Diagnostic Accuracy of Ultrasonography VS. Electromyography in Carpal Tunnel Syndrome: A Systematic Review of Literature

Ultrasound in Med. & Biol., Vol. 37, No. 10, pp. 1539–1553, 2011 Copyright Ó 2011 World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/$ - see front matter

doi:10.1016/j.ultrasmedbio.2011.06.011

d

Review DIAGNOSTIC ACCURACY OF ULTRASONOGRAPHY VS. ELECTROMYOGRAPHY IN CARPAL TUNNEL SYNDROME: A SYSTEMATIC REVIEW OF LITERATURE SHAWN C. ROLL, JANE CASE-SMITH, and KEVIN D. EVANS The Ohio State University, College of Medicine, Columbus, OH, USA (Received 26 January 2011; revised 29 May 2011; in final form 21 June 2011)

Abstract—A plethora of research investigates sonography vs. electrodiagnostic testing (EDX) for diagnosis of carpal tunnel syndrome (CTS). Through database searches, hand searches and communication with authors, 582 abstracts published from 1999 to 2009 were identified. A comprehensive systematic review process resulted in inclusion of 23 studies. Significant methodologic discrepancies among the studies limited the ability to complete a meta-analysis to identify specific diagnostic thresholds. Instead, the data were reviewed to provide implications for clinical utility of sonography as a screening tool as a compliment to EDX and to suggest continued and future research. The largest cross-sectional area of the median nerve within the carpal tunnel region has high potential for clinical screening, especially in individuals with severe CTS. Identifying swelling of the nerve through comparative measurements, qualitative analysis and Doppler techniques all require further investigation. Screening protocols may be enhanced through exploration of sonography in patients with mild CTS and false-negative EDX. (E-mail: [email protected]) Ó 2011 World Federation for Ultrasound in Medicine & Biology. Key Words: Median nerve mononeuropathy, Carpal tunnel syndrome, Musculoskeletal, Diagnostics, Ultrasonography, Nerve conduction study.

INTRODUCTION

testing, significant variability in sensitivity and moderate rates of false-negatives exist (Jablecki et al. 2002; Lew et al. 2005). In response to the variability of these tests, new methods continue to be developed and tested. Of these new methods, sonography has gained traction as a possible tool for use in clinical evaluation and research in median nerve pathology. Because sonography is pain-free, relatively inexpensive, allows for dynamic evaluation and is becoming more portable, this tool is perfectly situated as a diagnostic tool for musculoskeletal and neurological impairments (Kurca et al. 2008; Roll and Evans 2009). Researchers have investigated the diagnostic ability of sonography for carpal tunnel syndrome; unfortunately, similar to provocative tests and EDX, each of the sonographic techniques and tools suggested for diagnosing CTS varies in utility and quality criteria. Therefore, diagnostic measures remain controversial (El Miedany et al. 2008) and standard measurement techniques have not been determined (Hobson-Webb and Padua 2009). As the use of sonography for evaluation of the median nerve in CTS increases, periodic syntheses of the research evidence can provide direction to research

Carpal tunnel syndrome (CTS) is characterized by motor, sensory and autonomic nerve impairment resulting in numbness and tingling in the median nerve distribution of the hand, hypotrophy of the thenar musculature, dropping of items and decreased sweat function in the hand, with symptom exacerbation at night and after frequent or repetitive use of the wrist and hand (Kuhlman and Hennessey 1997; MacDermid and Doherty 2004; Marklin 1999; Marx et al. 1998; Rempel et al. 1998). Historically, clinical diagnosis of CTS is reliant upon a combination of these patient reported symptoms with positive clinical provocative tests and positive electrodiagnostic (EDX) results (Karadag et al. 2010; Rempel et al. 1998). Positive predictive values and overall sensitivity and specificity have been generally poor for provocative tests (Kuhlman and Hennessey 1997; Mondelli et al. 2001). Furthermore, while specificity may be high for EDX

Address correspondence to: Kevin D. Evans, Ph.D., The Ohio State University, School of Allied Medical Professions, 435 W. 10th Avenue, 340 Atwell Hall, Columbus, OH 43210, USA. E-mail: evans. [email protected] 1539

1540

Ultrasound in Medicine and Biology

and standardize methods and measures. The first such review of literature on the topic synthesized results of 7 studies published between 1991 and 2001 (Beekman and Visser 2003). While the review indicated that sonography was a promising tool for the diagnosis of CTS in place of EDX, standardization of techniques and determination of specific diagnostic values was needed to improve utility. Furthermore, these studies did not compare sonographic measures with EDX, did not differentiate carpal tunnel severity and did not establish cutoff values. A second review of literature was completed (Seror 2008) that compared results of 13 studies completed between 1999 and 2006 (three of the articles were included in the earlier review). Similar to the previous review, Seror indicated a lack of convincing evidence existed to establish cutoff values for diagnostic sonography due to variability across all studies and the methods by which the cut-points were calculated. This, along with numerous other arguments, resulted in a conclusion that sonography was not an acceptable alternative to EDX but complementary at best. While sonography can identify morphologic changes, entrapment and/or lesion of the median nerve, Seror posits that it has limited utility for differential diagnosis in idiopathic CTS. Despite the review by Seror, research reports continue to be published promoting the use of sonography in the diagnosis of CTS. Furthermore, while both reviews indicated that the diagnostic utility of sonography was limited by inconsistency of methods and a lack of identification of specific diagnostic cut-points, it is not clear if current research has moved toward resolution of these issues. Because disagreement exists between previous reviews, discrepancies continue to exist in current research, 3 years have elapsed since the last review and the methodology of previous reviews is not consistent with a full systematic search of literature, a full systematic review of published literature is needed to establish the current utility of sonography in the diagnosis of CTS. The initial purpose of this study was to complete a full meta-analysis of high-quality research to determine accuracy and specific diagnostic thresholds of sonographic measures for diagnosing CTS compared with electrodiagnostic testing. Diagnostic parameters evaluated by the analysis included: (1) cross-sectional area at the pisiform bone, (2) swelling ratio of cross-sectional area at the pisiform bone compared with area at the distal radius or forearm, (3) flattening ratio at the level of the hook of the hamate and (4) palmar bowing of the flexor retinaculum (Visser et al. 2008). Due to a lack of consistency in design and methods within the included studies, we could not statistically combine individual study results. Therefore, this report presents a full systematic

Volume 37, Number 10, 2011

synthesis of the utility of sonography, both for the aforementioned measures and other diagnostic criteria, relative to EDX, in diagnosing CTS.

METHODS A detailed methodology was utilized to ensure a comprehensive review occurred to minimize bias (Shea et al. 2007). Three search methods were used for a full systematic review of both published and unpublished literature, including (1) database search, (2) handsearch and (3) contact with authors. Database searches for studies were conducted using PubMed for MEDLINE, CINAHL, Cochrane, BIOSIS Previews, Health Source Nursing/Academic, PsycINFO, SPORTDiscus and ProQuest Dissertation Abstracts. Standardized search terms included any combination of ‘‘median neuropathy’’ or ‘‘carpal tunnel’’ and ‘‘ultrasonography,’’ ‘‘ultrasound’’ or ‘‘sonography.’’ Search results were limited to studies published in the past 10 years (i.e., 2000–2009); however, studies of any methodologic designs and of any language were included in the original searches. Figure 1 details the inclusion process for this review. Following initial search of all databases, 495 abstracts were obtained. A secondary hand-search occurred by searching bibliographies of review and key articles as well as a hand-search through results obtained from a world wide web search through Google Scholar. The secondary search resulted in the addition of 87 abstracts. Duplicates were removed and 407 abstracts were screened for inclusion in the review.

Fig. 1. Study selection flow chart.

Diagnostic accuracy of ultrasonography vs. electromyography in CTS d S. C. ROLL et al.

To minimize rater bias, the first and third authors screened all abstracts. Screeners independently read each abstract to determine the possible fit of the article with the systematic review. The screeners marked any abstracts that included diagnostic sonography and a reference to CTS or median nerve pathology. Any abstract that was marked for inclusion by either screener was moved forward for review of the full text; therefore, no method was required for reconciliation between results of the two independent screeners. Based on the screening of abstracts, the full-text reports of 135 articles were obtained for rating and final determination of inclusion in the review. Prior to reviewing the full-text articles, inclusion criteria were discussed by the authors and the methodology for rating the articles was agreed upon. Inclusion criteria included: (1) Sonographic measurements of the median nerve for one or more of the following: cross-sectional area, swelling ratio (radius to pisiform or forearm to pisiform), flattening ratio or palmar bowing of the retinaculum; (2) At least one electrodiagnostic (EDX) measure for comparison to sonography or classification of CTS; (3) Linear array ultrasound transducer with frequency of $10 MHz; (4) Clearly described sonography and EDX methodologies consistent with literature; and (5) Research designs that included: a detailed description of subject inclusion and exclusion criteria to ensure idiopathic CTS; adequate sample size for statistical power (e.g., no case study reports); detailed methodologic protocols allowing for replication; and data analysis that reported variability and used statistical tests relevant to the level of data collected. The first and third authors reviewed the full-text of all 135 articles based on these criteria. Each rater tracked responses to the inclusion criteria and made a final recommendation for inclusion or exclusion of the article. Disagreement between raters was settled through consensus and the second author was available to review articles when consensus by the two primary raters could not be reached. For any articles with limited data or unclear methods, attempts were made to contact authors to ensure all data were available for final inclusion in this review. Following rating of all full-text, 23 articles were included into the final review process. The primary reason for article exclusion was due to methodologic limitations that did not meet inclusion criteria or for a lack of data reporting comparative results. Articles that were methodologically sound but did not include EDX testing (n 5 10) or researched the median nerve

1541

in specific diagnoses other than idiopathic CTS (n 5 19) were excluded from the aggregate. Data were extracted from all included articles for methods, measures and outcomes of the studies. During data extraction, the type of extracted data and comparative tables were continually modified such that a valid interpretation and synthesis of the included articles could be completed based on variations in the studies. Data extraction was completed by the first author and random reliability checks were completed by the third author to ensure data had been extracted completely and correctly. All final data tables are referenced within the results section of this systematic review. Important research design components extracted from the studies were patient and equipment information, reference standards, and quality control within the study. The age, gender and total number of individual wrists within the patient and control subject groups was determined and the type of ultrasound equipment and transducer were recorded. Studies were classified based on the use of clinical symptoms and/or EDX as the reference standard. Finally, data were extracted relative to reported quality assurance of equipment and standardization of measurement protocols to ensure reliability and validity of results within the articles. Both sonographic and EDX measurements chosen by the authors within each study were recorded. While the studies often collected numerous ultrasound and EDX measurements, only those measurements that were used in final comparative analysis were extracted for purposes of this review. For each of the measures, the reference standard used within the study was recorded and ultrasound measures were further classified based on how the authors had derived the standard (i.e., previous literature, calculated by receiver operating characteristic [ROC] curves or control mean12 SD). Finally, data were extracted relative to the use of EDX to create dichotomous or severity classifications within subjects and/or for independent diagnostic accuracy based on clinical symptoms. Once all methodologic and measurement data were extracted, articles were reviewed for statistics relative to diagnostic accuracy or of a comparative nature. Sensitivity and specificity were extracted from each article and sorted according to the test measures and reference standard utilized. When these statistics were not reported, but adequate data were available in the article, sensitivity and specificity were calculated. Correlation and comparative statistics between control and patient groups were not recorded; however, articles that reported comparative statistics between ultrasound and EDX measures were recorded. While this standardized methodology was intended to minimize bias, results of this review were limited to

1542

Ultrasound in Medicine and Biology

consideration of studies that included EDX for comparative or classification purposes. Correlation studies that used clinical diagnosis or other methods as a reference standard may provide additional valid information relative to the diagnostic utility of sonography for CTS. Additionally, this review focused on idiopathic CTS and, therefore, did not include studies in subject populations with specific diagnoses or known CTS etiology. RESULTS The findings include a description of the study populations, number of subjects, methodology of diagnostic criteria measured (sonography and reference standard) and comparison results (sensitivity and specificity) from each of the 23 studies. A summary of these data is presented. Subjects Studies with similar inclusion and exclusion criteria were used for analysis. Exclusion criteria included pregnancy, history of previous CTS surgery, presence of known neurologic or systemic disorder that could contribute to CTS (e.g., diabetic neuropathy, thyroid disorders, other polyneuropathies). Additionally, all studies indicated that subjects with anatomical abnormalities such as bifid median nerve and persistent median artery were excluded. Case reports and studies with less than 20 subjects in the patient group were excluded from the review to ensure appropriate power was achieved within each individual study report. One study varied as recruitment included only those subjects with CTS symptoms but negative EDX results (Koyuncuoglu et al. 2005). Methodologic characteristics of all studies included in this review are included in Table 1. Seventeen studies deemed wrists to be independent of the subject factors for statistical analysis, whereas the remaining six studies completed statistical analysis at the level of the individual, only using data from either the dominant hand or the most severely affected hand. Sixteen of the studies utilized a control group. One study recruited only female subjects (Saracgil et al. 2009) whereas the remaining studies included both male and female participants. The percentage of females in each of these studies ranged from 65% to 93% for patient groups and 56% to 87% in control groups. The mean age range for patients was 43 years to 58 years whereas the mean age for controls was 31 years to 55 years. Ultrasound equipment and methods Ultrasound equipment varied greatly across all studies but subject positioning was standardized. Equipment manufacturers included: Acuson, Aloka, BK Medical, Diasonics, Esaote, General Electric (GE),

Volume 37, Number 10, 2011

Philips, Siemens and Toshiba. To ensure adequate image resolution to obtain valid and reliable measures, transducer frequency of $10 MHz was required for a study to be included. All transducers used in the included studies were a linear array and upper limits of frequencies ranged from 10 MHz to 18 MHz. Subjects were all sitting with their forearm supinated and resting on a surface, the wrist in a relaxed, neutral position and the fingers in a relaxed, semi-flexed position. Reports of quality control of equipment and standardization of methods varied greatly across the included studies. No studies reported quality control measures or calibration of the ultrasound equipment. Half of the studies (12 of 23) reported timing of the delay between data collection with sonography and EDX data collection varying from the same day up to 2 weeks. The remaining studies did not report the timing between data collection for each measurement. Similarly, only 11 of the studies reported controlling for skin temperature of subjects during EDX data collection, reporting temperatures between 30
C and 34
C. Sonographic images were collected at various anatomical locations, primarily in a transverse plane. While the majority of studies collected images at multiple locations, only those locations that were used in comparative or diagnostic statistical analysis were recorded for the studies. The primary site for image collection was at the level of the pisiform (61%). In addition to being documented as the level of the pisiform, this level was alternatively described by authors as the ‘‘proximal carpal tunnel’’ or the ‘‘distal wrist crease.’’ Nine studies (39%) obtained a longitudinal image at the radio-ulnar joint of the wrist also described as the ‘‘carpal tunnel inlet’’ or ‘‘proximal to the carpal tunnel.’’ Five studies (21%) obtained images at the level of the hook of the hamate, also described as the ‘‘carpal tunnel outlet’’ and ‘‘distal carpal tunnel.’’ Three studies evaluated the entire region from the proximal tunnel through the distal tunnel and obtained an image in which the median nerve was perceived to have the largest cross-sectional area. Two studies obtained a cross-sectional image of the median nerve in the distal one-third of the forearm (6 cm from wrist crease) and one study at the middle one-third of the forearm (12 cm from wrist crease) as a comparative measure. Longitudinal images of the median nerve within the carpal tunnel region were obtained in two studies for qualitative analysis of the median nerve (Kele et al. 2003; Wang et al. 2008). The utility of EDX within the study design as a comparative measure or grouping mechanism was not consistently reported and measurements and thresholds varied greatly across all studies. At a minimum, all studies utilized sensory conduction velocity (SCV) and distal motor latency (DML) as comparative

Table 1. Characteristics and study design for all studies included in the analysis (listed alphabetically)

Control/wrists (age, % female)

Bayrak et al. 2007

27/47 (43.6 years, 78%)

20/40 (37.5 years, 65%)

El Miedany et al. 2004

78/96 (44.9 years, 65%)

78/156 (44.3 years, 64%)

Hobson-Webb et al. 2008

44/44 (52.4 years, 77%)

18/18 (39.2 years, 44%)

Hobson-Webb and Padua 2009 Site 1 Site 2

46/46 (52.6 years, -)

Subject diagnostic classification system (n) Normal (6) Minimal (8) Mild (6) Moderate (10) Severe (16) Extreme (1) Negative (6) Minimal/Mild (30) Moderate (33) Severe/Extreme (27)

US measures (dx threshold)

US machine/transducer Toshiba PowerVision 7000/10 MHz liner

CSAw CSAp CSAh FR RB

Diasonics Gateway Series/12 MHz linear

CSAw (.10.03 mm2) FRw FRh

–

Philips HDI 5000/15–7 MHz linear

CSAw (.10 mm2) CSAw-f Ratio (.1.4)

–

0–5

Philips HDI 5000/7–14 MHz liner

50/50 (52.8 years, -)

–

0–5

BK Medical Falcon Pro Focus 2202/5–12 MHz liner

CSAw (.14 mm2) CSAf @ 12 cm CSAw-f ratio (.1.5) CSAw (.10mm2) CSAf @ middle 1/3 CSAw-f Ratio (.1.5) - ellipsoid formula

Karadag et al. 2010

54/96 (43.3 years, 93%)

–

Kele et al. 2003

77/110 (52 years, 77%)

33/55 (44 years, 60%)

Normal (49) Mild (22) Moderate (15) Severe (11) –

Klauser et al. 2009

68/100 (57.9 years, 76%)

58/93 (55.1 years, 72%)

Koyuncuoglu et al. 2005

43/59 (43 years, 91%)

15/30 (40.7 years, 87%)

Kurca et al. 2008

37/74 (56 years, 76%) 29/41 (53.0 years, 86%)

25/50 (51 years, 72%) 29/41 (53.0 years, 86%)

Kwon et al. 2008

Esaote MyLab 70/6–18 MHz liner

CSAp

GE Logiq 500M/11 MHz linear

Compression sign CSAw (.11 mm2) CSAp (.11 mm2)

Mild-Moderate (41) Severe-Extreme (59)

Esaote MyLab 90/8–14 MHz or 6–18 MHz linear

Subjects with Clinical Symptoms but Negative EDX –

Philips ATL 1500 HDI/12–5 MHz linear

CSA – maximum within tunnel (.10, 11 or 12 mm2) CSAf CSA change (.2 or 3 mm2) CSAp (.10.5 mm2)

–

Esaote Megas CVX/10 MHz linear Philips HDI 5000/12–5 MHz liner

CSAp (.0.1cm2) CSAp (.10.7 mm2)

EDX measures (dx threshold) Z scores calculated based on a reference value not provided. Amp and velocity (Z , –2) DSL & DML (Z . 2) DSL (.3.6 ms) DSL median-ulnar (.0.4 ms) DML (.4.3 ms) MCV (,49 m/s) SCV (,49 m/s) DML (.4.4 ms) MNL (.2.2 ms) DML (.4.4 ms) Median vs. Ulnar SCV thumb-wrist (.42 m/s) SCV palm-wrist (.37 m/s) DML (.4.0 ms) SCV DML DML (.4.2 ms) SCV (,60 years: ,47 m/s .60 years: ,43 m/s .70 years: ,40 m/s) SCV DML

Quality control EDX-US within 1 week

EDX-US within 3 days

EDX-US on same day Skin temp 34
C Skin temp 34
C –

EDX-US within 2 days Skin temp 32
– 34
C –

EDX-US within 2 weeks

EDX-US within 1 week

DML (.4.6 ms) DSL (.3.8 ms) Sensory amp (,15 uv) DSL (.3.5 ms) DML (.4.0 ms)

Skin temp .30
C EDX-US within 1 week Skin temp .30.5
C (Continued )

1543

DML (.4.2 ms)

Diagnostic accuracy of ultrasonography vs. electromyography in CTS d S. C. ROLL et al.

Subject/wrists (age, % female)

Reference

Control/wrists (age, % female)

Subject diagnostic classification system (n)

Mallouhi et al. 2006

151/206 (58 years, 71%)

–

–

Mondelli et al. 2008

85/85 (46.8 years, 82%)

(controls used only for creating reference values)

Naranjo et al. 2007

68/105 (47.0 years, 82%)

–

Padua et al. 2008

54/54 (53.3 years, 80%)

–

Pastare et al. 2009

66/97 (51 years, 72%)

9/18 (31 years, 56%)

Saracgil et al. 2009

54/100 (55 years, 100%)

25/45 (50 years, 100%)

Swen et al. 2001

63 (52 years, 70%)

20 (49 years, 75%)

Visser et al. 2008

168/168 (52 years, 77%)

137/137 (46 years, 61%)

0 (28) 1 (11) 2 (37) 3 (9) 4 (0) 5 (0) Normal (25) Mild (13) Moderate (30) Severe (37) Negative (3) Minimal (12) Mild (13) Moderate (9) Severe (12) Extreme (5) Negative Minimal Moderate Severe –

–

Normal (26) Minimal (28) Mild (11) Moderate (53) Severe (40)

US machine/transducer Philips HDI 5000/7–15 MHz linear

US measures (dx threshold) Nerve Edema (hypoechoic signal) Max CSA w-h (.0.11 cm2) FR (.3) RB (.2 mm) Hypervascularization (presence of intraneural vessel) CSAw (.10.5 mm2) CSAp (.12.2 mm2) CSAh (.10.1 mm2)

EDX measures (dx threshold)

Quality control

SCV (,62 m/s) DML (.3.9 ms)

Retrospective Case Analysis

SCV (,45.2 m/s) DML (.4.4 ms)

–

CSAw (.10.1 mm2) CSAp (.9.7 mm2) CSAh (.11.5 mm2) FRh (.2.77) RB (.2.76 mm) CSAp (.10 mm2)

DSL (.3.4 ms) DML

–

SCV DML

EDX-US on same day

GE E Logiq book/12 MHz liner

CSAp (.0.09 cm2)

SPL (.3.3 ms) DML (.4.0 ms)

–/VFX 13.5 MHz linear

CSAw (.14 mm2) CSAp (.14 mm2) CSAh (.14 mm2) FRw (.4) FRp (.4) FRh (.4) CSAp/CSAw (.1.5) RB (.3.5 mm) CSAp (.10 mm2) - ellipsoid formula

DML (.4 ms) DSL (.3.41 m/s) SCV (,35.9 mm/s)

EDX-US on same day Skin temp .31
C Skin temp .32
C

Esaote Technos Mp/5–10 MHz linear

GE Logiq 5 Pro/12 MHz liner

BK-Medical Falcon Pro Focus 2202/12–5 MHz linear

Aloka SSD 2000/10 MHz linear

Philips HDI 5000/5–12 MHz linear

CSAw (.0.1 cm2)

DSL (.3.6 ms) DSLu-DSLm (.0.4 ms) DML (.4.3 ms) MCV (,49 m/s) SCV (,49 m/s) DML (.4.2 ms) DSL (.3.5 ms)

–

Skin temp . 32
C

Volume 37, Number 10, 2011

Subject/wrists (age, % female)

Ultrasound in Medicine and Biology

Reference

1544

Table 1. Characteristics and study design for all studies included in the analysis (listed alphabetically) (Continued )

1 – Certainly normal (26) 2 – Probably normal (0) 3 – Mild CTS (30) 4 – Moderate CTS (27) 5 – Severe CTS (24) 74/104 (51 years, 65%) Ziswiler et al. 2005

–

– 45/76 (42.7 years, 87%) 86/148 (49.8 years, 88%) Yesildag et al. 2004

CTS 5 carpal tunnel syndrome; CSAf 5 cross-sectional area at the distal forearm; CSAh 5 cross-sectional area at the carpal tunnel outlet/hamate bone; CSAp 5 cross-sectional area at the proximal carpal tunnel/pisiform bone; CSAw 5 cross-sectional area at the wrist/tunnel inlet/radioulnar joint; DML 5 distal motor latency; DSL 5 distal sensory latency; EDX 5 electrodiagnostics; FR 5 flattening ratio; LCS 5 longitudinal compression sign; MCV 5 motor conduction velocity; MNL 5 mixed nerve latency; RB 5 retinacular bowing; SCV 5 sensory conduction velocity; SPL 5 sensory peak latency; US 5 ultrasonography.

Skin temp . 33
C SCV (,41–53 m/s) DML (.3.9-4.1 ms) SNAP (,5 mV) Largest CSA from inlet to outlet (.10 mm2)

EDX-US within 1 week CSAp (.10.5 mm2) FR

CSAw (.10 mm2) CSAh (.12 mm2) –

Philips HDI 5000/12/5 MHz liner Siemens Sonoline Elegra/13–5 MHz linear Philips ATL 1500 HDI/12 MHz linear Philips ATL 3500/5–12 MHz linear –

26/44 (56 years, 81%) 120/95 (49 years, 82%) Wiesler et al. 2006 Wong et al. 2004

43/86 (36 years, 47%) –

37/61 (44.0 years, 92%) Wang et al. 2008

20/40 (43.7 years, 75%)

–

Acuson Sequoia 512/8–15 MHz linear

CSAp (.9.875 mm2) RB (.2.11 mm) LCS (.1.5) CSAp (.11 mm2)

SCV (,41 m/s) DSL (.3.5 ms) DML (.4.1 ms) DSL (.3.5 ms) DML (.4.5 ms) DSL medial-ulnar (.0.4 ms) DML (.4.0 ms) SCV (,50 m/s) DML (.4.2 ms)

EDX-US within 7 days Skin temp .32
C Skin temp . 32
C EDX-US within 1 week

Diagnostic accuracy of ultrasonography vs. electromyography in CTS d S. C. ROLL et al.

1545

measurements. Diagnostic thresholds for these measurements varied, but were referenced to previous literature or standardized laboratory values. One study modified diagnostic criteria based on the age of the subjects (Kele et al. 2003). Recent studies utilized EDX results to create diagnostic groups by severity (i.e., negative, mild, moderate and severe); whereas, older studies categorized EDX results into dichotomous groups (i.e., negative or positive).

Ultrasound measurements For valid interpretation and comparison of diagnostic statistics, methods for measuring the median nerve were reviewed in each study. Because the outer edge of the epineurium of the median nerve can at times be challenging to determine, only those studies that completed measurements along the inner hyperechoic border of the median nerve were included. Some studies completed each measure only one time while other studies completed measurements multiple times and used the mean or median measure. The most extensive methodology encountered among the studies involved completing five measurements, the highest and lowest measures were eliminated and the remaining three measures were averaged (Kwon et al. 2008). At least one cross-sectional area (CSA) measure was obtained in every study included in this analysis. Two studies (Hobson-Webb and Padua 2009; Swen et al. 2001) calculated CSA by an indirect method using a mathematical ellipsoid formula and all other studies completed a direct trace to obtain CSA of the median nerve. The most recent studies published in 2008 and 2009 used a combination of multiple CSAs to evaluate the swelling of the nerve through a ratio or change score. Ratios were calculated between the wrist and forearm (Hobson-Webb et al. 2008; Hobson-Webb and Padua 2009) and between the pisiform and wrist (Saracgil et al. 2009). The mathematical difference between the maximum CSA within the carpal tunnel region and the CSA in the distal forearm was used as a diagnostic measure in one study (Klauser et al. 2009). Additional measurements primarily collected in study reports between 2000 and 2007 include flattening ratio of the median nerve at multiple levels and anterior bowing of the flexor retinaculum. The flattening ratio was measured in five studies, calculated by dividing the medio-lateral diameter (major axis) by the anteriorposterior diameter (minor axis) of the median nerve. Retinacular bowing, measured in five studies, was calculated by measuring the distance from the apex of the retinaculum to a point perpendicular to a straight line connected the insertion points on the trapezium and hook of the hamate.

Reference Kele et al. 2003 Mallouhi et al. 2006 Wang et al. 2008

Qualitative analysis

Sensitivity

Specificity

Longitudinal compression Longitudinal compression and cross-sectional area Evidence of hyper vascularization Edema in the carpal tunnel Longitudinal compression

50% 89.1%

100% 98%

95%

71%

80% 50%

65% 95.8%

– – r: [20.235–0.240] – – r: [20.129–0.416] Cohen’s K: 0.619 Spearman r: [0.635–0.714] r: [20.092–0.360]

– – – – r: 0.35 – r: 0.26 – – –

– – r: [20.258 – 0.323] – – r: [20.096–0.077]

– – r: [20.174–0.369]

– –

r: [20.062–0.025] r: [20.185–0.070] Spearman r: [0.50– 0.65] r: 0.37 r: [0.408–0.562] r: 0.25 r: 0.80 – –

Cross-sectional area hamate Cross-sectional area pisiform Cross-sectional area wrist

– – – – – r: 0.46 – Spearman r: 0.41 r: 0.40 – – – – r: 0.317 – – – – – – – – r: [0.089–0.403] – – – –

Yesildag et al. 2004 Koyuncuoglu et al. 2005 Ziswiler et al. 2005 Wiesler et al. 2006 Bayrak et al. 2007 Mondelli et al. 2008 Padua et al. 2008 Visser et al. 2008 Hobson-Webb and Padua 2009 Karadag et al. 2010 Pastare et al. 2009 Saracgil et al. 2009

Table 2. Diagnostic accuracy of various qualitative analyses of grey-scale sonographic images

Table 3. Correlation of ultrasound measures to electrodiagnostic testing (EDX)

Diagnostic accuracy of sonography Although similar sonographic measurements were used, the calculation of diagnostic accuracy of sonographic measurements across all studies was confounded by methodologies utilizing different reference standards and calculations of diagnostic thresholds. Correlations between sonographic measures and EDX were mild to moderate (Table 3) and accuracy statistics (i.e., sensitivity and specificity) varied greatly across all the studies. The methods utilized and reported statistics for all quantitative measurements are included in Table 4. Overall, diagnostic accuracy for the various sonographic measurements varied significantly. Sensitivity was noted to be as low as 2% and as high as 100%, with specificity ranging between 47% and 100%. Accuracy ratings were consistently low for measurement of flattening ratios (sensitivity ,65.4%) and retinacular bulging (sensitivity ,79%). Accuracy statistics for measurements of median nerve swelling (i.e., ratio or change) were noted to be exploratory, using purposeful, non-random patient samples to maximize accuracy to test methods. Additionally, the methodologies in studies interrogating nerve swelling varied, with some using changes from the forearm to pisiform and others used changes between the radius and pisiform measures. Therefore, these studies are not able to be easily combined to make conclusions regarding the utility of these measures. The most consistent and best sensitivity and specificity are reported for measures of CSA. The

CSA ratio

In addition to the quantitative data, three studies utilized qualitative measures as an indicator for a diagnosis of CTS (Table 2). Two studies evaluated longitudinal images of the median nerve at the carpal tunnel region for signs of compression. Both studies scored the longitudinal images based on observations of a uniform appearing median nerve vs. flattened or enlarged appearing nerves. The third study reviewed images for signs of edema within the carpal tunnel and recorded the presence of hypervascularization within the median nerve. In this retrospective study, the authors recorded hypervascularization as any color Doppler images with evidence of flow.

– –

Volume 37, Number 10, 2011

Retinacular bulge

Ultrasound in Medicine and Biology

Flattening ratio

1546

Table 4. Sensitivity and specificity of various quantitative grey-scale sonographic measures for CTS with symptoms vs. EDX as reference standards

Flattening ratio

Retinacular bulge

Cross-sectional area Wrist

Symptoms as reference Swen et al. 2001 – – Kele et al. 2003 – – ,z Koyuncuoglu et al. 2005* – – Naranjo et al. 2007 65.4% 47.8% Visser et al. 2008 – – Kurca et al. 2008 – – Kwon et al. 2008 – – Pastare et al. 2009 – – Saracgil et al. 2009 [2%–6%] – EDX as reference (or in combination with symptoms) Yesildag et al. 2004 37.2% 85.5% Wong et al. 2004 – – El Miedany et al. 2004 – – – – Ziswiler et al. 2005y Mallouhi et al. 2006 60% 76% Wiesler et al. 2006 – – Wang et al. 2008 – – Padua et al. 2008* – – Hobson-Webb et al. 2008 – – Mondelli et al. 2008 – – – – Klauser et al. 2009y Hobson-Webb and Padua 2009 Site 1 – – Site 2 – –

– 48.2% – 75% 78% – – – 19%

Cross-sectional area Hamate

CSA ratio

Specificity

Sensitivity

Specificity

Sensitivity

Specificity

Sensitivity

Specificity

– 96.1% – 59.1% 91% – – – –

70% 73.6% 30.5% 86.3% – 93% 66% 62% 33%

63% 98% 96.7% 48% – 96% 63% 100% –

– – – 63.6% – – – – 18%

– – – 78.3% – – – – –

– – – – – – – – – – – – – – – – 6% (wrist to pisiform)

– – – 79% – – – – 2%

– – – 52% – – – – –

– – – – 65% – 77% – – – –

– – – – 68% – 75% – – – –

– – 86% 74% [96.6–98.4] [96.8%–100%] – – – – – – – – – – – – 56.7% – – –

89.9% – – 82% 91% 91% 82% 72.5% 29.4% [94%–100%]

94.7% – – 87% 47% 84% 87.5% 66.7% – – [57%–95%]

– – – – – – – – – 64.7% –

– – – – – – – – – – –

– – – – – – – – – – – – – – – – 100% (forearm to wrist) – – [96%–99%] 100% (forearm to wrist)

– –

– –

37% false-negative 32% false-negative

– –

– –

– –

– –

2% false-negative 32% false-negative

Calculated diagnostic threshold values using sensitivity analysis, receiver operating characteristic curves, or other regression statistics. Calculated diagnostic threshold values based on characteristics of a control group (e.g., 2 standard deviations). Diagnostic thresholds (see Table 1) were selected based on published literature, calculated based on a deviation from a control group or statistically determined based on the data collected within the study. CTS 5 carpal tunnel syndrome; CSA 5 cross-sectional area; EDX 5 electrodiagnostics. * Statistics calculated based on presented data in report. y Used maximum CSA between wrist and hamate levels. z Subjects with symptoms and negative EMG only.

Diagnostic accuracy of ultrasonography vs. electromyography in CTS d S. C. ROLL et al.

Sensitivity Specificity Sensitivity Specificity Sensitivity

Cross-sectional area Pisiform

1547

1548

Ultrasound in Medicine and Biology

largest portion of the studies measured CSA within the mid- portion of the carpal tunnel at the level of the pisiform. Sensitivity and specificity were the highest for CSA measurements within the mid-portion of the carpal tunnel, with a sensitivity of 29.4% to 100% and specificity of 47% to 100%. Studies were subdivided into categories based on the reference standard and based on the methodology by which the diagnostic thresholds were determined. For CSA at the pisiform level, studies that used symptoms of CTS as the reference achieved sensitivity of 30.5%– 93% vs. 29.4%–100% for those utilizing EDX or a combination of the two references (specificity of 48%–100% and 57%–95%, respectively). Studies that set diagnostic thresholds based on previous literature report sensitivity of 30.5%–100% and specificity of 57%–100%. Those studies utilizing data from a control group to determine diagnostic thresholds report sensitivity of 29.4%–93% and specificity of up to 96%. The least variability for diagnostic accuracy was noted when thresholds were statistical calculated utilizing ROC curves or regression within the collected data (sensitivity of 66%–91% and specificity of 47%–87%). DISCUSSION Results of this systematic review of literature expose significant limitations in published research to document clear conclusions regarding the application of sonography to diagnose CTS. There is vast agreement from all the studies that sonography is valuable for detecting anomalies, such as bifid median nerves and persistent median arteries (Karadag et al. 2010), as well as secondary pathologic findings that may contribute to CTS, such as tenosynovitis, crystal or amyloid deposits, ganglia and tumors (Wang et al. 2008). Furthermore, sonography is noted to be less time consuming and better tolerated by patients than EDX (Kwon et al. 2008). While numerous benefits set sonography apart from EDX, a direct link to sonography and specific diagnostic measures for CTS have yet to be documented. Determining the diagnostic utility of sonography has been confounded by a lack of standardization among research methodologies/designs and variability in evaluation and measurement protocols. Results of previous and future research require improved association to pathology and clinical practice. Limitations in reviewed research Inclusion criteria for this review were discussed at length such that high-quality studies were; however, despite the quality, significant limitations are noted in the methods reported in many of these studies that may draw question to the results. The most notable of these

Volume 37, Number 10, 2011

limitations is a lack of reporting of quality control testing for the equipment utilized in any of the studies. Quality assurance measures are frequently cited and have become appropriate practice when utilizing sonography for image guided treatments (Drever and Hilts 2007). With diagnostic accuracy studies, precision of measurement is imperative and quality assurance testing of equipment should be completed on a regular basis, especially with studies that collect data longitudinally. Even if data are collected through a valid and reliable measure, inherent limitations of diagnostic criteria for CTS must not be overlooked. The foundation of CTS diagnosis is a subjective report of symptoms but no diagnostic reference or threshold is universally accepted (Ziswiler et al. 2005). Utilizing EDX as a reference standard in combination with symptoms is most often the process for clinical diagnosis; however, the rate of false-negatives with EDX can be as high as 20% (Yesildag et al. 2004). Knowing this, any false-positives that may be reported with sonographic measurements may in fact be a result of error due to EDX (Wong et al. 2004; Ziswiler et al. 2005). Despite these limitations, EDX continues to be a clinically acceptable means for diagnosis and, therefore, is a plausible reference standard; however, interpretation of the results obtained in these studies should be done with caution. A common problem with staging research is controlling extraneous variables. One of the most critical mediators for valid EDX results is temperature. As the temperature of the extremity increases the conduction velocities will also increase and latencies will decrease (Gooch and Weimer 2007). Only half of the reports indicate having controlled this factor. The skin temperature variability between 30
C and 34
C may have led to variability in the results. Studies failing to report temperature may be valid but caution should be used when interpreting results of these studies as it is unknown if temperature was controlled. Because changes in skin temperature could confound results when utilizing EDX, authors and reviewers of future research publications should be aware of the need to report control of this important variable. Similarly, for valid comparison of the EDX results to sonographic measurements, future research should be designed to minimize the time between sonographic and EDX data collection. Hobson-Webb et al. (2008) and Pastare et al. (2009) were the only studies reviewed to report both an appropriate control of skin temperature and short time lapse between data collection. However, each of these studies was further limited by variability in the comparison groups. Having a comparison group that tended to be younger and male (Hobson-Webb et al. 2008) and with limited subjects (Pastare et al. 2009) could have influenced the results. These were common problems across

Diagnostic accuracy of ultrasonography vs. electromyography in CTS d S. C. ROLL et al.

all studies, with small control groups often being younger and having a higher proportion of male subjects. Both age and gender have been suggested as correlates to the development of CTS; therefore, study designs with subjects and controls matched on age and gender may be more valid than studies with group differences in these variables. Sonography scanning and measurement protocols Combining the data from each of the studies in the review was difficult due to variations in the description of scanning protocols and process for selection of variables and thresholds for data analysis. These factors even varied within studies because clinical practice protocols were non-conformant between data collection sites, creating difficulty in collapsing data and comparing results (Hobson-Webb and Padua 2009). Variability in these factors was widespread across all studies. Successful future research will require standardization of protocols to ensure collection of valid and reliable data that can be generalized and translated to clinical practice. The advancement of technology allows for refined image collection, analysis and measurement. Measurement capabilities have become very efficient and more precise, which can increase the sensitivity when comparing to established diagnostic criteria. The ability to magnify images in postprocessing without losing quality can improve the ability of obtaining a precise measure of a small median nerve (Wiesler et al. 2006). Indirect calculation of the cross-sectional area (CSA) of the median nerve utilizing the ellipsoid formula was completed by Hobson-Webb and Padua (2009) and Swen et al. (2001). This work has been shown to be unreliable as the nerve has a tendency to become displaced by surrounding tissues and does not always take on a perfectly ellipsoid shape. Therefore, calculation with ellipsoid formula can result in poor diagnostic accuracy. For similar reasons, the moderate correlations between ultrasonongraphy and EDX reported by Bayrak et al. (2007) may have been affected by the use of an automatic ellipse to measure CSA. Utilization of a direct trace around the inner hyperechoic border of the median nerve as utilized in the majority of the studies appears to be the most reliable and precise method of measurement. Variability within a measurement can be reduced by calculating an average of repeated measurements. While multiple studies collected only one measurement for analysis, the majority of studies based comparisons on an average of multiple measures. Wiseler et al. (2006) approached the measurement from a conservative standpoint and retained the smallest of three measurements for further analysis. The most extensive measurement process attempted to reduce error by collecting five measurements, deleting the highest and lowest values

1549

and averaging the remaining three measures (Kwon et al. 2008). This methodology seems to account for the fact that human error may persist even in the most highly trained and skilled sonographer. While potentially time consuming, this comprehensive method could be especially important to reduce error in the analysis of images that may be of low quality or in patients that are difficult to scan. Standardization of sonography scanning protocols relies on very clear descriptions of anatomical landmarks. Anatomical landmarks chosen to define the carpal tunnel were inconsistent among the studies in this review. Mondelli et al. (2008) considered the distal edge of the radial-ulnar joint as the inlet to the carpal tunnel, whereas other studies described the inlet of the tunnel at the location of the pisiform bone (Visser et al. 2008). While the pisiform is considered a bone within the proximal row of carpal bones, the arching of these bones tends to place the pisiform slightly distal. At this slightly distal position of the pisiform, the median nerve will have already crossed under the transverse ligament. Therefore, the carpal tunnel region may be best described as the inlet at the distal edge of the radial-ulnar joint, the midcarpal tunnel at the level of the pisiform, and the distal outlet of the tunnel at the hamate bone. Standardization of these descriptions will be helpful in discussion and replication of research results but the clinical significance of various sonographic measurements also requires consistency. Measurement of CSA in the distal portion of the carpal tunnel has low diagnostic accuracy (sensitivity [18%–65%]). These low statistics may be a result of the challenge of obtaining reliable images in this region due to poor signal quality created by the curved palm and flat transducer and the deep oblique path of the nerve (Moran et al. 2009). The ability to accurately visualize and measure the nerve in this region has been reported to be as low as 42%, indicating that this measure is not reliable (Kele et al. 2003). Similarly, retinacular bulge measures at this location could also experience lower sensitivity. Studies that measured CSA immediately proximal to the carpal tunnel at the inlet had improved diagnostic sensitivity, but still slightly lower accuracy statistics than those measuring CSA within the carpal tunnel at the pisiform level. Measurement of CSA within the carpal tunnel, at the level of the pisiform, appears to be the most widely referenced and most accurate measure. Across these studies, three methods were used for determining diagnostic thresholds: (1) previous literature/lab standards, (2) control group comparison and (3) statistical calculation. Because this review and all previous literature reviews indicate that no specific thresholds have been determined, referencing previous literature to set a specific threshold is erroneous and may lead to invalid

1550

Ultrasound in Medicine and Biology

statistical analysis. Thresholds calculated based on a control group tended to be much higher. Determination of thresholds based on control group data is logical; however, due to the idiopathic state of CTS, it is challenging to determine a true, nonpathologic control group that is 100% matched to the patient group on all possible risk factors (e.g., age, gender, body mass index [BMI], wrist ratio). Furthermore, because of the precision of the measurements and the small changes that are being documented in mild cases of CTS, calculation of two standard deviations from a control group mean as a threshold may not correctly capture CTS cases of all severities. Clinical severity skewed toward mild cases in Mondelli et al. (2008) and lack of accounting for severity in Saracgil et al. (2009) may explain low sensitivity (29.4%–33%) with calculated respective thresholds of 12 mm2 and 14 mm2 for CSA. Further limits in the later study included measurements completed in cm, extrapolated to mm, for statistical analysis and the admitted lack by experience of those collecting data. Statistical calculation of thresholds based on the data collected in both a patient and control group, accounting for severity of CTS, had the best accuracy results. Thresholds calculated with ROC curve analysis or other regression methods ranged from 9.7 mm2 to 11 mm2, resulting in an average sensitivity of 81% and specificity of 70%. Two studies that accounted for the severity of CTS in these results indicated that instead of one threshold, multiple thresholds could be utilized to better rule in or rule out CTS. Naranjo et al. (2007) noted that 100% sensitivity could be achieved by raising the threshold to 13 mm2 and Ziswiler et al. (2005) suggest that CSA within the carpal tunnel smaller than 8 mm2 had adequate power to rule out CTS and measures larger than 12 mm2 had sufficient power to rule in CTS. These results further support the importance of pathologic severity and support the concept that one diagnostic threshold may not be clinically useful. Anthropometry is an additional important component in development and classification of many musculoskeletal disorders and may be a component to consider when attempting to diagnose CTS. Comparison of these CSA to current diagnostic standards is fraught with limitations that could be due to the variability that exists among subjects and not necessarily attributed to the measurements. Factors such as body mass index and wrist size may influence the nerve, not only in pathology, but in a normal state. Measuring the naturally occurring status of the individual’s median nerve for comparison may provide a more reliable measurement of pathology. While only three studies have evaluated this relationship of CSA in the forearm to size of the median nerve the carpal tunnel, these studies show promising results. Klauser et al. (2009) indicate that the calculation of the

Volume 37, Number 10, 2011

CSA change has been shown to improve accuracy over the singular measurement. Calculation of a ratio for these two measures may be an alternative methodology that has only been preliminarily tested (Hobson-Webb, Massey 2008, Hobson-Webb and Padua 2009). Another promising new method that requires further investigation involves the interrogation of vascularity within the median nerve. It has been suggested that the odds of a correct diagnosis of CTS may be 16 times better utilizing color Doppler to identify hypervascularization over other grey-scale measurements (Mallouhi et al. 2006). The authors noted that hypervascularization was present within the nerve for those patients with CTS symptoms, even when the nerve itself did not show signs of swelling or compression. This may indicate pathologic intraneural vascularization within and around the nerve before other swelling and edema of the connective tissues. Because this study was completed as a retrospective case analysis, conclusions that have been drawn regarding hypervascularization should be interpreted cautiously. Data collection and equipment settings were not controlled or described in the research methods, which could be a source for error in the results. However, as sonography continues to improve and be utilized for diagnosis of CTS, further interrogation of the intraneural vasculature of the nerve is appropriate and Mallouhi et al. (2006) provide a good template for developing future imaging protocols. Clinical utility The primary etiology of idiopathic CTS is believed to be due to compression on the median nerve, by either direct flattening or indirect reduction of space within the carpal tunnel. Low correlations (Wang et al. 2008) and sensitivity among studies in this review provided no evidence to support measurement of flattening of the nerve for clinical diagnosis. However, qualitative observation of compression of the median nerve may provide useful information for prevention of CTS in pre-pathological or very acute stages. Similarly, while measurement of bulging of the flexor retinaculum due to swelling is limited by the ability to obtain consistent images in the distal portion of the carpal tunnel (Kele et al. 2003, Moran et al. 2009), qualitative evaluation of edema and other pathology within the carpal tunnel region leading to compression of the median nerve may be useful in identifying etiology. Cross-sectional area of the median nerve had the most stable measures of sensitivity but the location of CSA measurement requires further standardization in clinical practice. The progression of median nerve swelling and its relationship to the stage of CTS development is not well known. Therefore, it may be erroneous to assume that this swelling will occur at the same location

Diagnostic accuracy of ultrasonography vs. electromyography in CTS d S. C. ROLL et al.

within all individuals and instead may occur at multiple or varied levels (Wong et al. 2004). Most literature has concluded that this swelling occurs within the tunnel at the level of the pisiform. However, if the carpal tunnel is not sufficiently large to allow for swelling of the nerve or other anthropometric or individual factors exist, excess swelling may occur immediately proximal to the tunnel. Multiple studies in this review suggested that measurement of the largest CSA within the entire carpal tunnel region may be a more accurate measure than focusing specifically at the level of the pisiform (Klauser et al. 2009; Mallouhi et al. 2006; Ziswiler et al. 2005). Instead of standardization of the data collection protocol to include measurement at specific bony landmarks, these landmarks could provide end points for a range of interrogation. Visualization of the largest CSA of the median nerve could be determined during completion of a dynamic scan through the region bordered by the radial-ulnar joint proximally and the hamate bone distally. Additional recommendations for this scanning protocol are needed to ensure that true cross-sections are being visualized and oblique scans are not being obtained (Kwon et al. 2008). Just as one measurement location may not be appropriate, this review supports the conclusion that a rigid diagnostic threshold for sonographic measurements does not appear to be valuable in the clinical setting (Wiesler et al. 2006). The size of the median nerve and swelling occur on a continuum and, therefore, diagnostics cannot be easily related to specific cutoff values (Wong et al. 2004). This continuum may shift based on individual factors or clinical differences leading to different outcome measures. Evaluating probabilities based on variations in these factors may lead to a sliding scale of flexible cutoffs to be applied to various situations, thereby improving clinical diagnosis (Ziswiler et al. 2005). The severity and stage of disorder progression appear to be the primary factors to consider when utilizing ultrasonography to make a clinical diagnosis. Although one clear cutoff was not identified, statistical analysis indicates that diagnostic accuracy increases as the CSA continues to increase beyond 10 mm2. In mild or relatively acute cases that tend to be closer to this statistically calculated threshold, more variability is noted in the data and diagnostic accuracy declines (Hobson-Webb and Padua 2009). One study has attempted to gain better understanding of this phenomenon, noting increased swelling in the median nerve of up to 30.5% of individuals with mild symptoms but negative EDX (Koyuncuoglu et al. 2005). This leaves a significant portion of symptomatic individuals without diagnosis, indicating further investigation of clinical screening protocols is needed. An evaluation technique such as sonography that identifies acute changes in connective tissue and vascular

1551

flow may be a more useful tool than EDX, which indicates more chronic nerve damage. When EDX testing is normal, sonography may indicate swelling in individuals who have not yet developed secondary dysfunction or damage of the nerve fibers (Moran et al. 2009). Early identification of these changes can allow for intervention to prevent nerve damage that occurs chronic stages. While data reported by Koyuncuoglu et al. (2005) indicates that sensitivity may remain low for the sonography measures in this population, further study is required. Because increased CSA is related to symptoms and declining hand function (Padua et al. 2008), individuals with large CSA, such as greater than 12 mm2 (Ziswiler et al. 2005) or 13 mm2 (Naranjo et al. 2007), may be diagnosed with mild or acute CTS that has not yet become pathologic based on EDX. While sonography may not be an absolute alternative to EDX, the two evaluation procedures may be complementary when utilized appropriately (Karadag et al. 2010; Naranjo et al. 2007; Padua et al. 2008). Individuals with significantly large CSA (e.g. .14 mm2) could be spared the time, expense and discomfort of EDX and only those individuals with mild cases and smaller CSA would require further evaluation (Mondelli et al. 2008; Naranjo et al. 2007; Pastare et al. 2009). CONCLUSIONS Continued research is necessary to better understand the utility of sonography for diagnosing CTS. Initial recommendations for research focus on implementing research designs that adequately control mediating factors and use standardized protocols to ensure that error is minimized. Prospective research designs should involve regular quality assurance checks on ultrasound equipment and appropriate control of measurement error due to skin temperature. For accurate comparison, data collection for sonographic measures and EDX should be completed on the same day and controls should be age and gender matched. Standardization in anatomical locations of data collection is needed to improve consistency in research and clinical reports. Future research studies should eliminate measurements that are no longer valid and begin investigating new techniques to determine clinical utility. Calculation of flattening of the median nerve does not appear to provide clinically useful information, and instead of measuring CSA at one specific location, the largest CSA within the entire carpal tunnel region may be a more appropriate technique. Furthermore, utilization of the CSA at the forearm as an internal control for observing swelling of the median nerve has potential to be a more precise diagnostic indicator. Qualitative observations of compression, waist-line effects or other

1552

Ultrasound in Medicine and Biology

observations of non-linear longitudinal appearance (Wang et al. 2008), as well as edema and fluid retention, both within the nerve and in the surrounding regions, requires further investigation (Klauser et al. 2009). Identifying changes in intraneural vasculature and flow with color or spectral Doppler may be useful for qualitatively or quantitatively measuring inflammation or compression (Evans et al. 2010; Mallouhi et al. 2006). Finally, since sonography allows for dynamic visualization of structures, observation of general mobility patterns of the median nerve may be a potentially useful parameter (Klauser et al. 2009). Despite the lack of solid research evidence for the independent use of sonography for diagnosis of CTS, screening with sonography appears to enhance current techniques. Sonographic evaluation of the median nerve may be best used as a screening tool in the first step of individuals with suspected CTS. As the CSA of the nerve increases to a severe stage, the accuracy of diagnosis increases; therefore, patients with severe swelling would not require the more expensive, uncomfortable and invasive EDX. Further investigation is required to identify the best combination of sonographic techniques for screening and diagnostic purposes. Studies are needed investigate diagnostic techniques for individuals with EDX falsenegatives, as well as longitudinal evaluation of individuals with sonographic false-positives to determine possible etiology of CTS (Wong et al. 2004). REFERENCES Bayrak IK, Bayrak AO, Tilki HE, Nural MS, Sunter T. Ultrasonography in carpal tunnel syndrome: Comparison with electrophysiological stage and motor unit number estimate. Muscle Nerve 2007;35: 344–348. Beekman R, Visser LH. Sonography in the diagnosis of carpal tunnel syndrome: A critical review of the literature. Muscle Nerve 2003; 27:26–33. Drever L, Hilts M. Daily quality assurance phantom for ultrasound image guided radiation therapy. J Appl Clin Med Phys 2007;8:2467. El Miedany Y, Ashour S, Youssef S, Mehanna A, Meky FA. Clinical diagnosis of carpal tunnel syndrome: Old tests-new concepts. Joint Bone Spine 2008;75:451–457. El Miedany YM, Aty SA, Ashour S. Ultrasonography versus nerve conduction study in patients with carpal tunnel syndrome: Substantive or complementary tests? Rheumatology (Oxford) 2004;43: 887–895. Evans KD, Roll SC, Li X, Sammet S. A holistic evaluation of risk factors for work-related musculoskeletal distress among asymptomatic sonographers performing neurosonology: A pilot study. J Diagn Med Sonography 2010;26:64–78. Gooch C, Weimer L. The electrodiagnosis of neuropathy: Basic principles and common pitfalls. Neurol Clin 2007;25:1. Hobson-Webb LD, Massey JM, Juel VC, Sanders DB. The ultrasonographic wrist-to-forearm median nerve area ratio in carpal tunnel syndrome. Clin Neurophysiol 2008;119:1353–1357. Hobson-Webb LD, Padua L. Median nerve ultrasonography in carpal tunnel syndrome: Findings from two laboratories. Muscle Nerve 2009;40:94–97. Jablecki CK, Andary MT, Floeter MK, Miller RG, Quartly CA, Vennix MJ, Wilson JR. Practice parameter: Electrodiagnostic studies in carpal tunnel syndrome. Report of the American

Volume 37, Number 10, 2011 Association of Electrodiagnostic Medicine, American Academy of Neurology, and the American Academy of Physical Medicine and Rehabilitation. Neurology 2002;58:1589–1592. Karadag YS, Karadag O, Cicekli E, Ozturk S, Kiraz S, Ozbakir S, Filippucci E, Grassi W. Severity of Carpal tunnel syndrome assessed with high frequency ultrasonography. Rheumatol Int 2010;30: 761–765. Kele H, Verheggen R, Bittermann H, Reimers CD. The potential value of ultrasonography in the evaluation of carpal tunnel syndrome. Neurology 2003;61:389–391. Klauser AS, Halpern EJ, De Zordo T, Feuchtner GM, Arora R, Gruber J, Martinoli C, Loscher WN. Carpal tunnel syndrome assessment with US: value of additional cross-sectional area measurements of the median nerve in patients versus healthy volunteers. Radiology 2009;250:171–177. Koyuncuoglu HR, Kutluhan S, Yesildag A, Oyar O, Guler K, Ozden A. The value of ultrasonographic measurement in carpal tunnel syndrome in patients with negative electrodiagnostic tests. Eur J Radiol 2005;56:365–369. Kuhlman KA, Hennessey WJ. Sensitivity and specificity of carpal tunnel syndrome signs. Am J Phys Med Rehabil 1997;76:451–457. Kurca E, Nosal V, Grofik M, Sivak S, Turcanova-Koprusakova M, Kucera P. Single parameter wrist ultrasonography as a first-line screening examination in suspected carpal tunnel syndrome patients. Bratisl Lek Listy 2008;109:177–179. Kwon BC, Jung KI, Baek GH. Comparison of sonography and electrodiagnostic testing in the diagnosis of carpal tunnel syndrome. J Hand Surg Am 2008;33:65–71. Lew HL, Date ES, Pan SS, Wu P, Ware PF, Kingery WS. Sensitivity, specificity, and variability of nerve conduction velocity measurements in carpal tunnel syndrome. Arch Phys Med Rehabil 2005;86:12–16. MacDermid JC, Doherty T. Clinical and electrodiagnostic testing of carpal tunnel syndrome: A narrative review. J Orthop Sports Phys Ther 2004;34:565–588. Mallouhi A, Pulzl P, Trieb T, Piza H, Bodner G. Predictors of carpal tunnel syndrome: Accuracy of gray-scale and color Doppler sonography. AJR Am J Roentgenol 2006;186:1240–1245. Marklin RW. Biomechanical aspects of CTDs. In: Karwowski W, Marras WS, (eds). The occupational ergonomics handbook. Boca Raton, FL: CRC Press; 1999. p. 795–832. Marx RG, Hudak PL, Bombardier C, Graham B, Goldsmith C, Wright JG. The reliability of physical examination for carpal tunnel syndrome. J Hand Surg [Br] 1998;23:499–502. Mondelli M, Filippou G, Gallo A, Frediani B. Diagnostic utility of ultrasonography versus nerve conduction studies in mild carpal tunnel syndrome. Arthritis Rheum 2008;59:357–366. Mondelli M, Passero S, Giannini F. Provocative tests in different stages of carpal tunnel syndrome. Clin Neurol Neurosurg 2001;103:178–183. Moran L, Perez M, Esteban A, Bellon J, Arranz B, del Cerro M. Sonographic measurement of cross-sectional area of the median nerve in the diagnosis of carpal tunnel syndrome: correlation with nerve conduction studies. J Clin Ultrasound 2009;37:125–131. Naranjo A, Ojeda S, Mendoza D, Francisco F, Quevedo JC, Erausquin C. What is the diagnostic value of ultrasonography compared to physical evaluation in patients with idiopathic carpal tunnel syndrome? Clin Exp Rheumatol 2007;25:853–859. Padua L, Pazzaglia C, Caliandro P, Granata G, Foschini M, Briani C, Martinoli C. Carpal tunnel syndrome: Ultrasound, neurophysiology, clinical and patient-oriented assessment. Clin Neurophysiol 2008; 119:2064–2069. Pastare D, Therimadasamy AK, Lee E, Wilder-Smith EP. Sonography versus nerve conduction studies in patients referred with a clinical diagnosis of carpal tunnel syndrome. J Clin Ultrasound 2009;37: 389–393. Rempel D, Evanoff B, Amadio PC, de Krom M, Franklin G, Franzblau A, Gray R, Gerr F, Hagberg M, Hales T, Katz JN, Pransky G. Consensus criteria for the classification of carpal tunnel syndrome in epidemiologic studies. Am J Public Health 1998;88: 1447–1451. Roll SC, Evans K. Feasibility of using a hand-carried sonographic unit for investigating median nerve pathology. J Diagn Med Sonography 2009;25:241–249.

Diagnostic accuracy of ultrasonography vs. electromyography in CTS d S. C. ROLL et al. Saracgil SN, Karatas M, Yerli H, Isiklar I, Karadeli E. Diagnostic significance of ultrasonography in carpal tunnel syndrome and comparison with electrodiagnostic tests. Turkish J Phys Med Rehabil 2009;55:13–18. Seror P. Sonography and electrodiagnosis in carpal tunnel syndrome diagnosis, an analysis of the literature. Eur J Radiol 2008;67: 146–152. Shea B, Grimshaw J, Wells G, Boers M, Andersson N, Hamel C, Porter A, Tugwell P, Moher D, Bouter L. Development of AMSTAR: A measurement tool to assess the methodological quality of systematic reviews. BMC Med Res Methodol 2007;7:10. Swen WA, Jacobs JW, Bussemaker FE, de Waard JW, Bijlsma JW. Carpal tunnel sonography by the rheumatologist versus nerve conduction study by the neurologist. J Rheumatol 2001;28:62–69. Visser LH, Smidt MH, Lee ML. High-resolution sonography versus EMG in the diagnosis of carpal tunnel syndrome. J Neurol Neurosurg Psychiatry 2008;79:63–67.

1553

Wang LY, Leong CP, Huang YC, Hung JW, Cheung SM, Pong YP. Best diagnostic criterion in high-resolution ultrasonography for carpal tunnel syndrome. Chang Gung Med J 2008;31:469–476. Wiesler ER, Chloros GD, Cartwright MS, Smith BP, Rushing J, Walker FO. The use of diagnostic ultrasound in carpal tunnel syndrome. J Hand Surg Am 2006;31:726–732. Wong SM, Griffith JF, Hui AC, Lo SK, Fu M, Wong KS. Carpal tunnel syndrome: Diagnostic usefulness of sonography. Radiology 2004; 232:93–99. Yesildag A, Kutluhan S, Sengul N, Koyuncuoglu HR, Oyar O, Guler K, Gulsoy UK. The role of ultrasonographic measurements of the median nerve in the diagnosis of carpal tunnel syndrome. Clin Radiol 2004;59:910–915. Ziswiler H, Reichenbach S, Vogelin E, Bachmann LM, Villiger PM, Juni P. Diagnostic value of sonography in patients with suspected carpal tunnel syndrome: A prospective study. Arthritis Rheum 2005;52:304–311.