Electrodiagnosis of ulnar neuropathy at the elbow (Une): a Bayesian approach.

INTRODUCTION: In ulnar neuropathy at the elbow (UNE), we determined how electrodiagnostic cutoffs [across-elbow ulnar motor conduction velocity slowing (AECV-slowing), drop in across-elbow vs. forearm CV (AECV-drop)] depend on pretest probability (PreTP).

METHODS: Fifty clinically defined UNE patients and 50 controls underwent ulnar conduction testing recording abductor digiti minimi (ADM) and first dorsal interosseous (FDI), stimulating wrist, below-elbow, and 6-, 8-, and 10-cm more proximally. For various PreTPs of UNE, the cutoffs required to confirm UNE (defined as posttest probability = 95%) were determined with receiver operator characteristic (ROC) curves and Bayes Theorem.

RESULTS: On ROC and Bayesian analyses, the ADM 10-cm montage was optimal. For PreTP = 0.25, the confirmatory cutoffs were >23 m/s (AECV-drop), and <38 m/s (AECV-slowing); for PreTP = 0.75, they were much less conservative: >14 m/s, and <47 m/s, respectively.

CONCLUSIONS: (1) In UNE, electrodiagnostic cutoffs are critically dependent on PreTP; rigid cutoffs are problematic. (2) AE distances should be standardized and at least 10 cm.