Review Reports - Association of Lower-Limb Asymmetry with Falls, Physical Performance and Frailty in Older Adults: A Systematic Review

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This is a well-conducted and clearly presented systematic review addressing a clinically relevant and relatively underexplored question: the association between lower-limb interlimb asymmetry and falls, physical performance, and frailty in older adults. The manuscript has several methodological strengths, including protocol registration, adherence to PRISMA 2020 and SWiM recommendations, independent study selection and risk-of-bias assessment, and the use of GRADE to assess the certainty of evidence across exposure–outcome pairs.
The decision not to perform a meta-analysis is adequately justified given the substantial heterogeneity in the definition, calculation, and categorization of asymmetry and the incompatibility of the reported effect measures. The narrative synthesis is appropriately structured, and I particularly appreciate that the interpretation gives greater weight to prospective studies, larger samples, adjustment for muscle capacity, and risk of bias rather than relying simply on the number of statistically significant findings.
The conclusions are appropriately cautious and consistent with the results. In particular, the authors correctly emphasize that lower-limb asymmetry should not currently be considered a standalone marker of fall risk and that no universal pathological asymmetry threshold can be established from the available evidence.
I have only a few minor comments that may further improve the manuscript:
1. The authors should consider highlighting even more clearly in the Abstract and Discussion that the certainty of evidence was rated as very low for all eight exposure–outcome comparisons. This is particularly important when discussing the apparently more consistent association between muscle mass asymmetry and gait speed.
2. The manuscript appropriately acknowledges that the search was not updated before submission. Given the rapidly evolving literature and the search date of 2 August 2026, the authors may wish to consider updating the search prior to final publication.
3. The use of different risk-of-bias tools according to study design is reasonable. However, the authors created numerical thresholds to classify NOS and AXIS scores as low, moderate, or high risk of bias. Since these thresholds are not intrinsic to the original instruments, their rationale and the potential implications of this categorization could be emphasized more explicitly as a methodological limitation.
4. Because several included studies evaluated multiple asymmetry indices and outcomes, the discussion of potential selective reporting and multiplicity is important. It would be useful to emphasize this issue when interpreting isolated positive findings from small studies.
5. The clinical implications section could distinguish even more explicitly between the potential value of asymmetry as a complementary assessment parameter and its currently unsupported use as an independent screening or diagnostic criterion for fall risk.
Overall, this is a comprehensive, methodologically careful, and clinically relevant systematic review. The manuscript provides a useful synthesis of a heterogeneous literature and appropriately avoids overstating the available evidence.

Author Response

Thank you for the time and effort you devoted to reviewing our work. We appreciate all your efforts to improve the quality of our article. We have addressed all comments and revised the manuscript throughout in accordance with your comments and recommendations. All changes are shown in tracked-changes mode in the revised manuscript. Thank you once again for reviewing our work.

Comment: The authors should consider highlighting even more clearly in the Abstract and Discussion that the certainty of evidence was rated as very low for all eight exposure–outcome comparisons. This is particularly important when discussing the apparently more consistent association between muscle mass asymmetry and gait speed.

A: It has been corrected. The Abstract now states very low certainty for all pairs, explicitly including muscle mass–gait speed; Section 4.2 and the Conclusions specify the cross-sectional design.

Comment: The manuscript appropriately acknowledges that the search was not updated before submission. Given the rapidly evolving literature and the search date of 2 August 2026, the authors may wish to consider updating the search prior to final publication.

A: The search is under three months old, within the Cochrane 12-month standard; we will rerun it before acceptance if requested.

Comment: The use of different risk-of-bias tools according to study design is reasonable. However, the authors created numerical thresholds to classify NOS and AXIS scores as low, moderate, or high risk of bias. Since these thresholds are not intrinsic to the original instruments, their rationale and the potential implications of this categorization could be emphasized more explicitly as a methodological limitation.

A: Thank you for your suggestion. It has been corrected. Section 2.5 explains the rationale and descriptive use of the thresholds; Section 4.6 discusses boundary misclassification and notes that GRADE relied on item-level limitations.

Comment: Because several included studies evaluated multiple asymmetry indices and outcomes, the discussion of potential selective reporting and multiplicity is important. It would be useful to emphasize this issue when interpreting isolated positive findings from small studies.

A: Thank you for your suggestion. Section 3.9 lists positive labels resting on single indices among several; Sections 3.9 and 4.5 state that such findings are hypothesis-generating.

Comment: The clinical implications section could distinguish even more explicitly between the potential value of asymmetry as a complementary assessment parameter and its currently unsupported use as an independent screening or diagnostic criterion for fall risk.

A: Thank you for your suggestion. Section 4.7 now opens by distinguishing asymmetry as an unvalidated complementary parameter from its unsupported use as an independent screening or diagnostic criterion.

Reviewer 2 Report

Comments and Suggestions for Authors

1. Eligibility criteria need clearer justification. This review excluded populations with unilateral pathologies such as stroke, brain injury, amputation and cerebral palsy. Rationale for these exclusions should be explicitly stated in Methods. Importantly, conclusions from this review apply only to older adults without overt unilateral neurological or musculoskeletal lesions. This limitation should be highlighted in Abstract, Introduction and Discussion, to prevent inappropriate extrapolation to neurorehabilitation cohorts e.g. post‑stroke hemiparesis. Formal definition of spontaneous inter‑limb asymmetry should be provided to distinguish it from asymmetry caused by structural tissue damage.
2. Reporting details for study screening should be supplemented. Although two reviewers performed screening and a third arbitrated disagreements, inter‑rater reliability (Kappa statistics) for title‑abstract and full‑text screening is missing, which is recommended under PRISMA‑2020. The frequency of disagreements resolved by third assessor should also be added. Grey literature and trial registries were not searched as pre‑specified; please further elaborate the associated risk of publication bias in Limitations.
3. Execution details for SWiM narrative synthesis need improvement. The manuscript tabulates studies simply by direction of effect (positive, no association, measure‑dependent, inverse). SWiM discourages simple “vote‑counting”. Explicitly emphasize in text that counting study numbers serves only for categorization and must not be interpreted as an indicator of evidence certainty. Please also describe how overlapping publications from the same participant dataset (e.g. two James et al. papers) were handled to avoid over‑weighting duplicated evidence.
4. Interpretation of risk‑of‑bias assessment requires refinement. The authors applied NOS for cohort/case‑control studies and AXIS for cross‑sectional studies with self‑defined cut‑offs. It should be clearly stated that total summary scores from these two instruments are not directly comparable. Individual item flaws are more informative than aggregate total scores. Many case‑control studies relied on retrospectively self‑reported falls with reversed‑causality risk and lacked adjustment for absolute muscle capacity. Please interpret the observation that larger, better‑quality studies tended to yield null findings, and discuss how selection and reporting bias may shape the existing literature landscape.
5. Justification for GRADE downgrading should be more granular. All exposure‑outcome pairs ended up with “very low” certainty. The tables only mark “serious / very serious” downgrade, but textual explanation for each downgrade decision remains insufficient. For the muscle‑mass‑asymmetry vs gait‑speed pair, all included studies are cross‑sectional, which cannot establish temporality; the indirectness concern deserves expanded discussion. Distinguish whether low certainty originates from primary‑study limitations or from review‑level synthesis constraints.
6. Methodological heterogeneity of asymmetry indices deserves deeper discussion. Primary studies adopted diverse formulas for asymmetry calculation, different reference‑limb choices; 15–20% cut‑offs originating from sports medicine are frequently imported without validation in older populations. Please systematically compare consequences brought by different calculation approaches. Highlight the critical interpretive pitfall: identical percentage asymmetry can originate from bilateral weakness OR one strong / one weak limb, which carry completely distinct clinical meanings; relative asymmetry indices alone cannot differentiate these two scenarios.
7. Temper causal claims and separate statistical association from causal inference. Most available evidence is cross‑sectional or case‑control; only a small number of prospective cohorts exist. Even significant association cannot answer whether asymmetry is a cause or a consequence of functional decline. While the practical‑implication section correctly warns against using asymmetry index alone for fall screening, several passages implicitly suggest that correcting asymmetry would improve function. Revise the full manuscript to remove causal‑intervention speculations; emphasize that whether reducing asymmetry improves patient‑relevant outcomes still awaits interventional evidence.
8. The threshold / non‑linear‑effect hypothesis must be consistently labelled as post‑hoc exploratory hypothesis. The authors hypothesize that adverse functional consequences of asymmetry may only emerge beyond a certain threshold, and moderate‑level asymmetry could even be beneficial. This is generated post‑hoc from observational data rather than pre‑registered in primary studies. Throughout the manuscript, repeatedly mark this as untested hypothesis rather than established clinical fact.
9. Heterogeneity across outcome measures should be further emphasized. Differentiate incident vs recurrent falls, prospectively monitored falls vs retrospectively self‑reported falls. For physical performance outcomes, gait speed, TUG and SPPB show markedly different sensitivity to asymmetry effects. Please discuss how task characteristics (balance demand, dual‑task load) modify detectability of asymmetry‑related effects, to inform outcome selection for future primary research.
10. Subgroup analyses are limited in the current synthesis. Included datasets show unbalanced sampling: the largest cohort exclusively enrolled men, several positive findings derive from female‑only samples; participants cover community‑dwelling, senior‑home and long‑term‑care older adults. Add a dedicated paragraph in Discussion to summarize potential modifying effects of sex, living setting and baseline functional status, and point out existing evidence gaps for subgroup‑specific effects.
11. Future‑research recommendations need to be more concrete and operational. Move beyond general statements. Suggest concrete directions: (1) prospective studies should pre‑register analytical plans to mitigate selective reporting driven by multiple testing; (2) report absolute bilateral limb values alongside relative asymmetry indices; (3) separate incident and recurrent falls as distinct outcomes; (4) threshold / non‑linear hypotheses must be pre‑specified rather than post‑hoc exploratory; (5) interventional trials should evaluate patient‑centered clinical endpoints (falls, physical function), not merely changes in asymmetry metrics.
12. Supplementary reporting requirements. Double‑check numbers in PRISMA flow‑diagram match text descriptions. Attach complete PRISMA‑2020 checklist as supplementary material with page‑number cross‑references. As LLM assistance was declared for manuscript preparation, fully disclose LLM scope, content and human revision process in Acknowledgements / Author Contributions per journal requirements.

Author Response

Comment: Eligibility criteria need clearer justification. This review excluded populations with unilateral pathologies such as stroke, brain injury, amputation and cerebral palsy. Rationale for these exclusions should be explicitly stated in Methods. Importantly, conclusions from this review apply only to older adults without overt unilateral neurological or musculoskeletal lesions. This limitation should be highlighted in Abstract, Introduction and Discussion, to prevent inappropriate extrapolation to neurorehabilitation cohorts e.g. post-stroke hemiparesis. Formal definition of spontaneous inter-limb asymmetry should be provided to distinguish it from asymmetry caused by structural tissue damage.

A: Thank you for your suggestion. It has been corrected. The rationale for E7/E4 was added, the target asymmetry defined, and the scope limitation added to the Abstract, Section 4.6 and Conclusions.

Comment: Reporting details for study screening should be supplemented. Although two reviewers performed screening and a third arbitrated disagreements, inter-rater reliability (Kappa statistics) for title-abstract and full-text screening is missing, which is recommended under PRISMA-2020. The frequency of disagreements resolved by third assessor should also be added. Grey literature and trial registries were not searched as pre-specified; please further elaborate the associated risk of publication bias in Limitations.

A: Thank you for your suggestion. It has been corrected. Kappa values and disagreement counts are now reported, although PRISMA 2020 does not mandate them. The consequences of not searching registers and grey literature are discussed.

Comment: Execution details for SWiM narrative synthesis need improvement. The manuscript tabulates studies simply by direction of effect (positive, no association, measure-dependent, inverse). SWiM discourages simple “vote-counting”. Explicitly emphasize in text that counting study numbers serves only for categorization and must not be interpreted as an indicator of evidence certainty. Please also describe how overlapping publications from the same participant dataset (e.g. two James et al. papers) were handled to avoid over-weighting duplicated evidence.

A: Thank you for your suggestion. It has been corrected. Sections 2.8 and 3.4 and the footnotes to Tables 2–3 state that counts are not vote counting. The James reports are not treated as independent replications.

Comment: Interpretation of risk-of-bias assessment requires refinement. The authors applied NOS for cohort/case-control studies and AXIS for cross-sectional studies with self-defined cut-offs. It should be clearly stated that total summary scores from these two instruments are not directly comparable. Individual item flaws are more informative than aggregate total scores. Many case-control studies relied on retrospectively self-reported falls with reversed-causality risk and lacked adjustment for absolute muscle capacity. Please interpret the observation that larger, better-quality studies tended to yield null findings, and discuss how selection and reporting bias may shape the existing literature landscape.

A: Thank you for your suggestion. It has been corrected. Section 2.5 states the totals are not comparable and that item-level judgements take precedence. A new Section 4.5 paragraph interprets the small-study pattern.

Comment: Justification for GRADE downgrading should be more granular. All exposure-outcome pairs ended up with “very low” certainty. The tables only mark “serious / very serious” downgrade, but textual explanation for each downgrade decision remains insufficient. For the muscle-mass-asymmetry vs gait-speed pair, all included studies are cross-sectional, which cannot establish temporality; the indirectness concern deserves expanded discussion. Distinguish whether low certainty originates from primary-study limitations or from review-level synthesis constraints.

A: Thank you for your suggestion. It has been corrected. Each downgrade is now explained in a lettered footnote to Table 4.

Comment: Methodological heterogeneity of asymmetry indices deserves deeper discussion. Primary studies adopted diverse formulas for asymmetry calculation, different reference-limb choices; 15–20% cut-offs originating from sports medicine are frequently imported without validation in older populations. Please systematically compare consequences brought by different calculation approaches. Highlight the critical interpretive pitfall: identical percentage asymmetry can originate from bilateral weakness OR one strong / one weak limb, which carry completely distinct clinical meanings; relative asymmetry indices alone cannot differentiate these two scenarios.

A: Thank you for your suggestion. It has been corrected. Section 4.5 shows how the formula changes magnitude for identical data. Section 4.1 illustrates how one index can reflect bilateral weakness or a unilateral deficit.

Comment: Temper causal claims and separate statistical association from causal inference. Most available evidence is cross-sectional or case-control; only a small number of prospective cohorts exist. Even significant association cannot answer whether asymmetry is a cause or a consequence of functional decline. While the practical-implication section correctly warns against using asymmetry index alone for fall screening, several passages implicitly suggest that correcting asymmetry would improve function. Revise the full manuscript to remove causal-intervention speculations; emphasize that whether reducing asymmetry improves patient-relevant outcomes still awaits interventional evidence.

A: Thank you for your suggestion. It has been corrected. The title now refers to association rather than impact, causal wording was revised and Section 4.7 states that asymmetry-targeted interventions cannot currently be recommended.

Comment: The threshold / non-linear-effect hypothesis must be consistently labelled as post-hoc exploratory hypothesis. The authors hypothesize that adverse functional consequences of asymmetry may only emerge beyond a certain threshold, and moderate-level asymmetry could even be beneficial. This is generated post-hoc from observational data rather than pre-registered in primary studies. Throughout the manuscript, repeatedly mark this as untested hypothesis rather than established clinical fact.

A: Thank you for your suggestion. It has been corrected.

Comment: Heterogeneity across outcome measures should be further emphasized. Differentiate incident vs recurrent falls, prospectively monitored falls vs retrospectively self-reported falls. For physical performance outcomes, gait speed, TUG and SPPB show markedly different sensitivity to asymmetry effects. Please discuss how task characteristics (balance demand, dual-task load) modify detectability of asymmetry-related effects, to inform outcome selection for future primary research.

A: Thank you for your suggestion. It has been corrected.

Comment: Subgroup analyses are limited in the current synthesis. Included datasets show unbalanced sampling: the largest cohort exclusively enrolled men, several positive findings derive from female-only samples; participants cover community-dwelling, senior-home and long-term-care older adults. Add a dedicated paragraph in Discussion to summarize potential modifying effects of sex, living setting and baseline functional status, and point out existing evidence gaps for subgroup-specific effects.

A: Thank you for your suggestion. It has been corrected. A dedicated paragraph in Section 4.5 covers sex, setting and baseline function, notes that sex and measurement domain are confounded, and identifies the missing subgroup-specific evidence.

Comment: Future-research recommendations need to be more concrete and operational. Move beyond general statements. Suggest concrete directions: (1) prospective studies should pre-register analytical plans to mitigate selective reporting driven by multiple testing; (2) report absolute bilateral limb values alongside relative asymmetry indices; (3) separate incident and recurrent falls as distinct outcomes; (4) threshold / non-linear hypotheses must be pre-specified rather than post-hoc exploratory; (5) interventional trials should evaluate patient-centered clinical endpoints (falls, physical function), not merely changes in asymmetry metrics.

A: Thank you for your suggestion. It has been corrected.

Comment: Supplementary reporting requirements. Double-check numbers in PRISMA flow-diagram match text descriptions. Attach complete PRISMA-2020 checklist as supplementary material with page-number cross-references. As LLM assistance was declared for manuscript preparation, fully disclose LLM scope, content and human revision process in Acknowledgements / Author Contributions per journal requirements.

A: Thank you for your suggestion. It has been corrected. The flow numbers were rechecked and are consistent with the text. The completed checklist is supplied and will reflect the revised line numbers. The AI-related statement is included in both the Acknowledgments and the Methods section.

Reviewer 3 Report

Comments and Suggestions for Authors

Summary

This systematic review examines associations between lower-limb asymmetry and falls, physical performance, and frailty in older adults, synthesizing evidence from 26 observational studies involving 10,167 participants. Although the review is interesting, there are severe methodological limitations that need to be addressed.

Abstract

Although the abstract acknowledges that the certainty of evidence is very low, the reporting of predominantly positive associations may give readers an overly confident impression of the findings. Please distinguish more clearly between exploratory associations and robust prognostic evidence, emphasizing that the current evidence does not establish the independent predictive value of lower-limb asymmetry.

The abstract should state that meta-analysis was not possible due to incompatible effect measures, that the majority of positive findings came from case-control studies with retrospective fall ascertainment, and that the largest prospective cohort found no association. Without these disclosures, the abstract overstates the coherence of the evidence and misleads readers about the certainty of the conclusions.

Introduction

The introduction identifies an important gap concerning whether lower-limb asymmetry provides information beyond absolute muscle strength, power, and mass. However, it would benefit from distinguishing these related constructs and explaining how incremental prognostic value should be established. Please clarify how the review's objectives and eligibility criteria address this distinction, particularly given the observational nature of the available evidence.

The introduction would be strengthened by articulating what conflicting findings or clinical dilemmas justify this review, and why the five-domain framework (A–E) was chosen a priori. Additionally, the introduction mentions that asymmetry indices were developed in sports science and may not apply to geriatric populations (lines 61–65), but this critical point is not developed into a clear rationale for the review.

Methods

The Methods section should include a dedicated subsection detailing deviations from the original study protocol. Despite the protocol being published prior to study commencement, the manuscript does not specify whether any deviations occurred, nor does it explain their rationale or how they were addressed. The authors should explicitly report any deviations from the prespecified protocol, or clearly state if none occurred, to ensure transparency and facilitate assessment of the study’s methodological rigor.

The decision not to conduct a meta-analysis is understandable given the incompatible exposure definitions and effect measures. However, classifying findings by direction of effect risks obscuring differences in effect magnitude, precision, statistical power, and clinical relevance. Please strengthen the synthesis by presenting comparable effect estimates and confidence intervals wherever possible, explaining how conflicting results were weighed, and clarifying how the direction-based approach informed the GRADE assessments. The decision to merge exposure domains after data extraction should also be justified transparently in relation to the registered protocol.

The methods section states that meta-analysis was abandoned because "the condition necessary for deriving a common effect measure was not met" (lines 248–250), and that this decision was made "at the data extraction stage" (line 248). However, the review was registered in PROSPERO with a protocol that presumably specified a quantitative synthesis. The authors do not report whether the protocol was amended, whether the deviation was documented, or whether the registered protocol anticipated this possibility. This is a significant methodological concern because post hoc changes to analysis plans can introduce bias and undermine the credibility of the review. The authors should: (1) report the original registered analysis plan; (2) explain when and how the decision to abandon meta-analysis was made; (3) document whether the protocol was amended and, if so, how; and (4) justify why a narrative synthesis based on direction of effect is the most appropriate alternative rather than, for example, a formal vote-counting or a narrative synthesis based on effect estimates with clear reporting of heterogeneity. Additionally, the methods state that "two sets of domains were merged" after data extraction (lines 274–282), which is another deviation from the registered plan. These deviations should be transparently reported and justified in the methods, not buried in the results section.

Eligibility criterion I3 admits asymmetry merely "co-occurring" with the outcome, which is incompatible with the stated prognostic aim. Only 4 of 26 included studies are prospective cohorts, and most fall outcomes were ascertained retrospectively; including cross-sectional co-occurrence as eligible evidence conflates association with prediction and compromises every prognostic claim (and indirectness is not consistently downgraded for this in GRADE).Please either restrict the prognostic synthesis to studies with a documented temporal sequence (presenting cross-sectional evidence separately as associative), or revise the aim, objectives (A3), title, and abstract so that the claims match the admissible evidence.

Results

Sixteen of the 26 included studies were cross-sectional, and the positive findings concerning falls were predominantly derived from case–control studies relying on retrospective self-report. These designs cannot reliably establish temporal relationships and are vulnerable to recall bias and reverse causation. Please make the distinction between prospective and retrospective findings more prominent throughout the results and avoid interpreting cross-sectional associations as evidence of predictive validity.

The results section classifies findings into four direction categories, but for many studies, the actual effect estimates and confidence intervals are either not reported or are reported inconsistently. For example, in Section 3.5.1, the authors report that Portegijs et al. found a "5.5-fold higher likelihood" (line 530), but for many other studies, only p-values or qualitative descriptions are provided (e.g., "no significant relationship," "weak relationship").

The synthesis tables report direction labels without any effect estimates or measures of precision, and the "measure-dependent" category can conceal within-study selective reporting. Tables 1–3 contain no effect sizes—despite Section 2.6 stating estimates would accompany the direction classification—and several "positive" or "measure-dependent" labels rest on a single significant index among many tested

Discussion

The discussion appropriately acknowledges the limitations of the evidence, but the clinical interpretation of the observed associations requires greater restraint. All eight exposure–outcome pairs were rated as having very low certainty, and adjustment for absolute neuromuscular capacity was limited across the included studies. Please strengthen the discussion of residual confounding, measurement reliability, and the inability to establish independent prognostic value. In particular, clarify that the observed associations do not establish whether asymmetry improves fall-risk prediction beyond established functional measures, and ensure that recommendations for clinical assessment remain proportionate to the available evidence.

The Discussion does not address footwear as a potential confounder or effect modifier, despite its direct relevance to the review's exposure and outcomes. All included studies calculated asymmetry indices from gait, loading, or functional measurements performed in participants' own shoes, yet none reported the type, fit, or condition of footwear worn during assessment. Footwear is an established independent determinant of plantar pressure distribution, gait symmetry, postural stability, and fall risk in older adults, as established here:

https://pubmed.ncbi.nlm.nih.gov/16645298/

https://pubmed.ncbi.nlm.nih.gov/41402434/

This should be acknowledged and briefly discussed.

Section 3.7 and 4.4 introduce a threshold/non-linear hypothesis that was explicitly not part of the registered analysis plan. The discussion presents this as a plausible explanatory model despite all GRADE ratings being very low and despite the largest high-quality studies showing no or inverse associations. Clinical implications are discussed without acknowledging that no universal threshold can be derived.

Author Response

Comment: Although the abstract acknowledges that the certainty of evidence is very low, the reporting of predominantly positive associations may give readers an overly confident impression of the findings. Please distinguish more clearly between exploratory associations and robust prognostic evidence, emphasizing that the current evidence does not establish the independent predictive value of lower-limb asymmetry.

A: Thank you for your suggestion. It has been corrected. We removed the vote-count sentence, framed the findings as exploratory associations, and now state that independent predictive value is not established.

Comment: The abstract should state that meta-analysis was not possible due to incompatible effect measures, that the majority of positive findings came from case-control studies with retrospective fall ascertainment, and that the largest prospective cohort found no association. Without these disclosures, the abstract overstates the coherence of the evidence and misleads readers about the certainty of the conclusions.

A: Thank you for your suggestion. It has been corrected.

Comment: The introduction identifies an important gap concerning whether lower-limb asymmetry provides information beyond absolute muscle strength, power, and mass. However, it would benefit from distinguishing these related constructs and explaining how incremental prognostic value should be established. Please clarify how the review's objectives and eligibility criteria address this distinction, particularly given the observational nature of the available evidence.

A: Thank you for your suggestion. It has been corrected. The Introduction now distinguishes absolute capacity from its interlimb distribution.

Comment: The introduction would be strengthened by articulating what conflicting findings or clinical dilemmas justify this review, and why the five-domain framework (A–E) was chosen a priori. Additionally, the introduction mentions that asymmetry indices were developed in sports science and may not apply to geriatric populations (lines 61–65), but this critical point is not developed into a clear rationale for the review.

A: Thank you for your suggestion. It has been corrected. We added the rationale for the five-domain framework and for questioning athlete-derived thresholds in older adults. The conflicting findings were already summarized in the Introduction.

Comment: The Methods section should include a dedicated subsection detailing deviations from the original study protocol. Despite the protocol being published prior to study commencement, the manuscript does not specify whether any deviations occurred, nor does it explain their rationale or how they were addressed. The authors should explicitly report any deviations from the prespecified protocol, or clearly state if none occurred, to ensure transparency and facilitate assessment of the study’s methodological rigor.

A: Thank you for your suggestion. It has been corrected. The new Section 2.12 reports all deviations from the protocol.

Comment: The decision not to conduct a meta-analysis is understandable given the incompatible exposure definitions and effect measures. However, classifying findings by direction of effect risks obscuring differences in effect magnitude, precision, statistical power, and clinical relevance. Please strengthen the synthesis by presenting comparable effect estimates and confidence intervals wherever possible, explaining how conflicting results were weighed, and clarifying how the direction-based approach informed the GRADE assessments. The decision to merge exposure domains after data extraction should also be justified transparently in relation to the registered protocol.

A: Thank you for your suggestion. It has been corrected. Sections 2.6 and 2.11 explain how conflicting results were weighed and how direction of effect informed GRADE.

Comment: The methods section states that meta-analysis was abandoned because "the condition necessary for deriving a common effect measure was not met" (lines 248–250), and that this decision was made "at the data extraction stage" (line 248). However, the review was registered in PROSPERO with a protocol that presumably specified a quantitative synthesis. The authors do not report whether the protocol was amended, whether the deviation was documented, or whether the registered protocol anticipated this possibility. This is a significant methodological concern because post hoc changes to analysis plans can introduce bias and undermine the credibility of the review. The authors should: (1) report the original registered analysis plan; (2) explain when and how the decision to abandon meta-analysis was made; (3) document whether the protocol was amended and, if so, how; and (4) justify why a narrative synthesis based on direction of effect is the most appropriate alternative rather than, for example, a formal vote-counting or a narrative synthesis based on effect estimates with clear reporting of heterogeneity. Additionally, the methods state that "two sets of domains were merged" after data extraction (lines 274–282), which is another deviation from the registered plan. These deviations should be transparently reported and justified in the methods, not buried in the results section.

A: Thank you for your suggestion. It has been corrected.

Comment: Eligibility criterion I3 admits asymmetry merely "co-occurring" with the outcome, which is incompatible with the stated prognostic aim. Only 4 of 26 included studies are prospective cohorts, and most fall outcomes were ascertained retrospectively; including cross-sectional co-occurrence as eligible evidence conflates association with prediction and compromises every prognostic claim (and indirectness is not consistently downgraded for this in GRADE).Please either restrict the prognostic synthesis to studies with a documented temporal sequence (presenting cross-sectional evidence separately as associative), or revise the aim, objectives (A3), title, and abstract so that the claims match the admissible evidence.

A: We revised the aim, title and abstract. Prognostic inferences are now restricted to prospective cohorts, and cross-sectional and case–control findings are explicitly treated as associative.

Comment: Sixteen of the 26 included studies were cross-sectional, and the positive findings concerning falls were predominantly derived from case–control studies relying on retrospective self-report. These designs cannot reliably establish temporal relationships and are vulnerable to recall bias and reverse causation. Please make the distinction between prospective and retrospective findings more prominent throughout the results and avoid interpreting cross-sectional associations as evidence of predictive validity.

A: Thank you for your suggestion. It has been corrected.

Comment: The results section classifies findings into four direction categories, but for many studies, the actual effect estimates and confidence intervals are either not reported or are reported inconsistently. For example, in Section 3.5.1, the authors report that Portegijs et al. found a "5.5-fold higher likelihood" (line 530), but for many other studies, only p-values or qualitative descriptions are provided (e.g., "no significant relationship," "weak relationship").

A: Thank you for your suggestion. It has been corrected.. Section 3.3 now states where only p-values or qualitative statements were available.

Comment: The synthesis tables report direction labels without any effect estimates or measures of precision, and the "measure-dependent" category can conceal within-study selective reporting. Tables 1–3 contain no effect sizes—despite Section 2.6 stating estimates would accompany the direction classification—and several "positive" or "measure-dependent" labels rest on a single significant index among many tested

A: Thank you for your suggestion. It has been corrected. Section 2.6 was corrected.

Comment: The discussion appropriately acknowledges the limitations of the evidence, but the clinical interpretation of the observed associations requires greater restraint. All eight exposure–outcome pairs were rated as having very low certainty, and adjustment for absolute neuromuscular capacity was limited across the included studies. Please strengthen the discussion of residual confounding, measurement reliability, and the inability to establish independent prognostic value. In particular, clarify that the observed associations do not establish whether asymmetry improves fall-risk prediction beyond established functional measures, and ensure that recommendations for clinical assessment remain proportionate to the available evidence.

A: Thank you for your suggestion. It has been corrected. A new Section 4.5 paragraph addresses residual confounding, reliability and the absence of incremental-prediction testing; Section 4.7 distinguishes complementary from standalone use.

Comment: The Discussion does not address footwear as a potential confounder or effect modifier, despite its direct relevance to the review's exposure and outcomes. All included studies calculated asymmetry indices from gait, loading, or functional measurements performed in participants' own shoes, yet none reported the type, fit, or condition of footwear worn during assessment. Footwear is an established independent determinant of plantar pressure distribution, gait symmetry, postural stability, and fall risk in older adults, as established here:

https://pubmed.ncbi.nlm.nih.gov/16645298/

https://pubmed.ncbi.nlm.nih.gov/41402434/

This should be acknowledged and briefly discussed.

A: Thank you for your suggestion. It has been corrected. Footwear does not affect dynamometry or body-composition measures, so ‘all studies’ is inaccurate. We added footwear as a potential confounder for gait and loading measures [49,50].

Comment: Section 3.7 and 4.4 introduce a threshold/non-linear hypothesis that was explicitly not part of the registered analysis plan. The discussion presents this as a plausible explanatory model despite all GRADE ratings being very low and despite the largest high-quality studies showing no or inverse associations. Clinical implications are discussed without acknowledging that no universal threshold can be derived.

A: Thank you for your suggestion. It has been corrected. The hypothesis is labelled post hoc throughout, with alternative explanations added. Section 4.7 now states that no universal threshold can be derived.

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

All comments have been well addressed, thank you