Duty Factor Reflects Lower Limb Kinematics of Running
Round 1
Reviewer 1 Report
The authors investigated the kinematic profiles of two groups of runners. These groups are those who had higher duty factor and those who had lower duty factor within their sample. The authors hypothesized that the supposedly distinct duty factor groups would demonstrate unique kinematics of the lower extremity, particularly the sagittal plane angles, during stance. This reviewer has concerns that there is a massive amount of data presented but very narrow hypotheses in the text. It appears that the authors wanted to present a narrow hypothesis but analyzed a broad set of data in an exploratory fashion. In addition, the novelty of the study seems lacking and these issues limit the value of the paper. There is no clear message that builds on what is already known regarding fore/rearfoot strikers and their temporal/kinematic patterns.
General Comments
Some repeated issues with repeating of the letter ‘n’ in words such as ‘additionnally’ in the first line of the paper, for example. This occurs throughout and needs attention.
The introduction does not set up the methods appropriately. Justification for all of the variables and the separation of stance into multiple phases needs addressed. An alternative approach could be statistical parametric mapping. For example, see:
Nuesch, C., Roos, E., Egloff, C. Pagenstert, G., & Mundermann, A. (2019). The effect of different running shoes on treadmill running mechanics and muscle activity assessed using statistical parametric mapping (SPM). Gait & Posture, 69, 1-7.
This reviewer recommends the authors eliminate unnecessary variables and segments of the stance phase and focus more concretely at the most important components of their research question. Breaking the stance phase into so many segments and presenting a large amount of variables, across multiple speeds, makes it difficult for the authors to truly connect a simple duty factor delineator as explanation for supposed differences.
Why use a male only reference data set for segment parameters when nearly one third of the sample was female?
There are numerous interactions between group and speed factors that are basically ignored in the discussion. Interactions must be addressed before main effects.
The tables are gargantuan. As mentioned elsewhere, the authors should be more judicious with when/what variables are used in the analysis. Tables should be able to fit within a single page.
Tables do not show the pairwise comparison results for the speed effect. Therefore, the table results do not stand alone without the text.
Effect sizes would better illustrate differences observed. With the numerous variables and various timepoints from which data were taken, the magnitudes of differences would be a helpful classifier to inform the reader where differences observed were more meaningful. This could support the authors’ conclusions regarding the unique kinematic profiles of these two supposed running groups.
The discussion does not address all data presented and this is related to the above comment that the authors need to be more judicious with variable selection.
Specific comments
The second hypothesis is just a restatement of the first and is unnecessary.
Line 99. How many strides were selected for analysis during the 15 second window? Which limbs were analyzed? Were the individual trials analyzed separately then combined for analysis or was the average profile within/between the limbs used?
Line 156. ‘Breaking’ is the render something in disrepair. ‘Braking’ means to slow. Also line 170.
Lines 456-458. The authors assume angles of the pelvis and trunk when they recorded full body kinematics. Why not verify the relative angles of the trunk and thigh segment or at least consider the absolute angle of the thigh instead of making assumptions?
Line 515. The authors indicate ‘his’ yet there were females in the sample. Were there sex specific outcomes that the authors did not provide here?
Author Response
General Response: We would like to thank the reviewer for the useful comments. The feedback provided was helpful in improving the quality of our manuscript during the revision process. We have adapted our manuscript as suggested by the comments of the reviewer. Indeed, we have provided an answer to each comment. In addition, to facilitate the review process, we have indicated all modifications in the manuscript in RED font color, except for where we removed text. We trust that the reviewer finds our responses meet the expectations.
Response to All Comments:
The authors investigated the kinematic profiles of two groups of runners. These groups are those who had higher duty factor and those who had lower duty factor within their sample. The authors hypothesized that the supposedly distinct duty factor groups would demonstrate unique kinematics of the lower extremity, particularly the sagittal plane angles, during stance. This reviewer has concerns that there is a massive amount of data presented but very narrow hypotheses in the text. It appears that the authors wanted to present a narrow hypothesis but analyzed a broad set of data in an exploratory fashion. In addition, the novelty of the study seems lacking and these issues limit the value of the paper. There is no clear message that builds on what is already known regarding fore/rearfoot strikers and their temporal/kinematic patterns.
Response: We thank the reviewer for this comment. First, we would like to mention to the reviewer that classifying runners based on their DF allows taking into account directly tf in the classification process in addition to tc. Indeed, a classification solely based on the foot strike pattern (forefoot vs. rearfoot) does take into account tc but only implicitly (tc being correlated to the footstrike pattern). Therefore, DF is a more global way to classify runners. Moreover, it was shown that DF and Vscore (a subjective evaluation of the running form) reflect similar constructs and lead to similar subgroupings of spontaneous running form. Knowing so, DF also encompasses, to some extent, vertical oscillation of the head, antero-posterior motion of the elbows, vertical pelvis position at ground contact, antero-posterior foot position at ground contact, and footstrike pattern when classifying runners. Notably, footstrike pattern is only one parameter among many others when classifying runners based on their DF.
Nonetheless, we agree with the reviewer that in average, a DFlow runner will tend to be a midfoot or forefoot striker while a DFhigh runner will tend to be a rearfoot striker.
Patoz, A.; Gindre, C.; Thouvenot, A.; Mourot, L.; Hébert-Losier, K.; Lussiana, T. Duty factor is a viable measure to classify spontaneous running forms. Sports 2019, 7, 233, doi:10.3390/sports7110233.
We have clarified this point in the introduction.
“Moreover, grouping runners based on their DF reflect their global running pattern which takes into account not only the FS pattern but also vertical oscillation of the head, antero-posterior motion of the elbows, vertical pelvis position at ground contact, and antero-posterior foot position at ground contact.”
A previous study depicted temporal and COM differences of the running pattern between runners classified based on their DF.
Lussiana, T.; Patoz, A.; Gindre, C.; Mourot, L.; Hébert-Losier, K. The implications of time on the ground on running economy: less is not always better. J. Exp. Biol. 2019, 222, doi:10.1242/jeb.192047.
However, the lower limb kinematics, particularly the sagittal plane joint angles as well as 3D pelvis and foot segment angles during stance were not studied, which is the novelty of this study. This point has been clarified in the aims of the manuscript. Nonetheless, we agree with the reviewer that lower limb kinematics was studied between forefoot and rearfoot strikers. However, as previously mentioned, classifying runners using their DF is different than using their footstrike pattern.
“Hence, the classification of runners based on their (DF) showed the existence of two different self-optimized global running patterns [17]. However, nothing is known about their lower limb kinematics, particularly their sagittal plane (hip, knee, and ankle) joint angles and 3D pelvis and foot segment angles during stance.”
We agree with the reviewer that there is a massive amount of data in this study. However, the rationale behind such large analysis is that we believe that the running gait follows a global strategy which needs to be addressed by inspecting a lot of parameters of the running gait at the same time. More explicitly, we believe that hip, knee, and ankle joints are working together during the running cycle, which means that a lot of data needs to be analyzed. However, as we believe these joints are working together, it should follow that only two options are left: 1) they produce flexion of the lower limb during stance, or 2) they produce extension of the lower limb during stance, leading to a very narrow hypothesis.
General Comments
Some repeated issues with repeating of the letter ‘n’ in words such as ‘additionnally’ in the first line of the paper, for example. This occurs throughout and needs attention.
Response (continued): We thank the reviewer for this comment. We have corrected the mistakes and checked the manuscript for misspelled words or grammatical errors.
The introduction does not set up the methods appropriately. Justification for all of the variables and the separation of stance into multiple phases needs addressed. An alternative approach could be statistical parametric mapping. For example, see:
Nuesch, C., Roos, E., Egloff, C. Pagenstert, G., & Mundermann, A. (2019). The effect of different running shoes on treadmill running mechanics and muscle activity assessed using statistical parametric mapping (SPM). Gait & Posture, 69, 1-7.
Response (continued): We thank the reviewer for this comment. We agree with the reviewer that statistical parametric mapping seems to be a very interesting approach to analyze statistical differences on time series data. Nonetheless, we have decided to keep our original analysis. Indeed, it is applied to very standard and well-known running events which makes it comparable with actual literature. Nevertheless, a future study involving statistical parametric mapping is worth being conducted.
Moreover, we agree with the reviewer that the introduction was lacking the justification of the use of several variables and separation of running cycle into sub-timings. Therefore, we have modified the introduction accordingly.
“In their analysis, the authors divided the running cycle in several sub-timings, namely, braking (tb) and pushing (tp) times, which are defined from footstrike (FS) to mid-stance (MS) and from MS to toe-off (TO), respectively, and elevation (te) and dropping (td) times, defined as the time from TO to mid-flight (MF) and from MF to FS of the contralateral foot, respectively. MS and MF events were calculated to divide tc and flight time (tf), respectively. The authors observed that the two subcomponents of the contact phase (tb and tp) were statistically longer for DFhigh than for DFlow and that both te and td were statistically shorter for DFhigh than for DFlow. In addition, the authors showed that the vertical displacement of the center of mass (COM) during the elevation phase (ΔzCOM,e) was statistically greater for DFlow than for DFhigh, but no statistical differences were obtained for the vertical displacement of the COM during braking (ΔzCOM,b), pushing (ΔzCOM,p), and dropping (ΔzCOM,d) phases.”
“This passive impact peak usually occurs at about 25 ms after initial ground contact and this event marks the middle of the passive peak impact attenuation (IA) phase [24–32].”
“These impact attenuation strategies between forefoot and rearfoot strike patterns induce different loadings on the lower limb and different three-dimensional (3D) stress patterns in the ankle, knee, and hip joint [40–42], as well as different sagittal plane (hip, knee, and ankle) angles and 3D pelvis and foot segment angles during stance [42,43], but with no global advantage of one strike pattern over the other [40].”
This reviewer recommends the authors eliminate unnecessary variables and segments of the stance phase and focus more concretely at the most important components of their research question. Breaking the stance phase into so many segments and presenting a large amount of variables, across multiple speeds, makes it difficult for the authors to truly connect a simple duty factor delineator as explanation for supposed differences.
Response (continued): We thank the reviewer for this comment. However, we believe that this is what is nowadays missing in the literature. Indeed, we believe that the running pattern is a global motion (for which we can identify 2 groups of runners based on their DF) and that this global motion is reflected by the center of mass position, sub-timings of the running cycle, and lower limb kinematics, i.e. ankle, knee, and hip joint angles as well as 3D pelvis and foot segment angles. However, as written in a previous comment, we believe that all these parameters are being articulated in the same way depending on the group, i.e., more flexion of the lower limb for DFhigh and more extension of the lower limb for DFlow.
Why use a male only reference data set for segment parameters when nearly one third of the sample was female?
Response (continued): We thank the reviewer for this comment. We did not do any sex distinction within our DF groups. However, we agree with the reviewer that a future work could study the impact of sex on global running pattern. For instance, one could wonder if, within a given DF group, females depict similar lower limb flexion than males. We have added this comment in the limitations.
“Moreover, no sex distinction was taken into account within our DF groups. Thus, future work should focus on the impact of sex on global running pattern. For instance, one could wonder if, within a DF group, females depict similar lower limb flexion than males.”
There are numerous interactions between group and speed factors that are basically ignored in the discussion. Interactions must be addressed before main effects.
Response (continued): We thank the reviewer for this comment. According to the reviewer suggestion, we have discussed interactions in the discussion.
“In addition, with increasing running speed, DFhigh were more prone to decrease tb and tp and to increase te and td than DFlow (Table 2). Increasing the running speed leads to the decrease of tc and increase of tf. Therefore, as DFhigh runners start from larger tc and smaller tf at 8 km/h than DFlow and as tc and tf cannot indefinitely decrease and increase, respectively, the observed results naturally follow.”
“In addition, DFlow slightly increased the foot supination at MS and TO (Table 10) and decreased the toe-in at IA, MS, and TO (Table 11) with increasing running speed while no impact of speed was observed for DFhigh. These findings suggest that DFhigh runners were able to maintain their propulsion mechanism made with the toes during tp to propel their body forward even if the increase of speed is leading to a more bouncy running pattern.”
“In addition, with increasing running speed, DFhigh depicted a larger increase of knee flexion at IA than DFlow as well as a larger increase in the range of motion of the knee joint during ta and tp (Table 4). These findings are potentially representing a more pronounced eccentric action of the quadriceps muscle for DFhigh than DFlow during tb. and to a better reliance of forward propulsion for DFhigh than DFlow during tp.”
“Furthermore, a larger range of hip extension as well as a greater increase in the range of motion of hip extension with increasing running speed during tp was observed for DFhigh than for DFlow (Table 4). These findings further suggest that DFhigh promote forward propulsion.”
“In addition, the pelvis showed a higher range of retroversion during ta and tb and no change of retroversion with increasing running speed (Table 6) despite a higher anteversion at FS for DFhigh than DFlow.”
The tables are gargantuan. As mentioned elsewhere, the authors should be more judicious with when/what variables are used in the analysis. Tables should be able to fit within a single page.
Response (continued): We thank the reviewer for this comment. We have split all tables which did not fit on a single page so that they now all fit on a single page. Please see the new tables: Tables 3 to 11.
Tables do not show the pairwise comparison results for the speed effect. Therefore, the table results do not stand alone without the text.
Response (continued): We thank the reviewer for this comment. However, we would like to point to the reviewer that we have decided to not show pairwise comparison for speed effect (this is not the main message of the manuscript) but only the pairwise comparison between DF groups at each speed. Indeed, the statistically significant pairwise comparisons between DF groups at the different speeds are depicted using an asterix (*) in the tables when there is statistical significance.
“* Significant difference between DF groups at a given speed as determined by Holm post-hoc tests”
or
“No significant difference between DF groups at a given speed was determined by Holm post-hoc tests.”
Effect sizes would better illustrate differences observed. With the numerous variables and various timepoints from which data were taken, the magnitudes of differences would be a helpful classifier to inform the reader where differences observed were more meaningful. This could support the authors’ conclusions regarding the unique kinematic profiles of these two supposed running groups.
Response (continued): We thank the reviewer for this comment. As suggested by the reviewer, we have added the effect size in the results section for all statistically significant DF groups main effect in the text.
“Cohen's d effect size was calculated when a significant DF group effect was observed [58], and classified as small, moderate, and large when d values were larger than 0.2, 0.5, and 0.8, respectively [58].”
and (only one occurrence is provided here)
“The two subcomponents of the contact phase (tb and tp) were longer for DFhigh than for DFlow (DF groups main effect, p < 0.001; tb: moderate effect size, d = 0.62 and tp: large effect size, d = 0.82; Table 2 and Fig. 1)”
The discussion does not address all data presented and this is related to the above comment that the authors need to be more judicious with variable selection.
Response (continued): We thank the reviewer for this comment. We have provided references to the tables presented in the results section in the discussion to better link between our say and our results. Moreover, we have work on the discussion so that it takes into account more results than it actually did. In addition, we refer the reviewer to the previous comment about statistically significant interaction effect.
“Accordingly, a significantly more pronounced forefoot angle was observed at IA and MS for DFlow than DFhigh, suggesting than DFhigh are relying more on the rearfoot during ta and tb than DFlow.”
“In addition, DFhigh depicted a significantly larger forefoot angle at TO than DFlow, which was associated with a significantly larger |∆θp| (Table 9). These findings further suggest that DFhigh promote forward propulsion by pushing more with the toes during tp.”
“Our study additionally observed that DFlow striked the ground with more supination than DFhigh (Table 10), as similarly observed by Williams and Cavanagh (1987), and Williams et al. (2000) for forefoot strikers and that more toe-in was present for DFhigh than DFlow at IA (Table 11).”
“When focusing on the entire lower limb, less knee flexion was observed for DFlow than DFhigh at MS but a similar one at FS, which led to a larger range of motion of the knee joint angle during tb and tp for DFhigh than DFlow (Table 4). Therefore, a larger leg compression and decompression was observed for DFhigh than DFlow despite the similar knee joint angle at FS. In addition, with increasing running speed, DFhigh depicted a larger increase of knee flexion at IA than DFlow as well as a larger increase in the range of motion of the knee joint during ta and tp (Table 4). These findings are potentially representing a more pronounced eccentric action of the quadriceps muscle for DFhigh than DFlow during tb. and to a better reliance of forward propulsion for DFhigh than DFlow during tp. The first part of these results are in line with the observations of Gerritsen et al. (1995), which showed that the eccentric action of the knee extensor muscles was playing a role to attenuate the impact when rearfoot striking.”
“Furthermore, a larger range of hip extension as well as a greater increase in the range of motion of hip extension with increasing running speed during tp was observed for DFhigh than for DFlow (Table 4). These findings further suggest that DFhigh promote forward propulsion.”
“Moreover, a larger pelvis drop during tb together with larger pelvis rotation during ta and tb (visually observable but not statistically significant due to high standard deviations; Fig. 3A) as well as a statistically more externally rotated pelvis at FS (Table 7) were observed for DFhigh than DFlow, suggesting that DFhigh have more pelvis mobility and thus are able to rotate during tb not only in the sagittal but also in the frontal and transversal planes.”
Specific comments
The second hypothesis is just a restatement of the first and is unnecessary.
Response (continued): We thank the reviewer for this comment. We have rephrased the hypothesis.
"As DFhigh and DFlow favor long tc and long tf (therefore shorter relative tc), respectively, we hypothesized that these two DF groups would rely on two different lower limb kinematic mechanisms. More specifically, we hypothesized that DFhigh would use more lower limb flexion during tc than DFlow."
Line 99. How many strides were selected for analysis during the 15 second window? Which limbs were analyzed? Were the individual trials analyzed separately then combined for analysis or was the average profile within/between the limbs used?
Response (continued): We thank the reviewer for this comment. We have added that 22 ± 2 running strides were analyzed during the 15 seconds.
"3D kinematic data were collected during the static trial and the last 15 s (22 ± 2 running strides) of the running trials."
We have performed a whole-body biomechanical analysis, meaning that both left and right side have been recorded and analyzed. However, our results report the average betwen left and right side. This has been clarified in the manuscript.
"Temporal characteristics and COM positions and displacements were given as the average between right and left values."
We would like to mention the reviewer that a similar statement was already present for joint and segment angles: " Joint and segment angles were given as the average between right and left joint and segment angles, respectively.".
Line 156. ‘Breaking’ is the render something in disrepair. ‘Braking’ means to slow. Also line 170.
Response (continued): We thank the reviewer for this comment. We have changed all occurrences of "breaking" to "braking".
Lines 456-458. The authors assume angles of the pelvis and trunk when they recorded full body kinematics. Why not verify the relative angles of the trunk and thigh segment or at least consider the absolute angle of the thigh instead of making assumptions?
Response (continued): We thank the reviewer for this comment. As correctly pointed out by the reviewer, we should have measured an extra parameter (e.g., the trunk angle) to avoid making assumptions. However, as we did not want to add an extra parameter, we have decided to remove the lines corresponding to this assumption.
Line 515. The authors indicate ‘his’ yet there were females in the sample. Were there sex specific outcomes that the authors did not provide here?
Response (continued): We thank the reviewer for this comment. We have replaced "his" by "the" in the manuscript to avoid any sex confusion as females represented one third of our sample. Moreover, we would like to mention to the reviewer that we did not differentiate male and female when classifying runners based on their DF. Threfore, there is no sex specific outcome.
"To conclude, this study observed that two lower limb kinematic mechanisms are involved in running and the one of an individual is reflected by the DF."
Reviewer 2 Report
The manuscript presents a throughout description of the kinematics of two groups of runners, divided by their Duty Factor. I think the manuscript would represent a great contribution to the community. I would suggest to improve the description of the methods and clarify some ideas in order to maximized its potential.
Comments:
Page 4, lines 144 -150 the method used to detect IC and FO was not chosen from the many algorithms that already exist and have been validated for this purpose (Handsaker et al., 2016; Leitch et al., 2011). Either a validated method should be used or the proposed method should be evaluated. This is particularly true for a paper that actually measures temporal parameters. Please explain the rationale behind this choice and I would suggest to include this issue in the limitations of this work, analysing the possible effects of the errors in detection.
Page 4, line 151: please explain the difference between ts and tf. From their definition both represent the swing time of the running cycle.
Page 4, line 158: The authors explain that the peak impact attenuation (IA) phase was selected as a fixed duration of 50 ms after foot strike. Is this still true for the slowest and the fastest running speed selected on this research? Normally, every gait phase changes its duration with speed (Hebenstreit et al., 2015), so presumably something similar happens with running. So please provide evidence that 50 ms is still a peak impact attenuation phase.
Page 4, line 160: Just after the explanation presented before, the authors mention that “Passive peak attenuation time (ta) was defined as the time from FS to IA”. So, was the IA phase fixed to 50 ms or measured somehow. If it was measure, please explain the method used.
Page 4, line 165: How is the COM calculated? Or is it extracted from Visual 3D directly?
Page 5, lines 181and 182: The authors mention: “The running gait cycle was defined as the vertical versus antero-posterior position of the average between the COM of the right foot segment and the one of the left foot segment.”
The sentence is a bit unclear, please rephrase it.
I assume the authors are talking about the “running cycle”, which would be defined as a time between consecutive FS of the same foot. Remember that “gait cycle” is a time measurement.
I don´t understand the “running gait cycle” phrase.
Also, it is not clear what they are describing (which seems to be a distance parameter, probably step length? Step high?).
Page 4, lines 151 onwards: the duration of each subphase of the running cycle (the running cycle described as tc + ts in the manuscript) depends on several variables that includes the duration of the running cycle itself. So, if interested in studying the proportion of each subphase and its change with speed, the subphases should be normalized to the running cycle (and expressed as a percentage of the cycle time). An analysis similar to the ones proposed by Hebenstreit et al could be done (Hebenstreit et al., 2015). Please explain the rationale behind the lack of normalization of the temporal parameters.
Conclusion
Page 4, line 515. The authors stated “DFlow runners, which rely on the re-use of elastic energy, were shown …”. Although it may be an idea derived from the results in this study, the fact that the runner would rely or not on the re-use of elastic energy is not a conclusion of this paper. I would suggest rephrasing the sentence.
References
Handsaker, J.C., Forrester, S.E., Folland, J.P., Black, M.I., Allen, S.J., 2016. a Kinematic Algorithm To Identify Gait Events During Running At Different Speeds and With Different Footstrike Types. J. Biomech. 49, 4128–4133. https://doi.org/10.1016/j.jbiomech.2016.10.013
Hebenstreit, F., Leibold, A., Krinner, S., Welsch, G., Lochmann, M., Eskofier, B.M., 2015. Effect of walking speed on gait sub phase durations. Hum. Mov. Sci. 43, 118–124.
Leitch, J.J., Stebbins, J.J., Paolini, G.G., Zavatsky, A.B.A.B., 2011. Identifying gait events without a force plate during running: a comparison of methods. Gait Posture 33, 130–132.
Author Response
General Response: We would like to thank the reviewer for the useful comments. The feedback provided was helpful in improving the quality of our manuscript during the revision process. We have adapted our manuscript as suggested by the comments of the reviewer. Indeed, we have provided an answer to each comment. In addition, to facilitate the review process, we have indicated all modifications in the manuscript in RED font color, except for where we removed text. We trust that the reviewer finds our responses meet the expectations.
Response to All Comments:
The manuscript presents a throughout description of the kinematics of two groups of runners, divided by their Duty Factor. I think the manuscript would represent a great contribution to the community. I would suggest to improve the description of the methods and clarify some ideas in order to maximized its potential.
Response: We thank the review and appreciate the very positive feedback he made.
Comments:
Page 4, lines 144 -150 the method used to detect IC and FO was not chosen from the many algorithms that already exist and have been validated for this purpose (Handsaker et al., 2016; Leitch et al., 2011). Either a validated method should be used or the proposed method should be evaluated. This is particularly true for a paper that actually measures temporal parameters. Please explain the rationale behind this choice and I would suggest to include this issue in the limitations of this work, analysing the possible effects of the errors in detection.
Response (continued): We thank the reviewer for this comment. We agree with the reviewer that the method has not been validated yet but this is an ongoing work in our laboratory. However, we would like to point to the reviewer that we have the following sentence in the next paragraph of our manuscript.
"All events were verified to ensure correct identification and were manually adjusted when required."
Nonetheless, we have also added the following statement in the limitations of this work.
"Finally, FS and TO events were defined from an algorithm which has not yet been validated. However, all events were verified to ensure correct identification and were manually adjusted when required. Nonetheless, a future study focusing on the validation of this algorithm should be conducted to avoid manual verification of all events."
Page 4, line 151: please explain the difference between ts and tf. From their definition both represent the swing time of the running cycle.
Response (continued): We thank the reviewer for this comment. As correctly pointed out by the reviewer, ts is the swing time of the running cycle. However, we have clarified the definition of tf. Indeed, tf represents the flight time of a step, i.e. from TO of a given foot to FS of the contralateral foot.
"tc and swing time (ts) were defined as the time from FS to TO and from TO to FS of the same foot, respectively, and tf as the time from TO of a given foot to FS of the contralateral foot."
Page 4, line 158: The authors explain that the peak impact attenuation (IA) phase was selected as a fixed duration of 50 ms after foot strike. Is this still true for the slowest and the fastest running speed selected on this research? Normally, every gait phase changes its duration with speed (Hebenstreit et al., 2015), so presumably something similar happens with running. So please provide evidence that 50 ms is still a peak impact attenuation phase.
Response (continued): We thank the reviewer for this comment. As correctly pointed out by the reviewer, we have kept 50 ms after foot strike to delimit the peak impact attenuation phase for all running speeds employed. We agree with the reviewer that the running gait changes its duration with speed. However, due to the absence of a force plate, we could not detect the presence of the passive impact peak for every runner and there exists a large variability between runners for the timing at which this peak occurs (Munro et al. 1987). Moreover, this peak does not exist for forefoot strikers. Therefore, we universally set this event to 50 ms for every runner. This specific time (50 ms) was selected following the consensus that the passive impact peak should appear within the first 50 ms of the running stride and should occur at 25 ms in average. Indeed, Munro et al. 1987 showed that the passive peak occurred around 25 ms for running speeds ranging from 3 to 5.5 m/s .
Carolyn F. Munro; Doris I. Miller; Andrew J. Fuglevand (1987). Ground reaction forces in running: A reexamination. , 20(2), 147–155. doi:10.1016/0021-9290(87)90306-x
Moreover, our results (not shown in the manuscript) have demonstrated that there was no difference in plantar flexion – dorsiflexion (x ankle in the following figure) between both DF groups after 50 ms for all running speeds, which confirm that the impact attenuation phase was ended at 50 ms.
We have added the following clarification in the manuscript.
"The end of the IA phase was set at 50 ms after FS for every runner due to several reasons: 1) the passive impact peak of forefoot strikers is not identifiable without a force platform, 2) runners showing a passive impact peak have a large variability in the timing of this peak around 25 ms, and 3) this peak was shown to occur around 25 ms for speeds ranging from 3 to 5.5 m/s [48].”
Page 4, line 160: Just after the explanation presented before, the authors mention that “Passive peak attenuation time (ta) was defined as the time from FS to IA”. So, was the IA phase fixed to 50 ms or measured somehow. If it was measure, please explain the method used.
Response (continued): We thank the reviewer for this comment. The reviewer correctly mentionned that passive peak attenuation time corresponds to 50 ms for every runner and all running speeds employed. We have modified the sentence in the manuscript to clarify our say.
"Passive peak attenuation time (ta) was defined as the time from FS to IA, i.e., 50 ms."
Page 4, line 165: How is the COM calculated? Or is it extracted from Visual 3D directly?
Response (continued): We thank the reviewer for this comment. The COM is extracted directly from Visual3D. In Visual3D software, segments are treated as geometric objects: segments are assigned inertial properties and COM locations based on their shape and attributed relative mass based on standard regression equations. Therefore, the whole-body COM is computed from the COMs and the masses of the segments. We have clarified the text to mention that COM is directly provided by Visual3D.
"Kinematic parameters were calculated using rigid-body analysis and whole-body COM location was calculated from the parameters of all 15 segments (COM was directly provided by Visual3D)."
Page 5, lines 181and 182: The authors mention: “The running gait cycle was defined as the vertical versus antero-posterior position of the average between the COM of the right foot segment and the one of the left foot segment.”
The sentence is a bit unclear, please rephrase it.
Response (continued): We thank the reviewer for this comment. We have changed the name running gait cycle to running wheel. Moreover, we have rephrased the sentence.
"The "running wheel", defined as the vertical position versus antero-posterior position of the COM of the foot segment (the motion of the foot COM in the sagittal plane), was calculated during the running stride."
I assume the authors are talking about the “running cycle”, which would be defined as a time between consecutive FS of the same foot. Remember that “gait cycle” is a time measurement.
I don´t understand the “running gait cycle” phrase.
Also, it is not clear what they are describing (which seems to be a distance parameter, probably step length? Step high?).
Response (continued): We thank the reviewer for this very useful comment. We agree with the reviewer that this part was not clear and that running cycle denotes a time measurement. In our case, we are describing the motion of the COM of the foot in the sagittal plane during the running cycle. Therefore, we have changed the name "running gait cycle" to "running wheel".
Page 4, lines 151 onwards: the duration of each subphase of the running cycle (the running cycle described as tc + ts in the manuscript) depends on several variables that includes the duration of the running cycle itself. So, if interested in studying the proportion of each subphase and its change with speed, the subphases should be normalized to the running cycle (and expressed as a percentage of the cycle time). An analysis similar to the ones proposed by Hebenstreit et al could be done (Hebenstreit et al., 2015). Please explain the rationale behind the lack of normalization of the temporal parameters.
Response (continued): We thank the reviewer for this comment. We agree with the reviewer that comparing temporal characteristics using relative (to running cycle) temporal values removes side effects due to variables such as cadence and/or amplitude. In our case, we have classified runners based on their DF. Therefore, runners are classified based on their relative tc, i.e., tc normalized to the running cycle. Having so, the statistical analysis of normalized tc and tf would not be allowed as they exactly represent the DF, i.e., the parameter used to construct the groups. Nonetheless, we agree with the reviewer that normalization could have been applied to the subphases while allowing statistical analyses. However, we have kept the temporal characteristics in absolute scale to make a distinction between the statistical analyses and the way our groups have been created (based on DF). Moreover, we would like to point out that when using DF to classify runners, the non-normalization of the subphases can only reduce the difference between the groups. For example, a DFhigh runner with a high cadence would get a smaller tc in absolute value while a DFlow runner with a high amplitude would get a smaller tf in absolute value. This means that these runners would be closer to each other when using tc or tf than when using DF. Finally, we agree with the reviewer that difference in temporal characteristics between groups constituted with something else than a temporal characteristic (e.g., sex, training status, ...) should be performed using relative (to running cycle) temporal values to remove side effects of variables such as cadence and/or amplitude.
Conclusion
Page 4, line 515. The authors stated “DFlow runners, which rely on the re-use of elastic energy, were shown …”. Although it may be an idea derived from the results in this study, the fact that the runner would rely or not on the re-use of elastic energy is not a conclusion of this paper. I would suggest rephrasing the sentence.
Response (continued): We thank the reviewer for this comment. We agree with the reviewer that this statement was not a direct result from the manuscript. Therefore, we have removed this statement from the conclusion and rephrase the sentence.
"DFlow runners were shown to exhibit a more extended lower limb than DFhigh due to a stiffer leg, which requires less range of motion at the knee and hip joints, and to be midfoot and forefoot strikers. On the contrary, DFhigh runners, which limit vertical displacement of the COM and promote forward propulsion, exhibited more lower limb flexion during tc than DFlow and were rearfoot strikers."
Round 2
Reviewer 1 Report
The authors investigated the kinematic profiles of two groups of runners. These groups are those who had higher duty factor and those who had lower duty factor within their sample. The authors hypothesized that the supposedly distinct duty factor groups would demonstrate unique kinematics of the lower extremity, particularly the sagittal plane angles, during stance. The revised version has addressed some but not all of the concerns of this reviewer. In particular, the choice to use a male-specific set of body segment parameters for all participants negatively impacts outcomes (i.e., center of mass) and confounds conclusions. The authors state that sex differences were not of interest here but the application of inappropriate body segment parameters to female participants is having an impact. Other methodological and interpretation shortcomings limit the potential impact of this work in its current form.
General Comments
The authors have made significant modifications to the tables and they are much more manageable.
The authors’ response to the review suggests the primary concern is whether there is greater flexion of the lower extremity in these two duty factor groups. In this light, this reviewer’s comment that unnecessary variables should be excluded stands. How does toe-in/out angle, or pronation/supination, etc. address, essentially, limb stiffness characteristics during stance? A focus on the primary research question is important. While interesting, any non-sagittal plane variables are not really relevant to answering the specific question the authors pose. COM displacement, running wheel, and sagittal plane variables alone with present a cohesive, logical addition to the literature. Indeed, only sagittal plane outcomes are listed in the conclusion paragraph.
Without an inverse dynamics analysis it is pure speculation as the source of power generation or absorption. The authors in several places propose sources of positive/negative work performed at various joints. This reviewer recommends not over-interpreting the kinematic results. A larger range of motion does not indicate greater torque at the joint, as suggested in Line 550 for example. This paragraph as a whole specifically has no citations provided to support the assertions.
The authors do integrate their work into the larger literature adequately in some places but less so in others. For example, the paragraph from lines 566-574 doesn’t provide satisfactory discussion of how the current results agree/contrast with previous work. This section could be broken into pieces and moved to other paragraphs on similar themes to bolster the discussion in those places and eliminate a free-floating list of prior outcomes.
Limitations include that the procedure for event selection is not validated. This is a major concern and should be the preliminary project the authors should undertake before attempting to get this work published. Would the outcomes have differed if all outcomes were manually selected? Perhaps a comparison of manual and automatic event detection could be a worthwhile effort.
Author Response
General Response: We would like to thank the reviewer for the useful comments. We have adapted our manuscript as suggested by the comments of the reviewer. Indeed, we have provided an answer to each comment. In addition, to facilitate the review process, we have indicated all modifications in the manuscript in RED font color, except for where we removed text. We trust that the reviewer finds our responses meet the expectations.
Response to All Comments:
The authors investigated the kinematic profiles of two groups of runners. These groups are those who had higher duty factor and those who had lower duty factor within their sample. The authors hypothesized that the supposedly distinct duty factor groups would demonstrate unique kinematics of the lower extremity, particularly the sagittal plane angles, during stance. The revised version has addressed some but not all of the concerns of this reviewer. In particular, the choice to use a male-specific set of body segment parameters for all participants negatively impacts outcomes (i.e., center of mass) and confounds conclusions. The authors state that sex differences were not of interest here but the application of inappropriate body segment parameters to female participants is having an impact. Other methodological and interpretation shortcomings limit the potential impact of this work in its current form.
Response: We thank the reviewer for this comment. We agree that the male-specific set of body segment parameters could be a limit and this is now acknowledged in the limitations section.
"Moreover, no sex distinction was taken into account within the full-body biomechanical model. More specifically, a male-specific set of body segment parameters was used in Visual3D [48,49], which could have impacted the outcomes such as the COM and could have confounded conclusions."
General Comments
The authors have made significant modifications to the tables and they are much more manageable.
The authors’ response to the review suggests the primary concern is whether there is greater flexion of the lower extremity in these two duty factor groups. In this light, this reviewer’s comment that unnecessary variables should be excluded stands. How does toe-in/out angle, or pronation/supination, etc. address, essentially, limb stiffness characteristics during stance? A focus on the primary research question is important. While interesting, any non-sagittal plane variables are not really relevant to answering the specific question the authors pose. COM displacement, running wheel, and sagittal plane variables alone with present a cohesive, logical addition to the literature. Indeed, only sagittal plane outcomes are listed in the conclusion paragraph.
Response: We thank the reviewer for this comment. We have removed the results which do not correspond to sagittal plane from the main text and put them in supplementary materials. As correctly pointed out by the reviewer, these outcomes do not directly reflect more flexion for DFhigh than DFlow. However, as these results are still very interesting, we refer the reader to supplementary materials in the main text.
"Nonetheless, segment angles in the frontal and transversal planes were presented in Supplementary Materials as they do not represent flexion-extension of the lower limb."
"The lower vertical kinetic energy at impact (shorter td and smaller ∆zCOM,d) and thus softer (less stiff) lower limb allows a higher pelvis mobility in DFhigh than DFlow and to rotate in the three anatomical planes (see Supplementary Materials) while DFlow requires, due to the higher vertical kinetic energy at FS, a stiffer lower limb than DFhigh, which limits the rotation in the sagittal plane."
Without an inverse dynamics analysis it is pure speculation as the source of power generation or absorption. The authors in several places propose sources of positive/negative work performed at various joints. This reviewer recommends not over-interpreting the kinematic results. A larger range of motion does not indicate greater torque at the joint, as suggested in Line 550 for example. This paragraph as a whole specifically has no citations provided to support the assertions.
The authors do integrate their work into the larger literature adequately in some places but less so in others. For example, the paragraph from lines 566-574 doesn’t provide satisfactory discussion of how the current results agree/contrast with previous work. This section could be broken into pieces and moved to other paragraphs on similar themes to bolster the discussion in those places and eliminate a free-floating list of prior outcomes.
Response: We thank the reviewer for this comment. We have removed our over-interpretation in several places. Moreover, we have grouped both paragraphs together.
"When focusing on the entire lower limb, a larger range of hip extension as well as a greater increase in the range of motion of hip extension with increasing running speed during tp was observed for DFhigh than for DFlow (Table 4). Besides, DFhigh demonstrated more hip flexion at FS, IA, and MS (Table 3) but similar range of motion during ta and tb than DFlow. In addition, the pelvis showed a higher range of retroversion during ta and tb and no change of retroversion with increasing running speed (Table 6) despite a statistically higher anteversion at FS for DFhigh than DFlow. The lower vertical kinetic energy at impact (shorter td and smaller ∆zCOM,d) and thus softer (less stiff) lower limb allows a higher pelvis mobility in DFhigh than DFlow and to rotate in the three anatomical planes (see Supplementary Materials) while DFlow requires, due to the higher vertical kinetic energy at FS, a stiffer lower limb than DFhigh, which suggests limiting the rotation in the sagittal plane. Furthermore, less knee flexion was observed for DFlow than DFhigh at MS but a similar one at FS, which led to a larger range of motion of the knee joint angle during tb and tp for DFhigh than DFlow (Table 4). Therefore, a larger leg compression and decompression was observed for DFhigh than DFlow despite the similar knee joint angle at FS. In addition, with increasing running speed, DFhigh depicted a larger increase of knee flexion at IA than DFlow as well as a larger increase in the range of motion of the knee joint during ta and tp (Table 4). These findings are potentially representing a more pronounced eccentric action of the quadriceps muscle for DFhigh than DFlow during tb. These results are in line with the observations of Gerritsen et al. (1995), which showed that the eccentric action of the knee extensor muscles was playing a role to attenuate the impact when rearfoot striking. Additionally, Knorz et al. (2017) have shown that different FS patterns change the distribution of loadings in a runner’s body. The authors concluded that a forefoot strike pattern was generally associated with a greater vertical maximum peak force but significantly lower loading rates in the ankle, knee, and hip joints, which is in line with the lower flexion and compression of the lower limb observed for DFlow than DFhigh. Similar studies were performed by other researchers and demonstrated that a forefoot strike pattern reduces knee loads whereas a rearfoot strike pattern reduces loads at the Achilles tendon [65–67] and ankle joint [68], again agreeing with less lower limb flexion for DFlow than DFhigh."
Limitations include that the procedure for event selection is not validated. This is a major concern and should be the preliminary project the authors should undertake before attempting to get this work published. Would the outcomes have differed if all outcomes were manually selected? Perhaps a comparison of manual and automatic event detection could be a worthwhile effort.
Response: We thank the reviewer for this comment. We agree with the reviewer that the events detection algorithm has not been validated yet but this is an ongoing work in our laboratory. Indeed, we are currently measuring whole-body 3D biomechanics using 3D cameras and a force platform in order to be able to perform a complete validation of our proposed algorithm. However, we would like to mention to the reviewer that we would have obtained exactly the same events by having only manually detecting them. This is due to the fact that all events were manually adjusted, if necessary, after having used our algorithm. Nonetheless, we agree with the reviewer that having a validated algorithm would then avoid the tedious process of manual adjustment.
Finally, we would like to point to the reviewer that, due to the fact that all events were manually adjusted, the validation of the algorithm was not necessary in this specific case. This statement was already written in our manuscript.
"All events were verified to ensure correct identification and were manually adjusted when required."
"Finally, FS and TO events were defined from an algorithm which has not yet been validated. However, all events were verified to ensure correct identification and were manually adjusted when required. Nonetheless, a future study focusing on the validation of this algorithm should be conducted to avoid manual verification of all events."
Reviewer 2 Report
I thank the authors for the clarifications made on the paper.
I would still recommend to include the rationale behind the lack of use of normalization of the espatio temporal parameters.
Author Response
General Response: We would like to again thank the reviewer for the useful comments. We have adapted our manuscript as suggested by the comments of the reviewer. Indeed, we have provided an answer to each comment. In addition, to facilitate the review process, we have indicated all modifications in the manuscript in RED font color, except for where we removed text. We trust that the reviewer finds our responses meet the expectations.
Response to All Comments:
I thank the authors for the clarifications made on the paper.
I would still recommend to include the rationale behind the lack of use of normalization of the spatio-temporal parameters.
Response: We thank the reviewer for this comment. We have added the rationale behind the lack of use of normalization of the spatio-temporal parameters.
Note of Table 2
"Duration were kept in absolute scale, i.e., were not normalized, to make a distinction between the statistical analyses and the way our groups were created (based on DF, i.e., tc normalized to the running cycle)."
Round 3
Reviewer 1 Report
The authors investigated the kinematic profiles of two groups of runners. These groups are those who had higher duty factor and those who had lower duty factor within their sample. The authors hypothesized that the supposedly distinct duty factor groups would demonstrate unique kinematics of the lower extremity, particularly the sagittal plane angles, during stance. The revised version has addressed the major concerns of this reviewer. Some minor English language edits would improve readability.
General Comments
One instance where readability would be improved through minor modification is Lines 28-30. See below an alternative to eliminate the ‘his’ pronoun and ‘its’ when referring to the runners. Consider:
This highlights that runners seem to self-optimize their running gait to minimize the metabolic cost of running [1,9–11].
Another example is number agreement. One example is on line 59. …no difference in running economy *has* been reported… Verb and noun should both be singular or both be plural.
Author Response
General Response: We would like to thank the reviewer for the useful comments. We have adapted our manuscript as suggested by the comments of the reviewer. Indeed, we have provided an answer to each comment. In addition, to facilitate the review process, we have indicated all modifications in the manuscript in RED font color, except for where we removed text. We trust that the reviewer finds our responses meet the expectations.
Response to All Comments:
The authors investigated the kinematic profiles of two groups of runners. These groups are those who had higher duty factor and those who had lower duty factor within their sample. The authors hypothesized that the supposedly distinct duty factor groups would demonstrate unique kinematics of the lower extremity, particularly the sagittal plane angles, during stance. The revised version has addressed the major concerns of this reviewer. Some minor English language edits would improve readability.
General Comments
One instance where readability would be improved through minor modification is Lines 28-30. See below an alternative to eliminate the ‘his’ pronoun and ‘its’ when referring to the runners. Consider:
This highlights that runners seem to self-optimize their running gait to minimize the metabolic cost of running [1,9–11].
Response: We thank the reviewer for this comment. We have adapted the manuscript according to the suggestion of the reviewer. Moreover, we went through the entire manuscript and try to improve its readability.
"This highlights that runners seem to self-optimize their running gait to minimize the metabolic cost of running [1,9-11]."
Another example is number agreement. One example is on line 59. …no difference in running economy *has* been reported… Verb and noun should both be singular or both be plural.
Response: We thank the reviewer for this comment. We have adapted the manuscript according to the suggestion of the reviewer.
"Similarly to what was observed when grouping runners based on their global running pattern, no difference in running economy has been reported while using more local and finer parameters, such as FS angle [20] or strike pattern [21–23]."