This protocol is prepared in accordance with the Preferred Reporting Items for Systematic
Review and Meta-analysis Protocols (PRISMA-P) guidelines. The review was prospectively registered with the International Prospective Register of Systematic Reviews (PROSPERO; registration number: CRD42020173172).
Question 1. Is corticomotor excitability, and/or intracortical, transcortical and sensorimotor modulators of CME, altered in response to experimentalpain?
Question 2. Is there a relationship between CME during experimental pain and pain severity?
Searches will be conducted in the following databases: Scopus, PubMed, MEDLINE, PsycINFO and Web of Science. Key words relating to experimental pain and TMS will be used. These terms will be used as either free or controlled terms to identify literature of interest and will be adjusted for each database. The full search strategy for each database can be viewed in Appendix 1. The reference lists of eligible studies will be searched to identify additional studies.
Studies will be included based on the following criteria:
- English Language
- Published in peer reviewed journal articles.
- Healthy adult human participants (age >18 years, with no gender restrictions).
- Experimentally induced cutaneous (skin) or musculoskeletal (muscle or joint) pain delivered to any area of the body. Examples of cutaneous pain paradigms include capsaicin heat, electrical stimulation, laser stimulation, cold water immersion or thermal heat models. Examples of musculoskeletal pain paradigms include nerve growth factor, delayed onset muscle soreness, absorbic acid and hypertonic saline models.
- Outcomes include assessment of one or moreof: CME, intracortical, transcortical or sensorimotor circuits. These measures are described in detail in the primary outcomes section.
- Inclusion of a pre-pain baseline or pain-free control condition (in the same participants).
Studies will be excluded based on the following criteria:
- Experimental pain is induced in the presence of another intervention (e.g. during a motor or cognitive task) or active stimulation (e.g. repetitive TMS). Studies that use a shamstimulation condition during pain will be included as this is not expected to interact with corticomotor output.
- Grey literature including data from unpublished studies (e.g. graduate theses), published abstracts, book chapters and conference proceedings.
- Pain severity is not assessed, or it is not clear whether participants experienced pain.
- The timing of TMS outcomes relative to pain induction is not stated.
Eligible studies should reportone of the following measures:
1) Measures of CME including the amplitude or area of MEPs elicited when TMS is delivered to the “hotspot” (the scalp location that generates the largest MEPs) or map volume of the corticomotorrepresentationof amuscle[5, 6, 13].
2) Measures of intracortical circuitry within M1, including short-interval intracortical inhibition (SICI), intracortical facilitation (ICF) and long-interval intracortical inhibition (LICI). These are assessed using paired-pulse TMS, where a conditioning pulse to M1 is delivered prior to a test stimulus to M1. Depending on the interval between stimuli, this causesinhibition or facilitationof CME relative to the test stimulus alone. SICIoccurs at an interstimulus interval (ISI) of 1-6ms, ICF at an ISI of 10-30ms andLICI at an ISI of 50-200ms[18, 19].
3) Measures of transcortical circuitry, including short-interval interhemispheric inhibition (SIHI) or long-interval interhemispheric inhibition (LIHI). These are assessed using dual-coil TMS, where a conditioning stimulus to M1 is delivered prior to a test stimulus to the opposite M1. This causes inhibition ofCME relative to the test stimulus alone. SIHIoccurs at an ISI of 10ms andLIHI at an ISI of 40ms.
4) Measures of sensorimotor circuitry, including short afferent inhibition (SAI)and long afferent inhibition (LAI). These are assessed by delivering a conditioning electrical stimulus to a peripheral nerve prior to a test magnetic stimulus to M1. This causes inhibition of CME relative to the test stimulus alone. SAI occurs at an ISI of 20ms and LAI at an ISI of 200ms [21, 22].
References will be managed using Endnote X9. Duplicate studies will be removed. The title and abstract of each study identified through the search will be evaluated against the inclusion and exclusion criteria. If there is uncertainty about the eligibility of a study, the full text of that article will be retrieved for further information. The full texts of the remaining articles will then be screened to determine study eligibility. Articles that are excluded during full text screening and the reasons for their exclusion will be recorded. Study identification will be conducted by two independent reviewers. Any disagreements will be resolved by consensus, and if unresolved, a third reviewer be consulted.
Information such as sex, age, and main findings/outcomes (described in the data synthesis section) will be extracted. In addition, specific characteristics of the experimental pain paradigm will be extracted such as: pain model (transient, tonic or transitional) and painfultissue (cutaneous  or musculoskeletal). TMS data to be extracted will include: stimulation intensities, interpulse intervals, target and non-target muscles, whether measures are taken with the muscle at rest or undergoing active contraction,muscle contraction level (if under active contraction) and relative timeto pain onset/resolutionthat TMS measures were assessed. A custom extraction form (Appendix 2), piloted by two independent reviewers on two studies not directly related to this review, will be used to extract data from the included studies. Any disagreements will be resolved through a third reviewer.
Risk of Bias Assessment
To assess risk of bias and methodological quality of the included studies, we will use a custom 7-item checklist (Appendix 3). Some of these items are drawn from the Downs and Black checklist, while others were custom-made to capture the quality of each study’s pain methodology. We will also use another checklist to assess the quality of the TMS methodology (Appendix 4). After piloting these checklists, two independent reviewers will use them to assess risk of bias, with any disagreement resolved by a third reviewer.
Group Level Data. For each type of pain model (e.g. transient, tonic or transitional), data will be grouped as musculoskeletal or cutaneous pain, and into data obtained from target and non-target/control areas. Data will then be organised into windows to capture commonly assessed time frames during and after recovery from pain (Figure 1). For example, in tonic pain models, MEPs have been measured 0-10, 11-20 and 21-30 minutes after pain resolution.
To compute the effect size (Cohen’s dav), the following equation will be used:
Where Mdiffrefers to the mean difference between a pain/post-pain condition and a baseline/pre-pain condition, and SD1 and SD2 are the standard deviations for each of these conditions. dav will be adjusted to Hedge’sgav, using the following equation (n = sample size):
Lastly, the variance ingav (Vgav) will be computed using the following equation (n1 and n2 = sample size for the baseline and pain/post-pain condition):
For within subjects designs, the correlation between dependent measures is usually required for effect size and variance calculations. However, these values are usually not reported, in which case Equations 1-3,typically used for between subjects designs, can be used instead[30-32]. Where means and SDs of the outcome measures are not retrievable, an email will be sent to the corresponding author requesting these data. If there is no response, a decline to send data, or if authors are no longer contactable,means and SDs will be calculated from t-values, p-values and F-values, or estimated from illustrations.
Individual level data. For each study, the Pearson correlation coefficient (r)between CME change during pain (relative to baseline) and pain ratings will be determined. These data will be extracted from the study or obtained from raw data made available online (e.g., through an Open Science Framework). Where data are not openly available, an email will be sent to the corresponding author requesting this data.If there is no response, a decline to send data, or if authors are no longer contactable, then these studies will not be included in the individual level analysis. Individual level data will also be separated into pain model, painful tissue and target area.To estimate effect size from correlation coefficients, the value of r for each study is not used. Instead, r is converted to Fisher’s z scale[29, 33] using the following equation:
The variance in z (Vz) is computed using the following equation :
Meta-Analyses. The Meta-Analyses will be conducted using R and several packages (e.g., metafor, meta) using the effect size (gav or z) and effect size variance values (Vgavor Vz)obtained from Equations 1-5. Each effect size will be assigned a weight based on its variance, and the weighted k effect sizes will be summed and then divided by the weights of the k studies to calculate weighted mean effect size. The 95% confidence intervals of each weighted mean effect size will also be obtained. A z-transformation will be conducted to test the null hypothesis that the weighted mean effect size is 0,with a two-tailed significance set at .05. The Q statistic, to test whether there is significant heterogeneity amongst effect sizes, will be assessed against a k-1 Chi Squared distribution, with a Q< .05 indicating that the dispersion of effect sizes is substantial. The I2 statistic will also be computed to estimate the percentage of variability due to effect size heterogeneity. Substantial heterogeneity will be considered existent when I2 > 50%. A minimum of 2 studies will be required for a meta-analysis. If there is only one study for an analysis, a narrative description of the findings will be provided.
Sensitivity Analysis. For each meta-analysis, a sensitivity analysis will be conducted by removing studies with high-risk of bias and re-running the meta-analysis. Issues related to sensitivity analyses are usually identified during the systematic review process. For this reason, the exact cut-off value for what constitutes a high-risk study will be determined after piloting of the methodological checklists.