Accuracy and Reliability of Physical Signs as a Diagnostic Tool for Cervical Cord Compression: A Cross-sectional Study

Abstract

Introduction: Cervical myelopathy diagnosis is based on specific symptoms, physical signs, and imaging findings. However, information on the accuracy and reliability of physical signs, particularly the Wartenberg reflex and the finger escape sign (FES), is lacking. Therefore, this study aimed to assess the validity and reliability of the Hoffmann and Trömner signs, FES, Wartenberg reflex, and combination of any one positive of these four physical signs. Methods: We reviewed the Hoffmann and Trömner signs, FES, and Wartenberg reflex from the medical records of patients with cervical cord compression who underwent surgery. We included those who underwent lumbar spine surgery as controls, except those with upper extremity symptoms or a history of cerebrospinal disease. Subsequently, we calculated the sensitivity and specificity of cervical cord compression. The primary and secondary observers performed two and one trial, respectively, to measure the intra- and interobserver reliabilities. Results: This study included 46 cases and 42 controls. The diagnostic sensitivities for the Hoffmann sign, Trömner sign, Wartenberg reflex, FES, and combination of any one positive were 46%, 72%, 63%, 22%, and 83%, respectively; the diagnostic specificities were 98%, 79%, 95%, 98%, and 79%, respectively; the intraobserver kappa value (κ) was 0.80, 0.82, 0.86, 0.66, and 0.95, respectively; and the interobserver κ was 0.84, 0.51, 0.51, −0.02, and 0.60, respectively. Notably, all κ values, except the interobserver κ for the FES, were obtained with P<0.01. Conclusions: Each physical sign had high specificity but low sensitivity in predicting cervical cord compression. Therefore, they may be useful for definitive diagnosis but not for screening tests. The combination of the four physical signs exhibited improved sensitivity and may be useful for screening tests. However, the results of these physical signs should be carefully interpreted owing to the low level of interobserver reliability.