Participants
Recruitment and study procedures were approved by the University of Florida Institutional Review Board. Written informed consent was obtained from all participants. The volunteers were required to have (1) no significant spontaneous pain anywhere in the body; (2) no ongoing pharmacotherapy with narcotics or antidepressants; (3) no disease that might significantly affect pain perception or unduly increase risk of injury (e.g., neurological disorders, serious psychiatric disorders, diabetes, hypertension, serious cardiovascular disorders, and chronic pain). Twelve of the sixteen participants were included in a previous study [1].
Pain measurement
Experimental pain was measured with an electronic version of a visual analog scale [11]. The electronic visual analog scale (eVAS) consisted of a low-friction sliding potentiometer of 100 mm travel. The left endpoint of the scale was identified as “no pain”, while the right endpoint was defined as “intolerable pain”. There were no divisions between these two anchors. The position of the slider was electronically converted into a pain rating between 0 and 100%. The slider automatically returned to the left (“no pain”) position when so required by the protocol. The eVAS was mounted into the surface of a small inclined desk positioned to facilitate precise operation with minimal fatigue. The custom-built testing system (Neuroanalytics Corporation, Gainesville, FL) integrated all inputs (temperature process value, eVAS signal) and outputs (stimulus temperature control, stimulus timing) and allowed automated execution of test protocols with preprogrammed parameters, including limits for temperature.
Response-Dependent Stimulation (REDSTIM) method
Thermal stimuli were administered with a flat copper contact thermode of 23x23mm in size. The thermode was electronically held at the desired temperature by a Peltier thermoelectric device. It was brought into light skin contact of reproducible force by solenoid activation, which was preprogrammed in the stimulator control software, allowing fully automated data collection. A pain intensity set-point was defined, and an algorithm in the stimulator control software calculated the deviation of the patient’s actual pain rating from the set-point as well as the derivative of this error. These data were the basis for automatic adjustments of the stimulus temperature to maintain an average pain rating that equaled the set-point.
In summary, REDSTIM is composed of the induction and the maintenance phase. During the induction phase, the thermode temperature was increased from non-painful levels (35°C) with temperature steps decreasing in size as pain ratings approach the set-point. During the induction phase, the temperature could increase or stay the same, but never decrease. Once pain intensity reached the set-point the maintenance phase began. When pain intensity exceeded the set-point the temperature was stepped down. Conversely, when the pain intensity was below the set-point, the stimulus temperature was stepped up. A temperature limit is set to prevent thermal injury of subjects that are pain insensitive. More detailed information on the system, software and paradigm has been previously reported elsewhere [1,2]. For the present study, continuous heat application was used to examine the psychometric properties of the oscillations occurring during the REDSTIM methodology.
Testing protocol
Thermal stimulation was conducted during three separate and identical daily sessions each separated by two-days apart. During each session REDSTIM experiments were conducted on the thenar eminence of the hand. Two set-points were tested on each subject in every session: 20/100 and 35/100 pain ratings. REDSTIM oscillations around the set-point were sampled every 1.5 seconds for a period of 120 seconds during all sessions and at both 20/100 and 35/100 set-point parameters. We used mild to moderate pain rating set-points to avoid reaching intolerable levels of pain in the most sensitive subjects. All test conditions were the same for the three experimental sessions and subjects were not informed of the pain rating set-points used.
Pain ratings above the set-point progressed in positive half-cycles that began within ascending temperature progressions, progressed to a high peak pain rating and then returned to the set-point. Similarly, pain ratings below the set-point in negative half-cycles began within descending temperature progressions progressed to a low peak pain rating and then returned to the set-point. For each half-cycle (positive and negative), deviations from the set-point were summed over time and the average area under the curve (AUC) was calculated for each subject for each day of testing. In addition, the software program automatically calculates the average temperature as the cycle progresses. The output provides a running average temperature needed to maintain the set-point at 60 seconds (i.e., for the first 60 seconds of the trial) and at 120 seconds of the cycle (i.e., average of the entire trial). Test-retest reliability was calculated for each of these variables.
Data analysis
Measures must have at least “Fair” test-retest reliability (ICC > 0.41) in order to be acceptable for its use, but ideally coefficients should fall within the “Moderate” (ICC > 0.61) to “Substantial” (ICC > 0.81) range [6,12]. Based on this premise, we estimated that a sample size of 16 subjects was required to measure an intra-class correlation coefficient (ICC) of at least 0.61 across three days, with a power of at least 0.80 at an alpha level of 0.05. We also assessed differences between days using a repeated measures ANOVA with “Days” and “eVAS set-point” (20 or 35) as within-subjects variables. First, data were examined for distribution, presence of extreme outliers and the Mauchly’s test of Sphericity was performed. If the sphericity assumption was violated, then Greenhouse-Geisser degrees of freedom corrections were applied. In addition, Bonferroni was used to adjust for multiple comparisons. Statistical analyses were conducted with IBM SPSS 22 for Windows. For all tests, 2-sided p values less than 0.05 were considered statistically significant.