Journal Article Review, June 2020

JUNE 2020  How accurate is the scratch collapse test for diagnosing carpal tunnel syndrome and cubital tunnel syndrome?

Montgomery K et al: Evaluation of the Scratch Collapse Test for Carpal and Cubital Tunnel Syndrome-A Prospective, Blinded Study. J Hand Surg Am 2020; Jun;45(6):512-517.

The scratch collapse test (SCT), first described in 2008, is a physical examination maneuver to identify compression neuropathy. Sitting face to face, the examiner lightly scratches the patient’s skin over the area of possible nerve compression and then quickly tests resisted shoulder external rotation. The test is positive if the examined shoulder cannot resist the examiner’s force of internal rotation. The mechanism responsible for the SCT is unknown.

The original study showed that the SCT had a 64% sensitivity (accurately identifying those who indeed had the condition) and a 99% specificity (accurately diagnosing those who indeed did not have the condition). The SCT also showed an interrater reliability of .98, meaning that two examiners almost always obtained the same results.

Subsequent studies have reported sensitivities of 24 and 31%, specificities of 60 and 61%, and an interrater reliability of .63 (substantial agreement). A major weakness in all of these studies was that the subjects were already presumed to have or were already confirmed to have carpal tunnel syndrome (CTS) or cubital tunnel syndrome (CuTS) prior to testing with the SCT. And although the SCT is said to be useful in the hands of experienced observers, its performance has not been tested in a blinded fashion. Furthermore, if a diagnostic test is to be widely used, it should be easily taught to and administered by less experienced health care providers. The current study sought to overcome such issues in testing the reliability of the SCT.

METHOD: A third-year physiatry resident and one of two nerve conduction technicians tested 92 patients referred to a hospital clinic for electrodiagnostic studies (EDS). The two examiners performed the SCTs separately and without knowledge of the patient’s reason for referral or history and physical examination findings. The resident completed a brief physical assessment of each patient consisting of the CTS-6 tests (points toward diagnosis with history of night waking, numbness in median nerve distribution, positive Tinel’s, Phalen’s, thenar atrophy, loss of 2PD) and two maneuvers for CuTS (Tinel’s at the elbow and elbow flexion test). Then the technician conducted electrodiagnostic studies. Ninety-one out of 170 wrists met EDS criteria for CTS, and 11 out of 89 elbows met EDS criteria for CuTS.

RESULTS: When the results of the SCTs were compared against the EDS standards for CTS, the sensitivities for the three examiners ranged from 0 to 14%, and the specificities ranged from 68 to100%. When the results of the SCTs were compared against the clinical standards (a positive CTS-6) for CTS, the sensitivities for the three examiners ranged from 1 to 25%, and the specificities ranged from 73 to 100%. The findings for the SCT in accurately identifying CuTS when it was actually present (sensitivity) were similarly low, and the findings for the SCT accurately identifying the absence of CuTS when it was actually absent (specificity) were similarly high.

DISCUSSION: This study, prospective and blinded and testing both symptomatic and asymptomatic limbs, eliminated the possibility of observer bias and preconception and found a far lower sensitivity than originally reported, regardless of whether the EDS or clinical reference standard was used. In other words, by comparison to the reference standards, the SCT poorly identified CTS or CuTS when it was actually present. For comparison, other studies have shown that the sensitivity of the Phalen test averages 46%, the Tinel test at the wrist averages 29%, and the Tinel test at the elbow averages 27%. The specificities of these tests range from 62 to 97%.

COMMENT: I love to see prospective, blinded clinical studies. Randomization makes them even better, but that element was not relevant here. This study was well designed and put the SCT on trial much as it would be used in a general medical clinic or therapy office. Like so many things in medicine (and maybe life in general), what seems so great at introduction often loses some luster when scrutinized. For the SCT, the glowing initial results may have been a case of unblinded examiners unconsciously finding what they were looking for. Furthermore, a test that relies on grading both the examiner’s and the patient’s strength is highly subjective. I do not use the SCT.