Abstract
INTRODUCTION: We compared sensory nerve conduction studies (NCS) using surface and near-nerve recording electrodes in 53 patients with clinical probable painful neuropathy. Our aim was to validate the use of both recording techniques in that limited patient group. METHODS: Patients had sensory NCS using two established recording methods and quantitative sensory tests (QST). We compared normalised amplitudes of sensory sural nerve action potentials (SNAP) and sensory thresholds and used receiver operated curve (ROC) analysis of absolute SNAP amplitudes to find discriminatory levels predicting abnormal sensory thresholds. RESULTS: Mean sural SNAP z-scores differed depending on recording techniques (surface -1.0: SD 1.9; near-nerve -2.5: SD 1.7) with a numeric mean difference of -1.49 (Bland-Altman test: CI -1.872 to -1.12) with surface technique giving the z-value closest to zero. We documented a significant bias between the methods. Fifteen patients (28.3%) and 30 (56.6%) patients had abnormal results, respectively (χ2 test: p<0.001).Sural SNAP amplitudes correlated significantly with vibration thresholds using the near-nerve (p<0.02) but not using the surface technique (p=0.11).ROC analysis gave an optimal discriminative value of SNAP amplitudes for each QST measure, which were similar to our lower limit of normal values from investigating normal controls using near-nerve but not surface recording. CONCLUSION: In patients with probable painful neuropathy, choosing sensory NCS technique introduces a bias in the diagnostic outcome. Differences in test performance suggest that using a normal sural NCS alone to delineate small fibre neuropathy from mixed neuropathy could result in poorly defined diagnostic groups.