Does This Patient Have Carpal Tunnel Syndrome?

In the following patient, the clinician would like to know which items from the patient interview and physical examination accurately predict the diagnosis of carpal tunnel syndrome (CTS):

A 55-year-old woman has difficulty sleeping because of numbness and tingling in her right hand for 6 months. On a hand diagram, she uses a pencil to locate precisely her numbness and tingling over the dorsal and palmar aspects of all 5 fingers, sparing the palm. On inspection the patient has no evidence of thenar atrophy, but thumb abduction is weak on the affected side. Sensory examination using monofilaments and a vibrating tuning fork is normal. Tinel sign is positive, and Phalen sign is negative.

Carpel tunnel syndrome is an important cause of pain and functional impairment of the hand due to compression of the median nerve at the wrist (Figure 1). Patients are usually between their third and fifth decades when diagnosed, and women are affected 3 times as often as men.^{2 - 3} About 0.5% of the general population reports being diagnosed with CTS.² It is likely, however, that few affected patients consult clinicians because population-based studies reveal that about 3% of adults have symptomatic electrodiagnostically confirmed CTS.⁴

In many patients, symptoms are self-limited or resolve with conservative measures, such as splinting the wrist, using anti-inflammatory medication, and modifying their activities. Corticosteroid injection into or near the carpal tunnel results in improvement for 49% to 81% of those affected, although 50% to 86% of those experience recurrence.^{5 - 9} In patients whose condition fails conservative treatment, surgical division of the transverse carpal ligament, either by an open or endoscopic procedure, promptly improves or relieves sensory complaints (dysesthesias) 75% to 99% of the time.^{10 - 18} Permanent complications from surgery occur in less than 1%,¹⁹ but the subsequent recovery often requires leave from work lasting days to several weeks.¹⁸

Many conditions, including pregnancy, rheumatoid arthritis, diabetes mellitus, and previous wrist trauma, are associated with CTS,¹⁹ although histologic sections from the carpal tunnel of most affected patients are normal.^{20 - 21} Many patients have an abnormally high tissue pressure within the carpal tunnel,²² which presumably causes intraneural ischemia that leads to dysesthesias and abnormal results of sensory testing.^{23 - 25}

This article systematically reviews the diagnostic accuracy of bedside findings for CTS. Presentation of this information, however, first requires understanding some of the issues surrounding electrodiagnosis, the current CTS diagnostic standard.

In his original definition of CTS, Phalen²⁶ required patients to have 1 or more of 3 bedside findings: sensory changes restricted to the median nerve distribution of the hand, a positive Tinel sign, and a positive Phalen sign (Table 1). Though electrodiagnosis was not part of Phalen's definition, clinicians now use electrodiagnosis frequently to confirm the diagnosis, and some third-party payers require it before compensating claims.³⁴ Consensus committees from professional societies have endorsed electrodiagnosis as the diagnostic test of choice.^{35 - 36} Diagnostic standards for nerve conduction studies in CTS have been developed, which report sensitivities of 49% to 84% and specificities of 95% to 99%.³⁷

The sensitivity and specificity of electrodiagnosis in CTS need to be carefully interpreted. For the sensitivity calculation, the criterion standard was bedside findings alone (eg, compatible symptoms plus a positive Tinel sign),^{38 - 40} which then begs the question whether electrodiagnosis or bedside findings are the more accurate standard. False-negative test results probably occur because the condition is intermittent⁴¹ or because the patient's symptoms emanate from small, unmyelinated fibers that are invisible to surface electrodes (electrodiagnosis detects only larger myelinated fibers).⁴²

The high specificity figures in these studies are also misleading, being arbitrarily set at 2 SDs above the mean of observations of normal hands. The values of 95% to 99% are based on the assumption that nerve conduction recordings follow a standard gaussian distribution, which has been shown to be inaccurate.^{43 - 44} False-positive test results are well documented when these test thresholds are applied to other populations.^{10 ,45 - 47}

It is well documented that many hand surgeons perform carpal tunnel release successfully in patients with normal electrodiagnostic findings.^{15 ,34 ,48 - 50} Even in patients with positive electrodiagnostic findings who undergo surgery, symptoms usually resolve within days despite nerve conduction abnormalities that persist for months or longer.^{11 ,17 ,42 ,51 - 52}

Nonetheless, most physicians rely on electrodiagnosis as the best available diagnostic standard. Electrodiagnostic studies may help identify other conditions that also cause hand dysesthesias, such as cervical radiculopathy, polyneuropathy, or other median nerve entrapment syndromes.^{41 ,53 - 55} Furthermore, the overwhelming majority of patients in surgical studies have compatible symptoms and electrodiagnostic studies positive for CTS.^{10 ,12 ,17 ,56} Electrodiagnosis may not predict recovery after carpal tunnel release, but neither does any other clinical variable with any certainty. The potential utility of computed tomography, magnetic resonance imaging, and ultrasonography is still being determined, and they remain primarily research tools.^{57 - 61} For these reasons, our review addresses the accuracy of the history and physical examination in diagnosing CTS, as confirmed by electrodiagnostic studies.

Using the MEDLINE database for articles from January 1966 to February 2000, both authors independently used the following search strategy, limited to the English language and human subjects, to retrieve all relevant publications on the diagnosis of CTS in adults: exp carpal tunnel syndrome and exp diagnosis. In addition text word searches were completed for Tinel or Tinels or Hoffman-Tinels, and Phalen or Phalens. Based on review of titles and abstracts, relevant publications were retrieved. To complete the search, the authors reviewed the bibliographies of these articles and retrieved all relevant articles.

To be included in this review, a study had to satisfy the following criteria: (1) the patients presented to a clinician for symptoms suggestive of CTS, (2) the physical examination maneuvers were clearly described, (3) there was an independent comparison with one or more electrodiagnostic parameters (which had to include at least some measurement of motor or sensory nerve conduction), and (4) the authors could extract from figures or tables in the articles the numbers needed to construct 2 × 2 tables and calculate sensitivity, specificity, and likelihood ratios (LRs).

Twelve articles met these criteria and are included.^{27 - 33 ,62 - 66} Thirty articles were excluded: 14 because the control group was asymptomatic,^{67 - 80} 8 because the data were incomplete,^{15 ,49 ,57 ,81 - 85} 4 because the subjects were identified by population surveys,^{45 ,86 - 88} 3 because the criterion standard was unacceptable (ie, electromyography alone,⁸⁹ electrodiagnosis and abnormal monofilament testing,⁹⁰ or criterion standard missing⁹¹ ), and 1 because the examination maneuvers were not clearly defined.⁹²

Sensitivity, specificity, and LRs and their confidence intervals (CIs) were calculated using conventional definitions.⁹³ When a cell of a 2 × 2 table was 0, 0.5 was added to all cells before summarizing the data for a particular test. Our summary measures pooled all the data using the Dersimonian and Laird random-effects model,⁹⁴ which considers both within-study variance and variability among studies. Our test for homogeneity between studies was the effectiveness score, a test of overall accuracy.⁹⁵

Likelihood ratios are the odds that a given finding would occur in a patient with CTS as opposed to one without CTS. If a particular LR, positive or negative, had a value close to 1 that outcome of the test is unhelpful in making diagnostic decisions at the bedside.

Table 1 summarizes how to elicit the physical examination signs of CTS analyzed in this review. When examining thumb strength, the clinician should focus on abduction of the thumb (Figure 2), not flexion or opposition, which sometimes can be accomplished by muscles innervated by nerves other than the recurrent motor branch of the median nerve.^{54 ,59} The Katz hand diagram is a self-administered diagram that depicts both the dorsal and palmar aspect of the patient's hands and arms (Figure 3). Patients use this diagram to mark the specific location of their symptoms, characterizing them as pain, numbness or tingling, or other. Diagrams are then graded as classic, probable, possible, or unlikely to be CTS based on criteria that appear in Figure 3.^{32 ,63}

Few studies have addressed the precision of findings for CTS. In one study, simple agreement was 84% for 2 physicians rating 54 of the Katz hand diagrams.⁶³ In another small study, the interobserver agreement was substantial for Tinel sign (κ=0.77) and Phalen sign (κ=0.65), moderate for vibration (κ=0.40), and fair for motor strength (κ=0.25).⁹⁶ Tinel test, however, is probably much less precise than these data suggest, because the proportion of healthy, asymptomatic hands with a positive Tinel sign ranges from 0%²⁸ to 45%.⁷¹ Some of this variability with Tinel sign may relate to technique; in one study, a greater percussion force increased sensitivity at the expense of specificity.⁸⁹

Table 2 summarizes the studies addressing the diagnostic accuracy of the history and physical examination for CTS. Based on the CIs of LRs, the following findings favor the electrodiagnosis of CTS when they are present in patients who present with hand dysesthesias: decreased sensitivity to pain (hypalgesia) in the median nerve territory (LR, 3.1; 95% CI, 2.0-5.1), classic or probable Katz hand diagram results (LR, 2.4; 95% CI, 1.6-3.5), and weak thumb abduction strength (LR, 1.8; 95% CI, 1.4-2.3). Using a slightly different system for grading hand diagrams, another study also found that the definite or possible hand diagram argued for CTS (LR, 2.1; 95% CI, 1.5-3.0).⁹² In our analysis, 2 findings argued against the electrodiagnosis of CTS: a Katz hand diagram classified as unlikely (LR, 0.2; 95% CI, 0.0-0.7; not shown in Table 2a) and normal thumb abduction strength (LR, 0.5; 95% CI, 0.4-0.7).

The following findings had limited or no value in distinguishing patients with CTS from those without it: the patient's age, presence of bilateral or nocturnal symptoms, thenar atrophy, other sensory abnormalities (2-point, vibration, monofilament), Tinel sign, Phalen sign, pressure provocation test, and the tourniquet test.

Several studies addressed the diagnostic accuracy of combined findings,^{32 ,65 ,90} but no combination consistently proved significantly more helpful than the individual findings themselves. One study did find that the combined finding of a positive Tinel sign and a classic or probable hand diagram was slightly more discriminating (LR, 3.6; 95% CI, 1.6-8.1) than either finding alone (LR, 1.8 for positive Tinel sign and 2.4 for classic or probable hand diagram),³² though this result requires validation given the problems with Tinel sign in other studies.

According to our analysis, several unconventional findings—flick sign, closed fist sign, and square wrist sign—show promise in diagnosing CTS. However, these maneuvers are not widely used and have been tested in only 1 or 2 studies. Two letters to the editor have suggested that the sensitivity of the flick sign is much lower (only 25%-36%) than indicated in Table 2.^{84 - 85} Therefore, before any of these 3 findings can be recommended for clinical practice, further supportive evidence is necessary.

There are several reasons why some findings are not as helpful diagnostically as traditionally thought. Thenar atrophy is probably not useful because it occurs only in long-standing or neglected cases of CTS and can also result from lower cervical radiculopathies or polyneuropathies. Tinel intended his sign to be used in patients after blunt traumatic nerve injury to follow the course of the regenerating nerve.^{30 ,76 ,87} The idea that patients with CTS would also have a stub of continually regenerating nerve at the distal wrist crease seems unlikely, limiting the diagnostic utility of this particular test. Our analysis shows that hypalgesia in the median nerve distribution is a more useful diagnostic finding than abnormalities of other sensory modalities, in part because hypalgesia is a more specific finding. It is not clear why this should be, though it may indicate that the threshold for abnormal results when testing sensation for vibration, 2-point discrimination, and monofilaments is set too low (in one study, for example, 20% of asymptomatic hands also displayed abnormal monofilament results⁷⁶ ).

In our analysis, only results for Tinel sign were heterogeneous. The heterogeneity is not explained by differences in the electrodiagnostic parameters used as criterion standards in the individual studies, variations in examination technique (ie, whether the clinician tapped over the median nerve using his index finger or a reflex hammer), differences in prevalence of CTS in each of the studies (mean prevalence was 57%), differences in the age and sex composition (mean age was 50 years, 77% were women), or by an apparent workup bias. Excluding the 2 studies that account for the heterogeneity^{62 ,64} does not change the summary measure in any meaningful way, and therefore these studies are included in our analysis.

When evaluating patients with hand dysesthesias, the findings most helpful in predicting the electrodiagnosis of CTS are hand symptom diagrams, hypalgesia, and weak thumb abduction strength testing. The square wrist sign, flick sign, and closed fist sign also show promise, but require validation by other investigators. Many traditional findings, including Phalen and Tinel signs, have limited ability to predict the electrodiagnosis of CTS.

The main limitation of the existing literature is the lack of an ideal criterion standard, which complicates all clinical research in the field of CTS. It is also important to note that these data are derived from symptomatic patients presenting to an orthopedic surgeon, physical therapist, or an electrodiagnostic laboratory. There are no data addressing the value of physical diagnosis in patients presenting to a primary care physician with symptoms suggestive of CTS. Our analysis, therefore, is most applicable to patients with severe enough symptoms to warrant such a referral.

Returning to the case presented at the beginning of the article, the findings of a classic hand diagram and thumb abduction weakness both support the diagnosis of CTS. The findings of a normal thenar eminence, a positive Tinel sign, and negative Phalen sign do not contribute significant diagnostic information. Her clinician believed she probably had CTS and chose to manage her symptoms by splinting her wrists and recommending anti-inflammatory medications. If the patient's symptoms fail to improve, nerve conduction testing, additional empiric therapeutic modalities (eg, corticosteroid injections), or referral for surgical assessment should be considered.