. The capability of the other tested measurements as discriminators of UNE was threatened by confounding influences of physical characteristics of the subjects and elbow position. An alternate technique to the use of reference values is to use the normal contralateral limb as a comparative control. References 1. Beekman R, Schoemaker MC, van der Plas JPL, et al. Diagnostic value of high-resolution sonography in ulnar neuropathy at the elbow. Neurology 2004; 62: Chiou HJ, Chou YH, Cheng SP, et al. Cubital tunnel syndrome: diagnosis by high-resolution ultrasonography. J Ultrasound Med 1998; 17: Okamoto M, Abe M, Shirai H, Ueda N. Diagnostic ultrasonography of the ulnar nerve in cubital tunnel syndrome. J Hand Surg [Br] 2000; 25: Wiesler ER, Chloros GD, Cartwright MS, Shin HW. Ultrasound in the diagnosis of ulnar neuropathy at the cubital tunnel. J Hand Surg [Am] 2006; 31: J Ultrasound Med 2008; 27:

7 Thoirs et al 5. Bianchi S, Martinoli, C. Elbow. In: Bianchi S, Martinoli C (eds). Ultrasound of the Musculoskeletal System. Berlin, Germany: Springer; 2007: Thoirs K, Williams M, Phillips M. Systematic review of sonographic measurements of the ulnar nerve at the elbow. J Diagn Med Sonogr 2007; 23: Heinemeyer O, Reimers CD. Ultrasound of radial, ulnar, median, and sciatic nerves in healthy subjects and patients with hereditary motor and sensory neuropathies. Ultrasound Med Biol 1999; 25: Jacob D, Creteur V, Courthaliac C, et al. Sonoanatomy of the ulnar nerve in the cubital tunnel: a multicentre study by the GEL. Eur Radiol 2004; 14: Okamoto M, Abe M, Shirai H, Ueda N. Morphology and dynamics of the ulnar nerve in the cubital tunnel: observation by ultrasonography. J Hand Surg [Br] 2000; 25: Thoirs KA, Williams MT. Reliability of measures of the cubital tunnel using high resolution ultrasound: a preliminary study. ASUM Bull 2001; 4: Thoirs K, Williams M, Wilkinson M. Sonographic measurements of the ulnar nerve and the cubital tunnel at the elbow: interobserver reproducibility. Radiography 2005; 11: Greenland S. Modeling and variable selection in epidemiologic analysis. Am J Public Health 1989; 79: Kim B, Date ES, Lee SH, Yoon JS, Hur SY, Kim SJ. Distance measure error induced by displacement of the ulnar nerve when the elbow is flexed. Arch Phys Med Rehabil 2005; 86: Tu YK, Blance A, Clerehugh MJ, Gilthorpe MJ. Statistical power for analyses of changes in randomized controlled trials. J Dent Res 2005; 84: Yu CH. Resampling methods: concepts, applications, and justification. In: Practical Assessment, Research, and Evaluation. PAREonline.net; 2003:8. Available at: Accessed October 10, Schmidt WA, Schmidt H, Schicke B, Gromnica-Ihle E. Standard reference values for musculoskeletal ultrasonography. Ann Rheum Dis 2004; 63: Peeters EY, Nieboer KH, Osteaux MM. Sonography of the normal ulnar nerve at Guyon s canal and of the common peroneal nerve dorsal to the fibular head. J Clin Ultrasound 2004; 32: J Ultrasound Med 2008; 27: