SUPPLEMENTARY MATERIAL FOR:
THE EFFECT OF ANODAL HIGH DEFINITION TRANSCRANIAL DIRECT CURRENT STIMULATION ON THE PAIN SENSITIVITY IN A HEALTHY POPULATION: A DOU-
BLE-BLIND, SHAM-CONTROLLED STUDY
Sebastian Kold, Thomas Graven-Nielsen*
Center for Neuroplasticity and Pain (CNAP), Aalborg University, Denmark
Running title: Effects of HD-tDCS on somatosensory and pain sensitivity
Keywords: Quantitative sensory testing; High-density transcranial direct current stimulation Original paper for: PAIN
Research funding: Center for Neuroplasticity and Pain (CNAP) is supported by the Danish Na- tional Research Foundation (DNRF121).
*Corresponding Author:
Prof. Thomas Graven-Nielsen, PhD, DMSc.
Center for Neuroplasticity and Pain (CNAP) Department of Health Science and Technology Faculty of Medicine, Aalborg University
Fredrik Bajers Vej 7 D3, DK-9220 Aalborg, Denmark Phone: +45 9940 9832, Fax: +45 9815 4008
E-mail: [email protected]
SUPPLEMENTARY MATERIAL
Bayesian statistical analysis
Bayesian statistical inference was used to quantify the probability of the hypothesis, that the active stimulations modulated the somatosensory pain and detection thresholds more than sham stimulation [1]. Using the statistical software JASP (v. 0.14.0.0) Bayesian mixed model ANOVAs were con- ducted as an addition to the frequentist ANOVAs. In doing so the probability of the alternative hy- pothesis can be graded on a continuous scale in terms of a Bayes factor (BF10), which illustrates the predictive capability of the model compared to the null hypothesis model [2]. When the null hypoth- esis shows stronger predictive value than the alternative hypothesis Bayes factor in favor of the null hypothesis (BF01) is presented due to easier interpretation. As this is a novel study design, the default prior was chosen in all analysis [3]. Bayesian analysis were conducted on the MPT and TDT after tDCS on each day and at Day1pre (parameters with ANOVAs approaching significance, P < 0.1).
RESULTS
Bayesian analysis of mechanical detection and pain thresholds after tDCS on each day
A Bayesian mixed model ANOVA with factors Groups (Sham-tDCS, DLPFC-tDCS, M1-tDCS and DLPFC+M1-tDCS) and Time (Day1pre, Day1post, Day2post and Day3post) determined that the data were best represented by the null model for both mechanical modalities. MPT (BF01) = 5.31 and TDT (BF01) = 3.44 for the model that included the factors Time and Group and the Group*Time interaction compared to the null model, indicating decisive evidence in favor of the null model.
DISCUSSION
The Bayesian approach strengthened the null hypothesis, that there were no differences between the four groups in the response of the MPT and TDT, following the respective interventions. This argues in favor of the conclusion, that active HD-tDCS has no stronger modulatory effects than sham-stim- ulation in the healthy population of this study. This is a valuable addition to the findings of the fre- quentist statistics used in the main analysis, as these are not able to describe the reliability of the null finding.