Locomotor Adaptation Training to Prevent Mobility Disability

Round 1

Reviewer 1 Report

Summary:

The manuscript describes an experiment in which the authors compared a control group with no exercise to split-belt and traditional treadmill interventions in healthy but sedentary older adults to determine which most improved mobility in the community. The main differences between conditions include a greater improvement in TUG from the split-belt group, whereas the traditional treadmill group, but not the split-belt group, improved gait efficiency and speed. I’m not sure if the improvement in efficiency is clinically significant, the improvement in gait speed is pretty close to a clinically meaningful difference, which you could maybe argue is meaningful for non-impaired population. Overall, however, the authors correctly state that there’s not a clear overall benefit to split-belt training.

Overall impression:

The authors present a clinical study of split-belt vs. traditional treadmill training that is well thought out in terms of methodology, addresses the important issue of mobility among older adults, and is very thorough in the measurements taken. It’s also well referenced, and I appreciated the opportunity to read it. It does have potential to be a meaningful publication, but I think major revision is necessary to improve it to that point.

The authors present this as a cautionary tale about split-belt walking interventions, but I don’t think a cautionary tale is necessary, particularly in this population. Split-belt interventions were initially proposed for asymmetric clinical populations like stroke survivors (see Reisman’s work with Bastian in 2007, 2010, 2013) and gait deficits in the elderly affect both sides about the same amount. The authors would therefore need to make a more compelling case we should have expected a unique benefit to split-belt walking in the elderly before the experiment. We already have cautionary tales about split-belt walking compared to other clinical interventions in populations it was actually designed to help (e.g. Lewek et al., 2018) and have known that transfer of learning from split-belt to overground can be difficult, especially with sudden changes to belt speed (e.g. Torres-Oviedo & Bastian, 2012).

The authors briefly discuss adaptability, but it’s not clear that the types of adaptability needed during community ambulation are best addressed by the split-belt intervention used in the study. The 3 split-belt treadmill interventions are described in basic detail, but the reasons these particular protocols were chosen is unclear. More detail about the noisy split-belt condition is necessary, but I think a rationale for all of them is necessary.

On a related note, the discussion mentions differences in tests (e.g. ceiling effects) when interpreting why some results are positive and others are not, but don’t discuss any reasons why the different conditions could have produced different test results (e.g. more improvement on TUG for the split-belt group, but improved gait speed for the traditional group).

Additionally the authors describe the split-belt intervention as feasible but difficult to control. Towards the end of the discussion (lines 454-460), you state that the split-belt participants had more trouble increasing speed or heart rate as the training progressed. First, these data should first be presented in the results, as should data about responders & non-responders. Also, in general, there is not enough information about absolute belt speeds during the interventions. A potential reason why traditional treadmill training improves speed more is that this subject group could walk faster during the actual intervention, but I can’t tell it that’s true from the results section.

I think this could be a useful addition to addition to the existing literature, but not simply as a cautionary tale. I have some other, more specific revisions below, but I would suggest that significant general revisions need to be made, and it’s likely that I would re-review a subsequent submission with further specifics suggested.

My comments

Intro:  general hypotheses, they state that adaptability to complex environments is important and claim that unilateral adaptability is necessary, but only cite a stroke study and a textbook (with no page #s indicated). Of course, stroke survivors need unilateral adaptability; they had an injury that causes asymmetric disability, which isn’t generally the case with the elderly.

Methods:

Line 93: how were gait speeds determined?

Section 2.2: were the functional tests conducted by a PT or someone with experience in this sort of testing?

Lines 191-192: Elsewhere, this is described as a 16-week intervention, but you have the split-belt group split into 3, 4-week cycles. I may be missing this, but what happened with this group during the other 4 weeks. Another figure or table to show how the conditions are broken up over the 16-week intervention could be helpful in addressing this issue.

Lines 191-192: It would be great to hear the rationale behind each split-belt condition somewhere in the manuscript. I imagine there was some basis for designing the 3 split-belt conditions this way to improve community ambulation. It’s necessary to briefly describe these to justify why?

Line 204: how was limb dominance determined?

Results:

Section 3.2: It’s unclear why this section or figure 4 are in the main section of the manuscript. I’d recommend putting it in an appendix. The intervention is walking, so there’s less reason to expect a group effect with cognition than with gait parameters. Do the cognitive test results make up some of the 26 variables that the authors consider when stating that this intervention didn’t “dramatically impact” mobility outcomes? If so, I don’t think their being fair to their own study. There is certainly evidence that exercise helps the brain, but I don’t know if I’d expect 16 weeks of walking 3x/week to significantly improve cognitive function by itself, and there seems to be no reason why split-belt training would improve cognitive function more than traditional treadmill training. Why then, is it the first figure in the results? In contrast, there are significant interactions in spatiotemporal parameters (e.g. pre vs. post walking speed in Appendix E) that are buried in appendices where readers are unlikely to look.

There’s not enough information about how fast, on average, participants walked during each condition. It looks like you determined the fast belt speed for the split-belt condition using a participant’s maximum comfortable walking speed (lines 196-197). If you averaged the speed between the 2 legs, my guess is it would be markedly slower that the traditional group. Could this contribute to why only the traditional group improved gait efficiency and increased gait speed (Appendix E)?

Discussion:

General comments:

It is true that your results demonstrate, overall, that split-belt is no better than traditional treadmill training, but there should be more discussion of differences between the significant results that you did see. For example, what is it about your split-belt intervention that could have caused it to improve TUG? Why did the traditional treadmill training improve gait efficiency and walking speed? It seems that the split-belt intervention may have been designed with the goal of challenging and therefore improving balance/adaptability, although there aren’t enough details about the rationale behind your split-belt intervention and the different types of split-belt conditions to really assess this.

Specific lines

Lines 333-334: Should we expect ‘dramatic’ improvement in a non-impaired population after a 16-week intervention? This seems like a minor word choice, but I think it’s important since it’s so early in the discussion & kind of sets expectations for the rest of the section.

Lines 384-386: walking speeds on the treadmill during the intervention are certainly a possible confounding factor. I recommend reporting speed differences between intervention groups so that the readers can figure out whether this confounds the gait speed effects or not

Lines 421-422: Since there seems to be no reason given by the authors why split-belt training would improve cognitive function more than traditional treadmill training, why is this finding counterintuitive? As you state in the next sentence, evidence indicates that locomotor adaptation is controlled by a completely different area of the brain.

Appendices:

If appendix A is to be included, it needs revision such that someone uninvolved with the study could understand it. For example, the first column lists “gradual” and “abrupt,” conditions which are mentioned nowhere else in the manuscript. What do these conditions mean?

Author Response

Please see attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

I appreciate negative results and the given articles shows a high quality, which probably results from multiple rejected submissions.

So I have not much to add to this sound article. Except:

* Please specify further characteristics in Table 1, such as MMSE

* Please include the results of your power calculation (e.g. via p-power), you have probaly conducted initially?

* Your discussion is rather long but totally meaningful. However, could you please add a conclusion chapter, where you briefly summarize the main points to be enhanced in future (e.g. randomising the groups not only by age and gender, but as well by the level of "fitness" based on the inital assessments)?

Congrats!

Author Response

Please see attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Abstract: I suggest including the information that the participants were randomly assigned to the three different groups.

Page 3, line 103: “Mini Mental State Exam” Please change to “Mini-Mental State Examination” as used on page 3, line 87.

Page 4, line 160: “Control participants were offered the option of partaking in the exercise program following completion of all testing sessions”: I assume that this means the control participants is offered this option after study completion?! Which “exercise program” was offered, the split-belt or normal treadmill training? Please specify.

Design: In my point of view, the study is designed as a randomized controlled trial, so I would suggest providing this information in in the title or abstract.

Statistical analyses: The authors presents only p-values for the results of their statistical analyses. As the groups are quite small, which is associated with reduced statistical power, I would suggest presenting also effect sizes for the results. E.g. Some F-values are quite high, but there is a lack of statistical power to reach significant.

Please include also a rational for the sample size. Was it based on a rather pragmatic approach (e.g. inclusion of as many as possible in a given time period) or an apriori sample size calculation?

Page 5, line 251/252: What were the reasons for the dropouts (58 enrolled but only 38 completed the study)?

Page 8, line 323: “…stride length changes are presented in Figure 7.” Figure 7 presents not only stride length changes but also those in walking speed, and step width. Please revise.

Page 10, line 342: “While split-belt treadmill training is generally feasible in individuals at risk for mobility disability”. It remains unclear to me how the authors examined the feasibility of the split-belt training? There is no information provided how well the split-belt training could be performed with the participants, e.g. were there any adverse events, problems during training; what about the training adherence or the  completion rate in the split-belt training group compared to the other groups, etc.?

Results: The sample description provided in table 1 should include more information no only gender, age, weight and height. I would suggest providing at least one assessment for each domain to give the reader a more detailed picture on the sample.

“As overground walking speed was our primary outcome”. If this was the case, please mention it at an earlier stage of the manuscript.  Please define this primary outcome in the method section.

Author Response

Please see attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Overall impression of the first revision:

The changes made to the manuscript greatly improved it. Revisions to the introduction and the addition of lines 229-236 help set up the rationale for conducting this study such that, even with negative results, it can still be a meaningful contribution to the literature. My revisions below don’t involve the broader themes of the study, which are acceptable, but focus on a few specific details which I believe will add clarity when the manuscript is published.

Abstract:

Lines 27-28: The final sentence is a bit vague as to what you refer to as being feasible. It would be clearer to state “while the intervention is feasible…”, but perhaps this was shortened to meet a word limit. If so, I suggest simply removing “while feasible”

Intro:

Line 58-59: Turning requires a few strides in which step lengths & times are asymmetric different lengths. This is not the same thing as “adaptation of step length…”  Adaptation involves a gradual change of a parameter over several strides in response to sensory feedback errors (see Bastian’s 2008 review in Current Opinions in Neurology for a more detailed description. Therefore, this sentence is not accurate. Possibly “adaptability” would be a better term to use here

Line 60: “Falls are more likely during turns” requires a reference

Line 95: I believe there’s a typo which should be “patterns” instead

Methods:

Line 227-229: I made a comment in my first review of this manuscript that it was unclear what happened in the last 4 weeks given that there are only 3 walking conditions and 4, 4-week cycles. Appendix A in your revision is much improved and helps address this, and I certainly acknowledge the difficulty in creating a figure to capture such a complex training paradigm in 1 page. However, the sentence itself could still be confusing. As a practical matter – since not all readers will go to your appendix – I would suggest clearing up any confusion by restating with something such as “The split-belt group was programmed into four, 4-week cycles, with each cycle including the three different walking conditions (1: constant split-belt, 2: noisy split-228 belt, 3: acute perturbation; See appendix A).”

Results:

The additional tables and inclusion of improved this section considerably. In hindsight, my comment on this section from my first review comes off as too dismissive of the cognitive results. They are fine to include in the main part of the manuscript, I just thought that the significant gait results should be included as well. I have no suggested revisions for this section.

Discussion:

The changes made to this section between the first draft and the current draft are satisfactory to address my concerns. I have no further suggested revisions here

Author Response

We thank the Reviewer for their suggestions - they have helped strengthen this manuscript. We have amended the word choices in our revised manuscript as recommended.

Reviewer 3 Report

The authors appropriatly addressed all my concerns and revised the manuscript accordingly. The revisions of the authors significantly improved the manusript. I do not have any further comments.

Author Response

We would like to thank the Reviewer for their comments on this manuscript - their feedback has helped strengthen it.