A comparison of two methods for measuring thermal thresholds in diabetic neuropathy.

Thermal thresholds can be measured psychophysically using either the method of limits or a forced-choice method. We have compared the two methods in 367 diabetic patients, 128 with symptomatic neuropathy. The Sensortek method was chosen for the forced-choice device, the Somedic modification of the Marstock method for a method of limits. Cooling and heat pain thresholds were also measured using the Marstock method. Somedic thermal thresholds increase with age in normal subjects, but not to a clinically significant degree. In diabetics Marstock warm threshold increased by 0.8 degrees C/decade, Sensortek by 0.1 degrees C/decade. Both methods had a high coefficient of variation in normal subjects (Sensortek 29%, Marstock warm 14%, cool 42%). The prevalence of abnormal thresholds was similar for both methods (28-32%), though Marstock heat pain thresholds were less frequently abnormal (18%). Only 15-18% of patients had abnormal results in both tests. Sensortek thresholds were significantly lower on repeat testing, and all thresholds were higher in symptomatic patients. Both methods are suitable for clinical thermal testing, though the method of limits is quicker. In screening studies the choice of a suitable apparatus need not be determined by the psychophysical basis of the test.