Considerations for stimulus-response curves in stroke: an investigation comparing collection and analysis methods

Stimulus-response curves (SRCs) provide information about the neurophysiological strength of corticospinal connections. The method for measuring and/or drawing meaningful conclusions of SRCs has not been well established in the stroke population. This study sought to characterize the use of an abbreviated SRC in stroke by comparing two data collection and two analysis approaches. SRCs were obtained using transcranial magnetic stimulation (TMS) in 25 survivors of stroke. The SRC data were collected with TMS intensities referenced to either motor threshold (MT; 5% increments between 100% and 140% of MT) or stimulator output (5% increments between 30% and 100% output) with peak-to-peak motor evoked potential amplitudes determined and averaged for each intensity level. Data were modeled with a three-parameter sigmoid function and a linear regression and were analyzed with a mixed design analysis of variance. There was no difference in the accuracy of prediction (r2) between the different analyses. There were significant differences in the slope parameters depending on the analysis method, but this was not dependent on the SRC data collection method. A linear regression of the SRC accurately represents the slope of the modeled data and is strongly correlated with the calculated peak slope, yet is less computationally complex than the sigmoid function. This study demonstrates the use of a linear analysis approach for SRCs as outcome measures for stroke studies.