Foam rolling (FR) and roller massage (RM) studies have increased dramatically in the literature recently, in parallel with their increased popularity within the training population. An acute session of rolling can increase static hip flexor (Behara and Jacobson, 2017; Bradbury-Squires et al., 2015; Mohr et al., 2014; Monteiro et al., 2017), hip extensor (MacDonald et al., 2013; Markovic, 2015; Monteiro et al., 2017; Sullivan et al., 2013) and ankle (Halperin et al., 2014; Kelly and Beardsley, 2016; Skarabot et al., 2015) range of motion (ROM), as well as dynamic hip extensor ROM during a lunge (Bushell et al., 2015). Su et al. (2016) reported greater hip flexor ROM with foam rolling versus static stretching. Improved flexibility can persist for up to 20 minutes after rolling (Junker and Stoggl, 2015; Kelly and Beardsley, 2016; Mohr et al., 2014) with increases in ROM ranging from 2.8% (Skarabot et al., 2015) to 23.4% (Grieve et al., 2015). Despite the abundance of findings of increased ROM, there is not unanimity in the rolling literature. Thoracolumbar fascia mobility significantly increased with foam rolling, but there was no significant effect on lumbar flexion (Griefahn et al., 2017). Some studies have reported no significant change in ROM of the hip extensors (hamstrings) (Couture et al., 2015), hip flexors (quadriceps) (Murray et al., 2016) and knee flexors (hamstrings) (Vigotsky et al., 2015) following rolling. Thus, the literature is not entirely consistent regarding the effects of rolling on ROM.

Rolling is often referred to as a self-myofascial release technique (Barnes, 1997; Beardsley and Skarabot, 2015; Cheatham and Kolber, 2017; Cheatham et al., 2015; Grieve et al., 2015; Healey et al., 2014; MacDonald et al., 2013; Okamoto et al., 2014; Peacock et al., 2014; Skarabot et al., 2015; Vaughan, 2014); however, it is unlikely that the predominant mechanism for rolling-induced increases in ROM is a modification of the myofascia. According to Schleip (Schleip, 2003a; 2003b), supra-physiological forces are needed to alter the mechanical properties of the fascia. Similar to an acute bout of stretching, a distinct possibility is that stretch (pain) tolerance (Magnusson, 1998; Magnusson et al., 1996) may be a primary mechanism underlying rolling-induced increases in ROM. Global pain reduction responses have been demonstrated with increased pain pressure threshold (PPT) in the plantar flexors (Aboodarda et al., 2015; Cavanaugh et al., 2017), quadriceps (Cheatham and Kolber, 2017) and hamstrings (Jay et al., 2014) following RM or manual massage (Jay et al., 2014) of the contralateral limb. Furthermore, rolling-induced improved flexibility has occurred in non-rolled muscles such as improved hamstring and lumbar spine flexibility after rolling the plantar surface of the feet (Grieve et al., 2015), improved dorsiflexion ROM with rolling of the contralateral plantar flexors (Kelly and Beardsley, 2016) and a tendency for contralateral (p = .095) increases in medial gastrocnemius PPT (Casanova et al., 2017). However, not all studies have found this effect with a lack of increase in sit and reach flexibility scores after rolling the plantar surface of the feet (Grabow et al. 2017a). Thus, the non-local rolling effects provide strong evidence for a global increase in pain or stretch tolerance.

If a central pain-modulatory system plays a role in mediation of perceived pain and stretch tolerance following RM (Aboodarda et al., 2015; Cavanaugh et al., 2017), is it possible to augment the analgesic effect in order to further improve ROM? Transcutaneous electrical nerve stimulation (TENS) is a form of electroanalgesia, which diminishes painful sensations (Sluka and Walsh, 2003; Vance et al., 2014) by activating either large (conventional TENS) or small (intense TENS) diameter afferents to block peripheral nerves associated with pain (segmental and extra-segmental analgesia respectively) (Jones, 2009). Magnusson and colleagues (Magnusson and Renstrom, 2006; Magnusson et al., 1996) have emphasized the role of increased stretch tolerance for the enhancement of ROM. If increased pain (diminution of stretch discomfort) tolerance with TENS is possible, either during the rolling or persisting thereafter, can there be additive effects when integrating RM with TENS?

The primary objective of the study was to examine the effects of RM, TENS and the combination of RM and TENS on ROM and PPT. It was hypothesized that a TENS-induced increase in pain tolerance would augment the proposed stretch tolerance mechanisms underlying RM to provide an additive improvement in ROM and PPT.

There was no significant main effect of the intervention (F_{(3, 33)} = 2.099, p = 0.11) or significant interaction effect between the intervention and time (F_{(6, 66)} = 0.325, p = 0.92) on ROM in the dominant limb.

There was a significant main effect of time on ROM in the dominant limb (F_{(2, 22)} = 49.478, p < 0.001). Post-hoc test revealed that ROM was significantly better during pre-intervention measures compared to baseline measures (Δ = 2.8 ± 0.49, (CI: 1.44, 4.19)), during post-intervention measures compared to baseline measures (Δ = 6.0 ± 0.67, (CI: 4.17, 7.92)), and during post-intervention measures compares to pre-intervention measures (Δ = 3.2 ± 0.65, (CI: 1.39, 5.07)). All the aforementioned comparisons exhibited MCID (Table 2) and large magnitude effect size increases (baseline to pre-test: 0.89, baseline to post-test: 1.97, pre- to post-test: 0.95).

There was no significant main effect of the intervention (F_{(3, 33)} = 1.374, p = 0.268) or significant interaction effect between the intervention and time (F_{(6, 66)} = 0.705, p = 0.644) on ROM in the non-dominant limb.

There was a significant main effect of time on ROM in the non-dominant limb (F_{(2, 22)} = 36.496, p < 0.001). Post-hoc test revealed that ROM was significantly better during pre-intervention measures compared to baseline measures (Δ = 2.2 ± 0.64, (CI: 0.37, 3.97)), during post-intervention measures compared to baseline measures (Δ = 5.3 ± 0.66, (CI: 3.48, 7.19)), and during post-intervention measures compares to pre-intervention measures (Δ = 3.2 ± 0.59, (CI: 1.52, 4.82)). MCID were evident when comparing non-dominant baseline to post-test (CI: 3.48-7.19 vs. SEM: 0.87-1.01) and pre- to post-test (CI: 1.52-4.82 vs. SEM: 0.99-1.01) (Table 2). Furthermore, effect sizes were moderate to large respectively (baseline to pre-test: 0.65, baseline to post-test: 1.62, pre-test to post-test: 0.81).

There was a significant main effect of the intervention on VAS during RM (F_{(1, 11)} = 18.279, p = 0.001). Post-hoc test revealed that there was more perceived pain associated with the RM during the RM only intervention compare to both the TENS and RM intervention (Δ = 1.6 ± 0.38, (CI: 0.78, 2.44)).

There was a significant main effect of time on VAS during RM (F_{(1.726, 18.982)} = 16.183, p < 0.001). Post-hoc test revealed that there was more perceived pain during the second round of RM compared to the first round (Δ = 0.9 ± 0.21, (CI: 0.19, 1.50)), during the third round of RM compared to the first round (Δ = 1.5 ± 0.29, (CI: 0.59, 2.42)), and during the fourth round compared to the first round (Δ = 1.8 ± 0.34, (CI: 0.66, 2.85)). All the aforementioned comparisons were MCID.

There was a near significant interaction effect between the intervention and time (F_{(1.437, 15.811)} = 3.388, p = 0.072). Table 2 illustrates that VAS scores exhibited moderate (first bout) to large (bouts 2-4) effect size magnitude higher scores with RM versus BOTH for all rolling bouts. Furthermore, the increase in VAS scores with RM condition were all large magnitude effect sizes when comparing the first to subsequent bouts of rolling whereas with the BOTH condition, the effect size magnitudes were small to moderate (Table 2). Table 3 illustrates MCID and provides further evidence for the increases in VAS scores with RM with all succeeding rolling bouts (i.e. 1 vs. 2-4, 2 vs. 4 and 3 vs. 4). Furthermore, there was very little overlap when comparing RM rolling bouts 2 vs. 3 (CI: 0.11-1.12 compared to SEM 0.12-0.13). In contrast the BOTH condition only exhibited a MCID (increased VAS) with one comparison (rolling bout 1 vs. 3) (Table 4).

The most important finding of the study was that the simultaneous use of RM and TENS did not cause any additional effects to pain tolerance (PPT) or ROM in either the treated or contralateral quadriceps. In fact, there were no significant differences between any of the interventions, including control, on pain tolerance or ROM. Additionally, measurements of pain tolerance and ROM did show improvements with time across all four interventions for both quadriceps. Finally, in terms of the VAS measures, there was a decrease in perceived pain associated with RM when it was accompanied by TENS, and RM was reported to be more painful in the last three rounds of RM compared to the first round.

This is the first study to combine RM and TENS; therefore, there are no previous findings to which these results can be compared. It is somewhat difficult to interpret the lack of significant differences to pain perception or ROM between interventions. However, since even the control, which did not use RM or TENS, was not significantly different from the other interventions, it appears possible that the time-dependent changes in pain perception and ROM were at least partly due to the actual testing consisting of the MTT and PPT. Stretching is an effective mechanism to increase ROM, due to several mechanisms including changes to muscle viscoelasticity and increased stretch tolerance (Behm et al., 2015). While the MTT is not a lengthy test, it still involved stretching the muscle to the point of maximal tolerable stretch. This could activate several stretch-induced adaptations, which may result in an increased ROM during subsequent tests. In terms of pain perception, Aboodarda et al. (2015) and Cavanaugh et al. (2017) showed that over a short testing period, initial PPT trials caused superficial nociceptors to have an increased sensitivity, resulting in higher PPT values. However, it is possible that over a longer testing period (i.e. between a baseline and pre-intervention test), the previous set of PPT trials may in fact desensitize the nociceptors to the sensation of the algometer. In accordance with this rationale, the present study’s, baseline to pre-test PPT increased by a moderate magnitude (ES : 0.61), whereas pre- to post-test showed trivial magnitude changes. Hence, the both legs’ PPT main effect for time was primarily driven by the repeated testing effect.

Albeit, there was no statistically significant interaction dominant leg PPT differences (p > 0.05), the results did approach significance (p = 0.079) with perceptible increases in effect size magnitudes. Whereas the control condition changes when comparing pre- and post-test to baseline remained at a trivial magnitude with no evidence of MCID, TENS and RM increased from trivial to small magnitude changes at pre- to post-test respectively. The increased PPT with TENS achieved MCID and 96% of MCID when comparing baseline and pre-test PPT to post-test respectively. However, the RM condition did not reach MCID for any of the comparisons. The BOTH condition had small magnitude PPT increases at pre- and post-test when compared to baseline values but only achieved a MCID when comparing baseline to pre-test. The small magnitude effect size changes with the experimental conditions versus trivial control changes might be interpreted as suggesting a possible pattern of evidence for increased PPT or decreased pain sensitivity but generally, the small magnitude changes predominately occurred in relation to the baseline rather than the subsequent pre-test. While an increased PPT finding with RM would be in accordance with the literature (Aboodarda et al., 2015; Casanova et al., 2017; Cavanaugh et al., 2017; Cheatham and Kolber, 2017; Grieve et al., 2015; Jay et al., 2014; Kelly and Beardsley, 2016), in the present study, there was limited clinical evidence for the analgesic effects of TENS (Sluka and Walsh, 2003; Vance et al., 2014) or RM, and no additive effect with the combination of TENS and RM (BOTH). As previously mentioned, a repeated testing effect seemed to diminish MCID or small magnitude changes in PPT. Prior and future RM pain tolerance results should be viewed with caution if only a single pre-test is conducted.

Previous studies measuring pain tolerance following RM have shown significant increases in PPT (Aboodarda et al., 2015; Casanova et al., 2017; Jay et al., 2014; Vaughan, 2014). However, none of these studies targeted the quadriceps muscle. There have been no evaluations of PPT following RM to the quadriceps muscle. It is possible that there are properties of the quadriceps muscle that may limit the magnitude of change in pain tolerance (less sensitivity) measured through repetitive pressure algometry. Similar to ROM, the changes in PPT may be attributed to a testing effect associated with repeated use and measurement.

The RM device used in the study has been used in previous studies (Bradbury-Squires et al., 2015; Casanova et al., 2017; Grabow et al., 2017b; Sullivan et al., 2013). These studies have all found changes in ROM to the muscle of interest; however, none of the previous studies used the MTT to measure changes to quadriceps ROM. When assessing ROM of the quadriceps, other studies have used the inline lunge test to assess ROM, which has revealed more significant findings (Grabow et al., 2017b; Macdonald et al., 2013; 2014); although it can be more difficult to control the position of the pelvis using this test. One previous study using foam rollers found no significant change in ROM assessed by the MTT (Vigotsky et al., 2015). Therefore, it is possible that the changes in ROM are not easily detected using the MTT, and that the present changes in the MTT seen in all interventions were simply due to a testing effect. All conditions including control for both legs demonstrated MCID increases in ROM from baseline to pre-test and pre- to post-test with moderate to large magnitude effect sizes.

Finally, both the intervention and the particular round of RM, as determined by the VAS, affected pain perception. Since perceived pain was lower during the rounds of RM while TENS was in use, TENS did indeed produce an analgesic effect. Although not statistically significant there was a near significant (p = 0.072) increase in pain tolerance during the intervention with both TENS and RM compared to RM alone. The increased magnitude effect sizes of VAS scores with RM compared to the BOTH condition was classified as moderate (first bout) to large (bouts 2-4). Further evidence for increased VAS scores with each bout of rolling, with the RM condition, was shown with the 95% CI exceeding the SEM indicating a MCID when comparing all subsequent rolling bouts (i.e. 1 vs. 2, 3, 4 or 2 vs. 4 and 3 vs. 4). However, with the BOTH condition, there was only a MCID for an increased VAS score when comparing rolling bouts 1 vs. 3. This analgesic effect appeared to be transient, and only present while TENS was being administered. During the second, third, and fourth rounds of RM, pain perception was reported as higher compared to the first round of RM across all conditions. These findings indicate that as the duration of RM increased, so too did perceived pain. This increase in perceived pain could be related to an increased sensitivity of nociceptors, similar to the effects of the PPT algometer (Aboodarda et al., 2015; Cavanaugh et al., 2017).

There are several limitations to consider with the current study. Across all RM trials, the average VAS score associated with the rolling was 4.9/10 for RM and 3.3/10 for BOTH. These values are lower than what would be anticipated at a load equal to 70% of a load that elicited a VAS score of 10/10 and may explain the lack of significant findings. The current literature is limited on comparing RM intensities and the associated magnitudes of change of testing measures. The available research shows mixed results as to whether an intensity-dependent relationship is present (Grabow et al., 2017b; Young et al., 2018)

Part of the study was accurately determining the required intensities for TENS and RM while limiting the impact on the targeted muscle prior to the actual intervention. By exposing the participant to the devices for a brief time, followed by performing the tests, and then finally performing the intervention, it is possible that some adaptations occurred. These adaptations may render the muscle less sensitive to the following intervention, resulting in a smaller and insignificant magnitude of change.

There is limited research assessing PPT of the quadriceps following rolling interventions. There is no prior RM research that uses PPT to assess pain tolerance in the quadriceps; however, there are two previous studies that found an acute increase in PPT following foam rolling of the ipsilateral (Cheatham and Baker, 2017) and contralateral (Cheatham and Baker, 2017; Cheatham et al., 2017) quadriceps muscle. Given the exploratory nature of these studies, the reliability, validity, and specificity is not totally clear.

Finally, there was the presence of some non-normally distributed data. Since there is no non-parametric equivalent for a repeated-measures ANOVA, the tests were conducted despite the lack of normal distribution. Therefore, there is an increased potential for the presence of type I errors in the ANOVAs containing non-normal data. This includes the pain tolerance tests for both the dominant and non-dominant limb, which each had a significant main effect for time. However, the inclusion of ES and MCID provide additional clarity for the results and interpretations. On the other hand, distribution-based approaches to MCID have limitations. They allow calculation of the MCID, but not the clinically important differences. Furthermore, they only define the minimum value below which a change in pain score is not due to measurement error (Katz et al., 2015).

In conclusion, the addition of TENS to RM of the quadriceps did not significantly improve pain tolerance or ROM with the affected or contralateral leg. Future studies should continue to observe the interactions of RM and TENS, however the RM protocol, the targeted muscle group, and the chosen test measures should more closely follow those of previous studies which have shown pain tolerance and ROM improvements with RM. The finding that TENS decreases the relative amount of perceived pain during RM is an important consideration for future research and eventually clinical application. Future studies should determine if the use of TENS can increase the maximum tolerable RM intensity an individual can maintain and analyze the resultant changes to pain tolerance and ROM measures.

There was no significant main effect of the intervention (F_{(3, 33)} = 0.242, p = 0.867). There was a significant main effect of time on PPT in the dominant limb (F_{(2, 22)} = 8.004, p = 0.002). Bonferroni correction post-hoc test revealed that PPT was significantly higher during pre-intervention measures compared to baseline measures (Δ = 1.2 ± 0.34, (95% Confidence Intervals (CI): 0.29, 2.18)), and significantly higher during post-intervention measures compared to pre-intervention measures (Δ = 1.5 ± 0.50, (CI: 0.08, 2.88)). However, none of these comparisons proved to be MCID (CI overlapped with SEM). Whereas the pre- to post-test PPT change posted a trivial (0.08) magnitude effect size, baseline to pre-test and post-test achieved moderate magnitude changes (0.61 and 0.67 respectively).

A near significant interaction effect was detected between the intervention and time (F_{(2.918, 32.097)} = 2.496, p = 0.079) on PPT in the dominant limb. Table 1 illustrates that the TENS and RM conditions increased from trivial to small magnitude effect sizes when comparing PPT at baseline to pre- and post-test respectively. With TENS, there were MCID (CI > SEM) exhibited when comparing baseline to post-test (CI: 0.75-4.16 vs. SEM: 0.46-0.62). In addition, with the TENS pre- to post-test comparison (CI: 0.51-3.29 vs. SEM: 0.56-0.62), the CI only exceeded the SEM by 3.9% indicating that 96.1% of the scores were assumed not to be due to measurement error. The BOTH condition exhibited small magnitude effect size increases in PPT at pre- and post-test when compared to baseline, whereas the Control session had trivial magnitude changes. Furthermore, with the BOTH condition, MCID were only detected when comparing baseline to pre-test. There were no MCID with the RM or Control conditions.

There was no significant main effect of the intervention (F_{(3, 33)} = 0.556, p = 0.648) or interaction effect between the intervention and time (F_{(6, 66)} = 0.106, p = 0.995) on PPT in the non-dominant limb.

There was a significant main effect of time on PPT in the non-dominant limb (F_{(2, 22)} = 11.226, p < 0.001). Post-hoc test revealed that PPT was significantly higher during post-intervention measures compared to baseline measures (Δ = 1.8 ± 0.40, (CI: 0.65, 2.88)). This baseline vs. post-test result was also evident as a MCID (CI: 0.65-2.88 vs. SEM: 0.54-0.58).