Components of variance for vibratory and thermal threshold testing in normal and diabetic subjects

Quantitative sensory testing (QST) is commonly used in the assessment of diabetic neuropathy. However, little data are available on the reliability of tactile and thermal testing devices. Reproducibility of QST measures between centers has not been previously reported. This study was designed to validate QST testing procedures and determine if these devices are suitable for large scale multicenter clinical trials. Finger and toe vibratory (V_f, V_t) and thermal (T_f, T_t) thresholds were determined for ten normal individuals by a two-alternative forced-choice procedure using the Optacon Tactile Tester (OTT) and Thermal Sensitivity Tester (TST). Threshold measurements were reproducible between technologists and had a day-to-day coefficient of variation of V_f 20%, V_t 23%, T_f 41%, and T_t 95%. Thresholds were determined for 140 normal individuals at six centers. Mean threshold values between centers were not significantly different. Center-to-center coefficents of variation (CV) were V_f 44%, V_t 45%, T_f 47%, and T_t 87%. There was no significant difference in threshold measures with regard to sex, side studied, presence of calluses, or skin temperature. V_f thresholds significantly correlated with age (p < 0.01). There was no correlation between either vibratory or thermal thresholds in normal individuals, and nerve conduction velocities (NCV). Thermal and vibratory thresholds were determined for 98 diabetic patients. Diabetic subjects without clinical evidence of neuropathy were not significantly different from normal individuals, but diabetic patients with neuropathy had increased thresholds compared to normals (p < 0.05). We conclude that thermal and vibratory threshold testing with the OTT and TST yields reproducible measurements in normal individuals and provides an objective measure of clinical diabetic neuropathy that is well suited for multicenter trials.