The purpose of this study was to examine the effect of localised tibialis posterior muscle fatigue on shank, rearfoot, and forefoot joint coupling and coupling variability during walking. Two main hypotheses were put forth: following a bout of fatigue-inducing exercise participants would demonstrate 1) altered and non-synchronous joint coupling between the respective segments as well as 2) reduced coupling variability. To test these hypotheses, a unique approach was utilised to selectively fatigue the tibialis posterior muscle.
The protocol for reducing the force output of this muscle was developed based on Kulig et al.  who showed that isolated activation of tibialis posterior is best achieved using closed chain resisted foot adduction. The results of the present study indicate that this fatigue protocol was successful in reducing the MVIC force by over 30%. Although eight participants did not achieve the targeted 30% reduction in force production, they did all achieve at least a 21% reduction and were unable to complete 2 consecutive sets of the 50-repetition exercise. Furthermore, there was no evidence that these eight participants differed systematically from the rest of the sample in terms of kinematic changes following fatigue based on the results of the current study and previous study . Finally, the reduction in strength was still apparent following the POST data collection indicating that the fatigue protocol was effective.
The decrements in isometric force are similar to previous fatigue studies and studies involving healthy runners and PTTD patients. Cheung and Ng  reported similar findings for fatigue of the invertor muscles in healthy runners following an exhaustive run. Moreover, Alvarez et al.  reported a 40% reduction in concentric ankle invertor strength for PTTD patients prior to a 16-week rehabilitation program. However, it should be recognized that the PTTD patients in this study included advance-stage PTTD patients who had symptoms for approximately 16.5 weeks prior to treatment. Pilot data from our laboratory shows that early-stage PTTD patients exhibit a 17% reduction in ankle invertor MVIC strength compared to healthy controls. Thus, we are confident that our fatigue protocol and a priori criteria for localised muscle fatigue is sufficient to induce tibialis posterior muscle fatigue and concomitant reductions in force output during a dynamic task such as walking.
In support of the first hypothesis, a change in joint coupling angle between 2.3° and 5.3° was measured during Phase 2 and 3. Moreover, the tibia and forefoot all increased their respective motions relative to the rearfoot. While we are not aware of another study that has investigated changes in foot and ankle joint coupling following a fatigue protocol, the pre- and post-fatigue coupling angle data in the current study are similar to Pohl et al.  who also reported joint coupling angles at or near 45° for the same coupling relationships while walking. Specifically, the pre-fatigue values for the TIBrot:RFi/ev coupling angle indicate a near 1:1 ratio in coupling for Phase 1 then greater overall motion of the rearfoot throughout the remainder of stance which is similar to previous studies [25, 26].
Post-fatigue, and during Phase 2 of stance, increased tibial motion was measured, relative to the rearfoot, suggesting that fatigue of the tibialis posterior disrupted the typical coupling mechanics between these two segments. This same relationship was observed for the RFi/ev:FFd/pf coupling relationship wherein a 1:1 coupling relationship is measured pre-fatigue and post-fatigue shows a change in forefoot motion relative to the rearfoot for Phases 2 and 3. Finally, greater overall rearfoot motion, relative of the forefoot (RFi/ev:FFab/d), was measured pre-fatigue, which is consistent with previous studies , and fatigue of the tibialis posterior disrupted this relationship resulting in greater motion of the forefoot relative to the rearfoot for Phases 2 and 3 of stance. Thus, it can be concluded that when the tibialis posterior is unable to produce sufficient force, there are significant alterations in coupling patterns for the shank and foot.
We postulate that the overall greater motion of the tibia and forefoot (relative to the rearfoot) following fatigue is the result of the functional anatomy of the tibialis posterior muscle itself. The tibialis posterior muscle originates from the tibia and the tendon does not attach directly to the rearfoot (calcaneus), but has several attachment points to the midfoot and forefoot. Thus, it is reasonable to speculate that greater relative motion of these segments is the result of the inability of the muscle, via fatigue and reduced force output, to control the individual motions of these foot segments.
In contrast to the second hypothesis, an increase in joint coupling variability was measured following the fatigue protocol for TIBrot:RFi/ev and RFi/ev:FFab/d during Phase 2 and 3. These results are in contrast to several other studies investigating joint coupling variability. Ferber et al.  studied different types of orthotics during running and reported no significant changes in TIBrot:RFi/ev variability across orthotic conditions or compared to a control group. Hamill et al.  studied patients with patellofemoral pain syndrome (PFPS) and reported overall reduced joint coupling variability for thigh and shank coupling variability compared to the uninjured leg and a control group. However, these authors measured thigh and shank coupling patterns or variability so it is difficult to compare their results with those of the present study. Also in contrast to the results of the present study, Miller et al.  reported that runners with a history of iliotibial band syndrome (ITBS) demonstrated reduced TIBrot:RFi/ev coupling variability while running on a treadmill compared to a control group. However, it is important to note that these studies involved runners who were either injured at the time of testing or had a long history of running-related injuries. The participants in the current study were healthy athletes with no history of chronic injury and involved an intrinsic perturbation rather than a cross-sectional comparison. In addition, these authors [24, 36] used a different measure for coordination (continuous relative phase), which may not directly compare to the present vector coding method and could explain the different findings of the present study. Thus, comparisons to previous studies must be made with caution.
While we are not aware of another study utilising a muscle-fatigue protocol to measure changes in either joint coupling or coupling variability, two studies have investigated changes in movement variability following an intervention of some type. Ferber et al.  reported that following a 3-week strengthening protocol, a reduced variability in stride-to-stride knee joint kinematic patterns was adopted by the PFPS group. These authors suggested that, from a clinical perspective, restoration of a more consistent and predictable movement pattern is expected with the increases in muscle strength and reductions in pain. Thus, the increase in coupling variability following fatigue in the present study is consistent with these authors . Further research involving changes in coupling variability following successful rehabilitation from a musculoskeletal injury is, however, warranted.
Few studies have investigated the effect of fatigue on changes in joint coupling. Miller et al.  studied changes in joint coupling variability during an exhaustive run for runners who had previously experienced ITBS. Compared with the control group, the ITBS runners were more variable in knee flex/extension-foot add/abduction at the start of the run, less variable in thigh add/abduction-foot inv/eversion at the end of the run, tended to be less variable in thigh add/abduction-tibia rotation at the end of the run but showed no change in TIBrot:RFi/ev coupling variability either during the entire stride cycle, swing, or stance phase. It is possible that since a variety of changes in coupling variability were reported, albeit the majority of joint coupling relationships showing reduced variability, that increased coupling variability for the shank and foot is a possible mechanism to explain injury aetiology similar to the results of the present study.
Based on the redundancy of the various muscles that serve to control frontal plane rearfoot and transverse plane tibial motion, a potential strategy for the foot may be to increase coupling variability to avoid injury. We postulate that a diminished ability of the muscle to produce a vigorous contraction, a concomitant reduction in joint contact force, and a resulting increase in joint coupling variability could result. In other words, the reduced function of the tibialis posterior muscle following fatigue would result in less control of joint movement since fewer muscles are functioning to achieve a desired movement pattern. Moreover, since tibialis posterior is a major invertor of the foot, and we successfully fatigued this muscle, other muscles must compensate to control foot pronation. Given these muscles are not as accustomed to localised fatigue conditions, this might also contribute to the increased variability that was observed. Finally, it is possible that reduced posterior tibialis function lead to increased activation levels of other inverters with the goal of compensating for the loss of force. Future studies are needed to improve our understanding of the lower extremity as a dynamical system in healthy and injured runners and how kinematic coupling and variability patterns may change for patients with chronic and more advanced PTTD.
There are factors that may have influenced the results of this study. While closed-chain foot adduction has been shown to be the best exercise at selectively activating tibialis posterior , as previously discussed, other muscles also play a role in this movement. Therefore, this study was limited in its ability to specifically quantify the degree of fatigue that was achieved in the tibialis posterior muscle. An alternative approach would be to quantify changes in muscle activity and fatigue via the use of electromyography . Subsequent EMG studies would also enable greater understanding of the compensation strategies employed by other muscles. Second, the order of conditions was the same for all subjects and ideally the order would be balanced to minimize the changes of a presentation bias. However, in a fatigue study, it is admittedly difficult to achieve randomization of order unless EMG, MRI, or some other valid measurement technique was used to ensure that participants recovered from the fatigue protocol prior to post-fatigue testing. Third, we chose to investigate potential changes in joint coupling and coupling variability using a vector coding technique. However, other techniques, such as continuous relative phase (CRP) are also available. Perhaps using a method such as CRP would yield different results but Miller et al.  stated that both vector coding and CRP methods seem to be valid metrics for assessing variability. However, future research using different methods of assessing joint coupling is warranted. Finally, our analysis was restricted to the stance phase of gait and did not include the swing phase. Previous studies have reported differences in coordination variability during swing or during the transitions between swing to stance [24, 28, 33]. However, since the ankle invertor muscles exhibit minimal or no activity during the swing phase of gait , we chose not to analyze these data.