Modeling nerve conduction criteria for diagnosis of diabetic polyneuropathy

Abstract

Introduction: In this study we aimed to determine which criteria are valid for nerve conduction (NC) diagnosis of typical diabetic sensorimotor polyneuropathy (DSPN).

Methods: Eight criteria were assessed from among diabetes databases, the Rochester Diabetic Neuropathy Study (RDNS, N = 456), and in healthy subjects (RDNS-HS, N = 330).

Results: In the RDNS, the most frequent abnormal attributes (≤2.5th/≥97.5th percentile) are: fibular motor nerve conduction velocity (MNCV; 26.3%); sural sensory nerve conduction velocity (SNAP; 25.4%); tibial MNCV (24.8%); ulnar MNCV (21.3%); fibular F latency (16.9%); and ulnar F latency (16.0%). Normal deviate (from percentiles) composite scores of NC included: representative of neurophysiological abnormalities; sensitive and specific for diagnosis and useful for epidemiological surveys; randomized trials; and medical practice. By contrast, abnormality of one or more attributes in any nerve or abnormally of two most sensitive attributes performed poorly.

Conclusions: Composite sum scores of normal deviates (from percentiles corrected for applicable variables) of sensitive NC attributes and with modifications, RDNS and AAN criteria performed acceptably for diagnosis of DSPN.