Skip to main content
Download PDF

The effect of exercise-induced muscle fatigue on gait parameters among older adults: a systematic review and meta-analysis

European Review of Aging and Physical Activity volume 22, Article number: 4 (2025) Cite this article

Abstract

Background

Exercise-induced fatigue is a common consequence of physical activities. Particularly in older adults, it can affect gait performance. Due to a wide variety in fatiguing protocols and gait parameters used in experimental settings, pooled effects are not yet clear. Furthermore, specific elements of fatiguing protocols (i.e., intensity, duration, and type of activity) might lead to different changes in gait parameters. We aimed to systematically quantify to what extent exercise-induced fatigue alters gait in community-dwelling older adults, and whether specific elements of fatiguing protocols could be identified.

Methods

This systematic review and meta-analysis was conducted in accordance with the PRISMA guidelines. In April 2023, PubMed, Web of Science, Scopus, Cochrane and CINAHL databases were searched. Two independent researchers screened and assessed articles using ASReview, Rayyan, and ROBINS-I. The extracted data related to spatio-temporal, stability, and variability gait parameters of healthy older adults (55 +) before and after a fatiguing protocol or prolonged physical exercise. Random-effects meta-analyses were performed on both absolute and non-absolute effect sizes in RStudio. Moderator analyses were performed on six clusters of gait parameters (Dynamic Balance, Lower Limb Kinematics, Regularity, Spatio-temporal Parameters, Symmetry, Velocity).

Results

We included 573 effect sizes on gait parameters from 31 studies. The included studies reflected a total population of 761 older adults (57% female), with a mean age of 71 (SD 3) years. Meta-analysis indicated that exercise-induced fatigue affected gait with a standardized mean change of 0.31 (p < .001). Further analyses showed no statistical differences between the different clusters, and within clusters, the effects were non-uniform, resulting in an (indistinguishable from) zero overall effect within all clusters. Elements of fatiguing protocols like duration, (perceived) intensity, or type of activity did not moderate effects.

Discussion

Due to the (mainly) low GRADE certainty ratings as a result of the heterogeneity between studies, and possible different strategies to cope with fatigue between participants, the only conclusion that can be drawn is that older adults, therapist, and researchers should be aware of the small to moderate changes in gait parameters as a result of exercise-induced fatigue.

Introduction

Muscle fatigue, or performance fatigability, is a natural consequence of physical activities and exercise (e.g., walking) and may lead to a reduction in physical performance [1], particularly in the older population [2]. Multiple reviews have been performed on the effects of exercise-induced (muscle) fatigue in older adults with regards to gait, functional tasks, dual task performance and standing balance. In general, these reviews show that exercise-induced fatigue negatively affects physical performance outcomes. For instance, gait stability-, gait variability-, standing balance, and spatio-temporal parameters change when fatigued [3,4,5,6,7], indicating that fatigue could lead to an increased risk of tripping and falls [8, 9]. However, the heterogeneity in fatiguing protocols and gait performance outcomes makes it hard to draw consistent conclusions on the consequences of exercise-induced fatigue on gait performance. As no meta-analysis has yet been conducted on the effects of exercise-induced fatigue on gait parameters in older adults, the pooled effects of gait parameters and their directions are currently unknown. Furthermore, in this study we aimed to evaluate the heterogeneity in fatiguing protocols between studies to pinpoint specific elements of fatiguing exercises that influence gait performance.

Generally, the heterogeneity stemming from fatiguing protocols accounts for the duration of the exercise, the (perceived) intensity of the exercise and/or the type of activity. For example, most fatiguing protocols either involve high intensity for short duration exercises or low intensity for long duration exercises [10,11,12,13,14,15]. It is suggested that these two combinations induce a different type of fatigue, namely peripheral fatigue or central fatigue, respectively [16,17,18,19], with different underlying pathways [20, 21]. The differences between fatiguing protocols is also found in the type of activity, which can be indicated by the number of muscles involved or the type of contraction(s). In previous reviews, a range from single muscle (isokinetic) exercises to more whole body, cyclic, activities such as walking are described [3,4,5,6,7]. Since muscle fatigue is specific to task demands, the type of activity, together with intensity and duration, are considered important factors for the limiting adjustments that come with fatigue [22,23,24].

This review aimed to make clear if, regardless of fatiguing protocol, exercise-induced muscle fatigue in general will change gait parameters in older adults. Furthermore, the variation in fatiguing protocols across studies could shed light on the question of whether elements of these protocols contribute to changes in specific gait parameters. It seems plausible that, depending on the type of activity, as well as the intensity or the duration of the exercise, exercise-induced fatigue leads to different outcomes of gait parameters. Therefore, the aim of our systematic review and meta-analysis was to evaluate and quantify to what extent exercise-induced muscle fatigue alters gait parameters in community-dwelling older adults and whether specific elements of fatiguing protocols (i.e., type of activity as well as intensity or duration of the exercise) lead to different outcomes. First, and in line with previous research, we expected that gait stability, gait variability, and spatio-temporal parameters change as a consequence of the fatiguing exercises, regardless of the elements of fatiguing protocols. Second, we hypothesized that when the intensity of the protocols was (perceived as) high, a more severe level of fatigue would result in significantly greater changes in gait parameters, compared to low or medium intensity protocols. Third, we expected to find greater changes in gait parameters after protocols with longer durations, since in general longer duration leads to more depletion and thereby more fatigue. Fourth, we hypothesized that protocols that used walking activities lead to greater changes in the gait parameters compared to non-walking protocols, due to task-specificity. Analyzing which elements of fatiguing exercises contribute to changes in gait could provide insight into how older adults should prevent themselves from adverse fatiguing effects. These findings might be of use for (preventive) interventions for falls in the older population.

Methods

This systematic review and meta-analysis were set up following the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) 2020 guidelines [25], the Cochrane Handbook for systematic reviews of interventions [26] and the Open Science Guidelines [27]. It was pre-registered with the International Prospective Register of systematic reviews (PROSPERO) (registration number: CRD42022357662).

Search strategy

In April 2023, a systematic literature search in PubMed, Web of Science, Scopus, Cochrane, and the Cumulative Index to Nursing and Allied Health Literature (CINAHL) was performed. The search strategy included a combination of medical subject headings (MeSH) terms and similar keywords concerning (1) the population: “old*”, “senior*”, “elder*”, (2) the intervention: “fatigue”, “muscle fatig*”, “peripheral fatig*”, (3) the outcomes: “walking”, “gait stability”, “step length” and (4) exclusion criteria: “Parkinson*”, “COVID*”, “Stroke”. The search strings were built up with the use of Boolean operators (“AND”, “OR”, and “NOT”) and adapted to specific search engines. The full search strategies for the different databases are found in the Supplementary Materials (1. Search plans). Although the Cochrane Handbook states that the “NOT” operator should be avoided [26], we did include the operator in this search and used the “NOT” operator in combination with the field code “Title” to filter out articles based on exclusion criteria in the title. Thereby, we aimed to remove a substantial number of irrelevant papers since many studies concerning muscle fatigue and gait parameters were conducted in populations with neurological and/or orthopaedical diseases and, hence, not relevant to our study’s aims. Additional filters or restrictions were not used. Next to the searches, all available reference lists were examined. Furthermore, a grey literature search by contacting authors from the field asking for unpublished data or ongoing data using standardized templates was performed, following the recommendations by Moreau and Gamble [27].

Study selection

We included experimental, quasi-experimental, and observational studies for further screening when they met the following inclusion criteria: (1) participants were non- or pre-frail community-dwelling older adults (55 + years of age) without comorbidities that would influence walking abilities, such as cardiovascular-, orthopedic- or neurological diseases, and (2) effects of exercise-induced fatigue or prolonged activity on gait parameters were compared with a non-fatigued state within participants.

The results of the search were imported into EndNote 20 [28]. After removing duplicates, all eligible articles were imported into the algorithm-aided open science software ASReview Lab [29]. For the learning phase of the selection process, we utilized input from a scoping search and reviews from other authors in the same field. After reading the full texts of 20 articles, we marked 10 articles as relevant and 10 articles (near misses) as irrelevant. Two authors (PV and RO) independently screened titles and abstracts using ASReview Lab. Since the software re-orders the articles after every decision and moves relevant articles up in ranking, only a fragment of the full set of articles needed to be screened. We decided to screen at least 10% and then screen up to 100 consecutive irrelevant articles [30, 31], applying the default learning model: “Naïve Bayes, TF-IDF, Max” [32]. The articles identified as potentially eligible after screening of titles and abstracts, were full-text analyzed independently by two authors (PV and RO) using the open science software Rayyan [33].

During the screening and full-text analysis, discrepancies were discussed by the two authors and when necessary, a third author from the team was asked for help (BV). Interrater reliability was calculated for both title and abstract screening and full-text analysis after the first search (0.66 and 0.60, respectively). The R script, calculations, and CIs of the interrater reliability are found in the Supplementary Materials (2. Interrater-reliability (irr) screening, 2. Interrater-reliability (irr) fulltext, 2. irr_screening, and 2. irr_fulltext).

Quality assessment

The articles that remained after the full-text analysis were assessed on risk of bias using the ROBINS-I tool [34, 35]. The risk of bias assessment and GRADE criteria were used to rate the quality of the studies found, from “very low” to “high” GRADE certainty ratings [36]. GRADE considerations include limitations of studies, inconsistencies, lack of precision, indirectness, and publication bias. Two authors (PV and RO) independently assessed the risk of bias, GRADE was done by one researcher (PV) and checked by another (RO). Differences were solved by discussion and when necessary, a third author (JB) was asked for help.

Data extraction

From the eligible articles, two authors (PV and RO) extracted information about study design, participant demographics (average age and percentage of females), elements of fatiguing activity and protocol (i.e., duration, type of activity, (perceived) intensity), gait parameters (i.e., spatio-temporal-, gait variability-, gait stability- or joint kinematic-parameters), pre- and post-fatigue and their standard deviations. The data were saved in a coding form that was built in Microsoft Excel. The two authors (PV and RO) checked each other’s work and added or adjusted data when necessary. When possible, the extracted outcomes were converted to the same unit of measure (e.g., centimeters converted to meters). Not every paper included the required data. If attempts to obtain data from the authors failed, data was imputed or hand-measured. Hand-measuring was done if figures were available. The imputation of data was based on averaged values per variable across available studies and implemented only if values were missing at random; otherwise, incomplete study data were deleted case-wise.

Data synthesis

To test the first hypothesis that gait parameters change in response to fatiguing protocols, first, the negative effect sizes were transformed into their absolute values, and second, the transformed negative effect sizes and positive effect sizes were pooled.

To test the other hypotheses on the large number of effect-sizes, the non-absolute (i.e., raw) effect sizes were analyzed in clusters of gait parameters. All gait parameters were grouped into six clusters. These clusters were made based upon the previous research by Lindemann [37] and Dapp et al. [38] (Fig. 1 (p. 2)). In our classification: (1) gait parameters were placed into the Symmetry cluster when they matched with Lindemann’s paper or when they could be labelled in the Phase cluster as defined by Hollman [39]. Since the Phase cluster overlaps with Lindemann’s Symmetry cluster [38] all data was merged into the same cluster. (2) Gait-stability outcomes were grouped into the Dynamic Balance cluster. (3) Furthermore, the Foot Movement cluster was renamed to Lower Limb Kinematic cluster, and combined foot movement parameters with other lower limb kinematic parameters. For clarity purposes, (5) the Walking Capacity cluster and (6) the Coordination cluster were renamed to Velocity cluster and Spatio-temporal Parameters cluster respectively.

Only if ten or more effect sizes from different studies were available in the same cluster, results were pooled for meta-analysis.

Statistical analysis

A priori analysis

Our a priori power analysis showed that at least 47 studies were needed (see the Supplementary Materials (6. Power analysis a priori) for more details). Regardless of the number of articles, the intended subgroup- (i.e., per cluster) and moderator analysis, via meta-regression, was performed.

Main analysis

It was expected that the included studies would vary in the methods used to induce fatigue, as well as in the measurements and type of gait parameters. Because of this, the analysis was conducted using the standardized raw mean change (SMCR) as the effect size measure and random-effects model was used to fit to the data. If between-measurement correlation related to a given SMCR was not reported, the correlation was imputed from an average correlation of the available data. In case of a substantial amount of missing correlation data, we rerun analysis with different imputed between measurements correlations (i.e., r = 0.00, r = 0.25, r = 0.50, r = 0.75, and r = 0.90), assuming that the true SMCR depends strongly on the value of the correlations between measurements (cf. [40]).

The level of heterogeneity (i.e., τ2) was estimated using the restricted maximum-likelihood estimator (REML) [41]. In addition to the estimate of τ2, the Q-test for heterogeneity [42], the I2 statistic [43] and prediction intervals were reported. The analysis was carried out using open science software R [44], RStudio (version 4.2.3) [45], and the metafor R package (version 4.4–0) [46]. Additional R packages that were used are: clubSandwich [47], dmetar [48], tidyverse [49] and gridExtra [50]. A multilevel model was applied to account for the dependencies between effect sizes reported in the same study.

Moderator analysis

Per cluster, moderator analyses were done to test if the duration of the protocols (longer than 10 min vs. shorter than 10 min), the rate of (perceived) exertion by the participants ((low vs. moderate vs. high), based on the RPE scale (low = lowest 33% of RPE scale, moderate = middle 33% of RPE scale, high = highest 33% of RPE scale), on heart rate, decrease in muscle capacity, on the description of the fatiguing protocol with words like until exhaustion, or a combination of these factors), and type of activity (walking vs. non-walking) affected the changes in gait parameters. For visualization of the findings, the orchaRd R package [51] was used to create orchard plots and metafor [46] to create forest plots.

Sensitivity analysis

To test the level of publication bias, for example, p-hacking analysis, the following R packages were used: metaplus [52], publicationbias [53], phacking [54, 55] and multibiasmeta [56].

Post hoc power analysis

A post hoc power analysis was conducted using the POMADE R package [57]. When the project started, power analysis for multilevel models was not available, so we decided to run a power analysis once POMADE became available. This additional power analysis is a more precise estimation of power than the initial power analysis conducted before the current project started, as the POMADE procedure specifically accounted for dependencies between the effect size.

Results

Study selection

The search resulted in 43,679 studies. Duplicates were removed, and 27,661 articles were screened using ASReview. After screening 10% of the found articles, we already had around 1500 consecutive irrelevant articles, leading to the decision to stop. The screening resulted in 50 articles that were manually labeled as relevant, and 2767 articles were manually labeled as irrelevant. All other articles were labeled irrelevant by the machine learning algorithm. After screening of full-text, 28 articles were included. Reasons for exclusion were wrong study population, inappropriate study design, irrelevant outcomes, or incomplete data. The citation search resulted in one additional study and e-mail contact with authors yielded two additional articles. Eventually, we included 31 articles (Fig. 1).

Fig. 1
figure 1

PRISMA flowchart showing how the 31 studies were included in this meta-analysis, figure created with PRISMA2020 [58]

Full size image

Quality Assessment

Five out of 31 studies had a moderate risk of bias rating, all other studies were considered to have a low risk of bias (see Fig. 2). The GRADE certainty ratings indicated low risk for all outcomes combined. In the Dynamic Balance and Spatio-temporal Parameters cluster the GRADE certainty rating was considered “very low”. The ratings were considered “low” in the Lower Limb Kinematics, Regularity, and Symmetry clusters and the GRADE rating turned out to be “moderate” in the Velocity cluster. More details on GRADE can be found in the Supplementary Materials (3. GRADE outcomes table).

Fig. 2
figure 2

Risk of bias assessed with the ROBINS-I tool, figure created with robvis [59]

Full size image

Study characteristics

A total of 573 effect sizes on gait parameters were included, with an average of 18.5 effect sizes per study. The included studies were conducted on a total population of 761 older adults with a mean age of 70.9 (± 2.8) years of age, and 57% of all participants were female. On average, there were 24.5 participants per study. Study characteristics are in the summary of findings table (Table 1; and for more details view the Supplementary Materials (4. Summary of Findings table)).

Table 1 Summary of findings table. Alphabetically ordered
Full size table

Since both Odonkor et al. [76] and Kushioka et al. [74] reported on the same dataset and two papers of Rocha dos Santos et al. [82, 83] also used the same dataset, the outcomes were treated in our analysis as if they were from one study. Therefore, the analysis was conducted on 29 datasets instead of 31 studies.

Meta-analysis outcomes

Since many between-measurement correlations could not be retrieved, the analyses were rerun using different imputed between-measurement correlations, namely r = 0.00, r = 0.25, r = 0.50, r = 0.75, and r = 0.90. The focus was placed on results which were based on a conservative assumption of the between-measurement correlations; we only report here the SMCR calculated based on r = 0.25. The analyses with the different various between-measurement correlations (r = 0.00, r = 0.50, r = 0.75, and r = 0.90) can be found in the Supplementary Materials (7. Outcomes meta-analysis).

Imputation of data was done for 9 effect sizes; only standard deviations and no mean effects were imputed.

Overall effects on gait parameters

When analyzing absolute values of gait parameters, a minimal to moderate change (SMCR = 0.31, 95% confidence interval (CI) [0.24, 0.37], p < 0.01, k = 573) was detected when post-fatiguing walking was compared to pre-fatiguing walking (Table 2 and Fig. 3).

Table 2 Changes in gait as result of exercise-induced fatigue, analysis on absolute data (r = 0.25)
Full size table
Fig. 3
figure 3

Orchard plot of the absolute data with all 573 gait parameters, data from 29 data sets. The orchard shows the estimate as an open circle and the confidence interval as the horizontal black line. The number of effect sizes is represented by k and the number of data sets is between parentheses. The size of the color filled circles represents precision of the effect size

Full size image

For visualization purposes, a forest plot with absolute gait parameters aggregated per study is presented (Fig. 4).

Fig. 4
figure 4

Forest plot with aggregated absolute outcomes per study, listed by effect size. Note that the estimate in this forest plot is different than the SMCR in Table 2; the outcomes in this forest plot were aggregated per study before analysis

Full size image

Clustering effect sizes

All the gait parameters were grouped into six different clusters: Dynamic Balance, Lower Limb Kinematics, Regularity, Spatio-temporal Parameters, Symmetry, and Velocity. See Table 3 for an overview of the clusters, underlying gait parameters, and distribution of effect sizes.

Table 3 Overview of clusters of gait parameters
Full size table

Effects on clustered data

Analysis of the clusters with absolute data showed that fatigue had the larger effect on Velocity (SMCR = 0.42, 95% CI [0.29, 0.54], p < 0.001, k = 26, n = 15)), and the least effect on Dynamic Balance (SMCR = 0.23, 95% CI [0.12, 0.34], p < 0.001, k = 45, n = 14). Nonetheless, moderator analysis showed no statistical differences between the clusters (p = 0.19; view Table 4 for more details).

Table 4 Moderator analysis of clustered absolute data (r = 0.25)
Full size table

Analysis of non-absolute data by means of the random effects model analysis, showed that in all the clusters the pooled estimate indicated “no to a small change” effects (minimum SMCR = –0.01, 95% CI [–0.10, 0.09], p = 0.90, k = 136, n = 26 (Spatio-temporal Parameters); maximum SMCR = 0.24, 95% CI [–0.01, 0.49], p = 0.06, k = 26, n = 15 (Velocity)). Nevertheless, pre- vs. post-fatiguing changes in all the clusters were statistically non-significant and thereby the effects are considered non-distinguishable from zero (Table 5).

Table 5 Changes in gait as result of exercise-induced fatigue, per cluster of gait parameters with non-absolute data (r = 0.25)
Full size table

The variance in the observed effects that reflects variance in true effect can be derived from the values of I2 [89]. The proportion of variance is shown to be “might not be important” for the clusters: Regularity, Dynamic Balance, and Symmetry (approximately 12%, 32%, and 39%, respectively) and “may represent substantial heterogeneity” for Spatio-temporal Parameters, Velocity and Lower Limb Kinematics (approximately 56%, 65%, and 67%, respectively) [26]. Although Q-statistic was not significant in the Dynamic Balance and Regularity clusters, we did assume that heterogeneity was present in all the clusters. One of the reasons was the range in the effect sizes, as visualized in the orchard plots, and another reason was the values of the prediction intervals (PIs), which included opposing effects in all the clusters (Fig. 5).

Fig. 5
figure 5

Orchard plots per cluster with non-absolute data, indicating that in all clusters the effects are considered non-distinguishable from zero and statistically non-significant. Besides, heterogeneity is assumed to be present in all clusters. This is reflected by the estimate (open circle), the confidence interval (horizontal thick black line), and the prediction interval (horizontal thin line). The number of effect sizes is represented by k and the number of data sets is between parentheses. The size of the color filled circles corresponds with the precision of the effect size

Full size image
3.4.3.1 Moderator analysis

An analysis of the three moderators (i.e., type of activity, duration, and (perceived) intensity) on each of the six clusters resulted in only one statistically significant moderator (Tables 6, 7 and 8). Within the Regularity cluster (perceived) intensity turned out to be a significant moderator (p < 0.01, k = 24, n = 8). In particular, the type of protocol that was (perceived) as Low Intensity showed an increase in regularity outcomes post-fatiguing protocol/task (SMCR = 0.49, 95% CI [0.18, 0.80], p < 0.01, k = 2, n = 1) compared to general outcomes of the Regularity cluster (SMCR = 0.03, 95% CI [–0.09, 0.15], p = 0.64, k = 60, n = 11). However, the number of the effect sizes from these data sets was low, which might be too low to achieve sufficient statistical power for such an analysis (Table 8).

Table 6 Outcomes of moderator analyses of ‘type of activity’ on different clusters of gait parameters (r = 0.25)
Full size table
Table 7 Outcomes of moderator analyses of ‘duration’ on different clusters of gait parameters (r = 0.25)
Full size table
Table 8 Outcomes of moderator analyses of ‘(perceived) intensity’ on different clusters of gait parameters (r = 0.25)
Full size table

Sensitivity analyses and analysis of bias

3.5.1 P-curve analysis

The p-curve analysis shows that most of the effect sizes were non-significant, and the analysis shows an ability to detect true effects with the power estimation: 88%, 95% CI [83%, 92%] (Fig. 6).

Fig. 6
figure 6

P-curve analysis on all (absolute) effect sizes (r = 0.25)

Full size image

3.5.2 Sensitivity analysis and publication bias

The (contour-enhanced) funnel plots of all the clusters appeared considerably symmetrical. Meta-analysis on bias-corrected pooled point estimates, showed findings similar to our analysis of the clusters. When excluding possible outliers, this again showed similar results to our cluster analysis. Analysis on selective reporting showed no signs of p-hacking. Overall, we did not find major indications of bias due to selective reporting or publication bias that changed the interpretation of our results. All the sensitivity analyses are presented in the Supplementary Materials (8. Outcomes meta-analysis – Sensitivity analysis).

3.5.3 Post factor power calculation

The post factor power calculation showed that with a mean effect size of SMCR = –0.016 the minimum number of studies needed ranged between 655 to 8053. Simulation input ranged from τ = 0.10, ω = 0.00, ρ = 0.20 to τ = 0.40, ω = 0.30, ρ = 0.70, where tau (τ) reflected heterogeneity, omega (ω) reflected within-study effect size range, and rho (ρ) indicated the correlation between effect size within studies. See Supplementary Materials (6. Power analysis Post factor R-script) for more details.

Discussion

In this study, we aimed to understand whether and how exercise-induced fatigue affects gait in community-dwelling older adults, and what specific elements of fatiguing protocols moderate the effects. Our meta-analyses indicated that exercise-induced fatigue in general affects gait parameters. However, these general effects could not be specified within any of the six clusters of the gait parameters that we identified (i.e., Dynamic Balance, Lower Limb Kinematics, Regularity, Spatio-temporal Parameters, Symmetry, and Velocity) as effects were not one-directional and hence non-distinguishable from zero. Furthermore, the (perceived) intensity, type of activity or duration of fatiguing exercises appeared not to be significant moderators of the general effects.

Overall effects were not reflected in clusters of gait parameters

As hypothesized, gait parameters change as a result of fatiguing activity. This is shown in the significant small to moderate changes on absolute data and is in line with existing literature. Nonetheless, analysis of absolute data makes it difficult to understand in what direction gait parameters are affected. Due to the large amount, and variety of gait parameters derived from all included studies in our meta-analyses, clustered gait parameters were needed to analyze the non-absolute data. Ideally, similar gait parameters would have been clustered, but the limited number of similar gait parameters found in the included studies made this not possible. As a result, different gait parameters were combined into clusters. Within the clusters, no significant changes in gait parameters were observed. These findings could potentially be explained by several factors. First, within clusters, there are conflicting or opposing parameters. We deemed that a pooled zero-effect as a result of (multiple) conflicting or opposing mechanisms may only appear in the Spatio-temporal Parameters cluster, where cadence was combined with single support time, stance time, step time, and swing time. So, in the case fatigue increases cadence, typically (one of) the five other spatio-temporal parameters will decrease. In the other clusters, less or no opposing effects of pooled parameters were expected. Second, between studies varying or opposite effects were reported. As shown in Table 3, only a minority of gait parameters had the same direction (all increase or decrease) post-fatiguing. The identified differences could be explained by the diverse ways to measure gait parameters. For example, in the analysis data from studies that used a treadmill was combined with studies that used overground walking to measure gait parameters. Although gait parameters on treadmill and overground are comparable, they are not always similar [90], and this combined analysis may have affected our findings. Third, besides conflicting parameters within clusters and between studies, an explanation could be the between-subject variation. The close-to-zero effects could also indicate that the pre- to post-fatiguing effects vary within studies. As shown in the summary of findings table the majority of findings are non-significant (Table 1). This could be explained by the possibility that the effects of fatigue on gait parameters, or the compensatory mechanisms that people show when fatigued [91], are not uniform. For example, a change in gait velocity could be a way to counter the feeling of instability [92]. In contrast to the review of Santos et al. [6], which showed an increase in gait velocity post fatigue, we found no significant changes in walking speed. Santos et al. [6] found this increase in speed with the four papers they included and mentioned that it could possibly be the result of a warm-up effect. In our meta-analysis, we used 26 effect sizes from 15 datasets and only 65% of the effect sizes indicated an increase in walking speed. This finding is in line with other research, indicating that participants potentially make changes in walking speed or other gait parameters to cope with (feelings of) fatigue or instability [93].

Potential moderators did not explain the (absence of) fatigue effects on clusters of gait parameters

In the moderator analysis of activity type, we distinguished walking activity from other activities. This was based on the assumption that walking is a rhythmic activity that affects multiple joints and muscles per cycle and therefore the fatiguing effect on gait may differ substantially from (isokinetic) single joint activities (i.e. knee flexion / extension). Moreover, walking activities are part of everyday life and relate close to the gait measures that we focused on in our study. Nonetheless, the type of activity appeared not to moderate the effects of fatigue in any of the clusters. Apparently, all types of activity (i.e., walking, and non-walking) have an effect on gait performance.

Furthermore, both (perceived) intensity and duration showed not to moderate the effects of fatiguing exercise on gait parameters. With respect to task dependent fatigue, taking only one moderator into account might not be sufficient to limit the variation in underlying mechanisms and sites associated with fatigue [24]. In daily living, activities show an interplay between the type of activity, the (perceived) intensity, and the duration of exposure, depending on the task requirements. Ideally, an interaction effect of moderators could give more insight into this ecological interplay and thereby possibly identify specific elements of fatiguing protocols. Unfortunately, due to the insufficient number of effect sizes, it was not possible to look at interaction effects between moderators in our meta-analysis.

Methodological aspects and limitations

The use of the algorithm-assisted ASReview resulted in a broader search, which probably resulted in including more papers than with a more conventional search strategy. In ASReview Lab, only 10% of the papers we found in our search were screened. Although a substantial number of papers was not screened by the authors, we are confident that all relevant papers are included. This latter was also underlined by simulations done in ASReview Lab. Furthermore, based on the algorithm, finding more relevant papers would not even out to the time spent [31]. In hindsight, the search could have been broadened even more. We now used the “NOT” operator to filter out based on title, but we could have trusted the software to filter for us. The use of the "NOT" operator to filter based on titles may have excluded studies where control group data could have been utilized. Nonetheless, including all these studies would possibly make the learning phase more complex, which could result in more studies that had to be assessed for full text but did not meet our inclusion criteria. Fortunately, contact with authors led to the inclusion of a control group from an article that had been filtered out based on the “NOT” operator. However, the extent of potentially missed control groups remains unclear. To analyze all the gait parameters in clusters with sufficient effect sizes, the clusters as introduced by Lindemann [37] were used. Although the work of Lindemann and other classifications is primarily focused on spatio-temporal gait parameters [37, 39, 94], it gave us tools to cluster all our gait parameters in six groups only. Although we tried to be as consistent as possible with the original clustering, we did include gait parameters that have not been clustered in the Lindemann’s framework before. Our clustering of the gait parameters thereby had a broader spectrum of the gait parameters per cluster than originally reported by Lindemann [37]. Another result of this clustering is that we could not compare our findings with literature on individual gait parameters other than velocity.

When comparing our a priori power analysis and post factor power analysis, two quite different numbers of studies that needed to be included for sufficient power were found. This is explained by two factors. First, prior to the start of this meta-analysis, there was no multilevel power calculation known by the authors. Second, in the a priori power calculation, an estimated effect size was used that was higher than the effect size that came out of the analysis. Therefore, it could be possible that our zero effect is a result of too little effect sizes that are included in our study.

An aspect that might have had an impact on our outcomes is the fact that we included eight articles (based on seven data sets) that reported on prolonged activity instead of a pre- vs. post-fatiguing exercise set up. In the studies with prolonged activity, fatigue was assumed as a consequence of activity and took data from the first minutes as pre-fatiguing measurement and from the last minutes as post-fatiguing measurement. As shown, gait speed is faster at the beginning and end of 400-m walk test and significantly lower in the middle, possibly because older adults are motivated to finish [95, 96]. The post-outcome measures might overestimate actual gait parameters, resulting in smaller differences between pre- and post-assessments.

An aspect influencing the calculated effect sizes, and consequently the pooled effect, was the need to impute correlations. This approach assigned the same correlation value across all pre- and post-measurements. While using actual correlations instead of imputed values would likely yield results closer to the true effects, the authors believe that such adjustments would not significantly alter our findings. The presence of heterogeneity influenced the precision of the true effect estimation. Our meta-analysis indicates that such variability in underlying studies (in terms of fatiguing protocols and outcome measures) limits the robustness of the findings. Thus, without more strict experimental control and without reducing the methodological differences across studies, the presence of heterogeneity would still be a serious problem for future meta-analytical attempts. Therefore, we feel that the GRADE certainty rating, which was (very) low in most cases, is a good reflection of the available papers in the field. Studies not only varied in the types of fatiguing exercises employed, but also in the methods used to measure the outcomes. As mentioned before, treadmill walking or overground walking could have different effects on gait parameters [90], but also the different overground walking distances could have had impact on gait parameters [97, 98].

Implications and recommendations

We now know that exercise-induced fatigue changes gait parameters in general (first hypothesis), but we could not pin-point these changes to specific directions, nor within clusters of gait parameters. Furthermore, no evidence was found to support our hypotheses that fatiguing exercises with a longer duration (second hypothesis), or a higher (perceived) intensity (third hypothesis), or that consisted of walking activities (fourth hypothesis) would show greater changes in gait parameters. This could be the result of the possible different strategies to cope with fatigue between participants, but is also a result of too many methodological inconsistencies in primary studies. Researchers in the field of gait analysis should report more similar gait parameters and measure these outcomes in more similar ways [98,99,100]. Similar recommendations for standardization of fatiguing protocols have been emphasized in reviews on the effects of exercise-induced fatigue addressing other target populations [101, 102]. More standardization in different aspects would make future systematic reviews and meta-analysis easier to conduct and could make such analysis more conclusive than the current work. Nonetheless, from current work we can conclude that it is important that older adults are aware of their changed physical (walking) abilities after fatiguing exercise in order to protect themselves from possible adverse effects. Although duration, type of activity and (perceived) intensity did not significantly moderate our findings, these moderators are building blocks when pre-scribing exercise [103]. Therefore, both researchers and therapists should know that regardless of the fatiguing exercise, older adults will show changes in walking performance. More practical, it seems that researchers can choose a fatiguing protocol that best suits their question, is possible in their lab settings, matches with their participants preferences or is the least invasive for participants, as long as they manage to fatigue participants. Therapists should be aware that the walking performance of older adults may be reduced or (un) intentionally improved, in relation to fall prevention, when fatigued. Thereby, underlining the need for tailor-made fall prevention, and taking sufficient rest after fatiguing exercises.

Research focusing on the effects of exercise-induced fatigue and gait parameters, should try to explain why exercise-induced fatigue could lead to non-uniform changes between participants. Do older adults actively choose a different movement behavior when fatigued? And do they make changes in gait parameters because they are aware of adverse fatiguing effects, are topics that are not well researched yet.

Conclusion

In walking, exercise-induced fatigue leads to small to moderate changes in gait parameters. These changes cannot be attributed to specific clusters of gait parameters. Furthermore, we could not identify specific elements of fatiguing exercise that lead to changes in gait parameters, as the type of activity, duration, or (perceived) intensity of the exercise did not moderate our findings. This may have a three-fold explanation: (1) older adults respond in both positive and negative ways to fatiguing exercise in their performance, (2) the type of gait parameters and how they are measured is too heterogeneous, or that (3) different moderators or interactions between moderators are necessary to explain these pooled close-to-zero effect sizes.

Data availability

The datasets generated and/or analyzed during the current study are available in the Figshare repository, https://doiorg.publicaciones.saludcastillayleon.es/10.21943/auas.25651671 and https://doiorg.publicaciones.saludcastillayleon.es/10.21943/auas.25652523.

The following materials can be found on our Figshare pages:

1. Search and inclusion related documents.

2. R-script and excel file for the interrater reliability.

3. Risk of Bias and GRADE documents.

4. Summary of Findings table.

5. Data file for the main analysis (coding form).

6. Analysis files (main, sensitivity, and power analysis).

7. Outcomes of the analyses with the different between measurement correlations.

8. Outcomes of the sensitivity analysis.

9. PRISMA checklist Abstract + Main.

Abbreviations

DB:

Dynamic balance

CI:

Confidence interval

CINAHL:

Cumulative index to nursing and allied health literature

df:

Degrees of freedom

ES:

Effect size

k:

The number of effect sizes

LB:

Lower bound

LLK:

Lower limb kinematics

MeSH:

Medical subject headings

n:

The number of datasets

NI:

No information

PI:

Prediction interval

PRISMA:

Preferred reporting items for systematic reviews and meta-analyses

PROSPERO:

International prospective register of systematic reviews

Reg:

Regularity

REML:

Restricted maximum-likelihood estimator

SE:

Standard error

SMCR:

Standardized mean change using raw score standardization

ST:

Spatio-temporal parameters

Sym:

Symmetry

UB:

Upper bound

Vel:

Velocity

References

  1. Enoka RM, Duchateau J. Translating fatigue to human performance. Med Sci Sports Exerc. 2016;48(11):2228–38. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC5035715/.

    PubMed  PubMed Central  Google Scholar 

  2. Senefeld J, Yoon T, Hunter SK. Age differences in dynamic fatigability and variability of arm and leg muscles: Associations with physical function. Exp Gerontol. 2017;1(87):74–83.

    Google Scholar 

  3. Barbieri FA, dos Santos PCR, Lirani-Silva E, Vitório R, Gobbi LTB, van Diëen JH. Systematic review of the effects of fatigue on spatiotemporal gait parameters. J Back Musculoskelet Rehabil. 2013;26(2):125–31. Available from: https://www.medra.org/servlet/aliasResolver?alias=iospressamp;doi=10.3233/BMR-130371.

  4. Ghamkhar L, Kahlaee AH. The effect of trunk muscle fatigue on postural control of upright stance: A systematic review. Gait Posture. 2019;1(72):167–74.

    Google Scholar 

  5. Helbostad JL, Sturnieks DL, Menant J, Delbaere K, Lord SR, Pijnappels M. Consequences of lower extremity and trunk muscle fatigue on balance and functional tasks in older people: A systematic literature review. BMC Geriatr. 2010;10(1):56. Available from: https://biomedcentral-bmcgeriatr.publicaciones.saludcastillayleon.es/articles/10.1186/1471-2318-10-56.

  6. Santos PCR dos, Barbieri FA, Zijdewind I, Gobbi LTB, Lamoth C, Hortobágyi T. Effects of experimentally induced fatigue on healthy older adults’ gait: A systematic review. PLoS One. 2019;14(12). Available from: https://pubmed.ncbi.nlm.nih.gov/31887182/.

  7. Chardon M, Barbieri FA, Penedo T, Santos PCR, Vuillerme N. The effects of experimentally-induced fatigue on gait parameters during obstacle crossing: A systematic review. Neurosci Biobehav Rev [Internet]. 2022;142(August):104854. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.neubiorev.2022.104854.

  8. Nagano H, James L, Sparrow WA, Begg RK. Effects of walking-induced fatigue on gait function and tripping risks in older adults. J Neuroeng Rehabil [Internet]. 2014 Nov 15;11(1):155. Available from: https://pubmed.ncbi.nlm.nih.gov/25399324/

  9. Pana A, Sourtzi P, Kalokairinou A, Pastroudis A, Chatzopoulos ST, Velonaki VS. Association between self-reported or perceived fatigue and falls among older people: A systematic review. Int J Orthop Trauma Nurs. 2021;1(43): 100867.

    Google Scholar 

  10. Barbieri FA, Lee YJ, Gobbi LTB, Pijnappels M, Van Dieën JH. The effect of muscle fatigue on the last stride before stepping down a curb. Gait Posture. 2013;37(4):542–6.

    PubMed  Google Scholar 

  11. Bahmanbegloo ZH, Budini F, Hassanlouei H, Farsi A, Tilp M. Submaximal fatiguing contractions reduce stability of voluntary postural control more than maximal fatiguing contractions. Gait Posture [Internet]. 2022;92(December 2021):407–12. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.gaitpost.2021.12.004

  12. Barbieri FA, Gobbi LTB, Lee YJ, Pijnappels M, van Dieën JH. Effect of triceps surae and quadriceps muscle fatigue on the mechanics of landing in stepping down in ongoing gait. Ergonomics. 2014;57(6):934–42.

    CAS  PubMed  Google Scholar 

  13. Hatton AL, Menant JC, Lord SR, Lo JCM, Sturnieks DL. The effect of lower limb muscle fatigue on obstacle negotiation during walking in older adults. Gait Posture [Internet]. 2013 Apr;37(4):506–10. Available from: https://pubmed.ncbi.nlm.nih.gov/23021990/

  14. Santos PCR dos, Lamoth CJC, Gobbi LTB, Zijdewind I, Barbieri FA, Hortobágyi T. Older Compared With Younger Adults Performed 467 Fewer Sit-to-Stand Trials, Accompanied by Small Changes in Muscle Activation and Voluntary Force. Front Aging Neurosci [Internet]. 2021 Jun 21;13. Available from: https://pubmed.ncbi.nlm.nih.gov/34267644/

  15. Donath L, Zahner L, Roth R, Fricker L, Cordes M, Hanssen H, et al. Balance and gait performance after maximal and submaximal endurance exercise in seniors: Is there a higher fall-risk? Eur J Appl Physiol. 2013;113(3):661–9.

    PubMed  Google Scholar 

  16. Smith JL, Martin PG, Gandevia SC, Taylor JL. Sustained contraction at very low forces produces prominent supraspinal fatigue in human elbow flexor muscles. J Appl Physiol [Internet]. 2007;103(2):560–8. Available from: https://www.physiology.org/doi/10.1152/japplphysiol.00220.2007

  17. Boyas S, Remaud A, Rivers E, Bilodeau M. Fatiguing Exercise Intensity Influences the Relationship between Parameters Reflecting Neuromuscular Function and Postural Control Variables. PLoS ONE. 2013;8(8):1–10.

    Google Scholar 

  18. Azevedo R de A, Silva-Cavalcante MD, Lima-Silva AE, Bertuzzi R. Fatigue development and perceived response during self-paced endurance exercise: state-of-the-art review. Eur J Appl Physiol [Internet]. 2021 Mar 2;121(3):687–96. Available from: http://link.springer.com/http://link.springer.com/10.1007/s00421-020-04549-5

  19. Thomas K, Elmeua M, Howatson G, Goodall S. Intensity-Dependent Contribution of Neuromuscular Fatigue after Constant-Load Cycling. Med Sci Sports Exerc. 2016;48(9):1751–60.

    PubMed  Google Scholar 

  20. Gandevia SC. Spinal and Supraspinal Factors in Human Muscle Fatigue. Physiol Rev [Internet]. 2001 Jan 10;81(4):1725–89. Available from: https://www.physiology.org/doi/10.1152/physrev.2001.81.4.1725

  21. Burgerhout W. Visies op vermoeidheid Deel 3: Perifere factoren - vermoeidheid op spierniveau. Sportgericht. 2018;72(1):12–7.

    Google Scholar 

  22. Hunter SK. Performance fatigability: Mechanisms and task specificity. Cold Spring Harb Perspect Med [Internet]. 2018;8(7). Available from: https://pubmed.ncbi.nlm.nih.gov/28507192/

  23. Taylor JL, Amann M, Duchateau J, Meeusen R, Rice CL. Neural contributions to muscle fatigue: From the brain to the muscle and back again. Med Sci Sports Exerc. 2016;48(11):2294–306.

    CAS  PubMed  PubMed Central  Google Scholar 

  24. Enoka RM, Stuart DG. Neurobiology of muscle fatigue. J Appl Physiol. 1992;72(5):1631–48.

    CAS  PubMed  Google Scholar 

  25. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: An updated guideline for reporting systematic reviews, vol. 372. The BMJ: BMJ Publishing Group; 2021.

    Google Scholar 

  26. Higgins J, Thomas J, Chandler J, Cumpston M, Li T, Page M, et al. Cochrane Handbook for Systematic Reviews of Interventions Version 6.3. The Cochrane Collaboration. 2022. Available from: www.training.cochrane.org/handbook

  27. Moreau D, Gamble B. Conducting a Meta-Analysis in the Age of Open Science: Tools, Tips, and Practical Recommendations. Psychol Methods. 2022;27(3):426–32.

    PubMed  Google Scholar 

  28. The EndNote Team. EndNote. Philadelphia, PA: Clarivate; 2013.

  29. ASReview LAB developers. ASReview LAB - A tool for AI-assisted systematic reviews (v1.1). Zenodo; 2022. https://doiorg.publicaciones.saludcastillayleon.es/10.5281/zenodo.7319063.

  30. ASReview. ASReview. 2023 [cited 2023 Sep 22]. Available from: https://asreview.nl/blog/active-learning-explained/

  31. Boetje J, van de Schoot R. The SAFE Procedure : A Practical Stopping Heuristic for Active Learning-Based Screening in Systematic Reviews and Meta-Analyses. PsycArxiv Prepr. 2023;1–28. https://doiorg.publicaciones.saludcastillayleon.es/10.21203/rs.3.rs-2856011/v1.

  32. ASReview LAB developers. ASReview LAB Software Documentation (v1.3). Zenodo; 2023. https://doiorg.publicaciones.saludcastillayleon.es/10.5281/zenodo.10066693.

  33. Ouzzani M, Hammady H, Fedorowicz Z, Elmagarmid A. Rayyan-a web and mobile app for systematic reviews. Syst Rev [Internet]. 2016;5(1):1–10. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13643-016-0384-4

  34. Sterne JA, Hernán MA, Reeves BC, Savović J, Berkman ND, Viswanathan M, et al. ROBINS-I: A tool for assessing risk of bias in non-randomised studies of interventions. BMJ. 2016;355:4–10.

    Google Scholar 

  35. Schünemann HJ, Cuello C, Akl EA, Mustafa RA, Meerpohl JJ, Thayer K, Morgan RL, Gartlehner G, Kunz R, Katikireddi SV, Sterne J, Higgins JP, Guyatt G, GRADE Working Group. GRADE guidelines: 18. How ROBINS-I and other tools to assess risk of bias in nonrandomized studies should be used to rate the certainty of a body of evidence. J Clin Epidemiol. 2019;111:105–14. https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.jclinepi.2018.01.012. Epub 2018 Feb 9.

  36. Schünemann HJ, Brożek J, Guyatt G, Oxman A. GRADE handbook for grading quality of evidence and strength of recommendations. [Internet]. October 20. The GRADE Working Group; 2013. Available from: https://guidelinedevelopment.org/handbook.

  37. Lindemann U. Spatiotemporal gait analysis of older persons in clinical practice and research. Z Gerontol Geriatr. 2020;53(2):171–8. Available from: https://link-springer-com.vu-nl.idm.oclc.org/article/10.1007/s00391-019-01520-8.

  38. Dapp U, Vinyard D, Golgert S, Krumpoch S, Freiberger E. Reference values of gait characteristics in community-dwelling older persons with different physical functional levels. BMC Geriatr [Internet]. 2022;22(1):713. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s12877-022-03373-0

  39. Hollman JH, McDade EM, Petersen RC. Normative spatiotemporal gait parameters in older adults. Gait Posture. 2011;34(1):111–8. Available from: https://linkinghub.elsevier.com/retrieve/pii/S0966636211001019

  40. Cuijpers P, Weitz E, Cristea IA, Twisk J. Pre-post effect sizes should be avoided in meta-analyses. Epidemiol Psychiatr Sci. 2017;26(4):364–8.

    CAS  PubMed  Google Scholar 

  41. Viechtbauer W. Bias and efficiency of meta-analytic variance estimators in the random-effects model. J Educ Behav Stat. 2005;30(3):261–93.

    Google Scholar 

  42. Cochran WG. The Combination of Estimates from Different Experiments. Biometrics [Internet]. 1954;10(1):101. Available from: https://www.jstor.org/stable/3001666?origin=crossref

  43. Higgins JPT, Thompson SG. Quantifying heterogeneity in a meta‐analysis. Stat Med [Internet]. 2002;21(11):1539–58. Available from: https://onlinelibrary.wiley.com/doi/10.1002/sim.1186.

  44. R Core Team. A Language and Environment for Statistical Computing [Internet]. R Foundation for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing; 2020. p. https://www.R-project.org. Available from: http://www.r-project.org.

  45. RStudio Team. RStudio: Integrated Development for R [Internet]. Boston, MA: RStudio, PBC; 2020. Available from: http://www.rstudio.com/.

  46. Viechtbauer W. Conducting meta-analyses in R with the metafor. J Stat Softw. 2010;36(3):1–48.

    Google Scholar 

  47. Pustejovsky JE. clubSandwich: Cluster-Robust (Sandwich) Variance Estimators with Small-Sample Corrections [Internet]. 2023. Available from: https://cran.r-project.org/package=clubSandwich.

  48. Harrer M, Cuijpers P, Furukawa T, Ebert DD. dmetar: Companion R package for the guide “Doing Meta-Analysis in R” [Internet]. 2019. Available from: https://dmetar.protectlab.org/authors#citation

  49. Wickham H, Averick M, Bryan J, Chang W, McGowan L, François R, et al. Welcome to the Tidyverse. J Open Source Softw. 2019;4(43):1686.

    Google Scholar 

  50. Auguie B. gridExtra: Miscellaneous Functions for “Grid” Graphics [Internet]. 2017. Available from: https://cran.r-project.org/package=gridExtra

  51. Nakagawa S, Lagisz M, O’Dea RE, Pottier P, Rutkowska J, Senior AM, et al. orchaRd 2.0: An R package for visualizing meta-analyses with orchard plots. EcoEvoRxiv. 2023;12:4–12.

  52. Beath KJ. Metaplus: An R-Package for the Analysis of Robust Meta-Analysis and Meta-Regression. R J [Internet]. 2016;8(1):5–16. Available from: https://journal.r-project.org/archive/2016-1/beath.pdf

  53. Braginsky M, Mathur M, VanderWeele TJ, Solymos P. PublicationBias: Sensitivity Analysis for Publication Bias in Meta-Analyses [Internet]. 2023. Available from: https://github.com/mathurlabstanford/PublicationBias

  54. Mathur MB. P-hacking in meta-analyses: A formalization and new meta-analytic methods. 2022; Available from: https://osf.io/ezjsx/

  55. Mathur M, Braginsky M. phacking: Sensitivity Analysis for p-Hacking in Meta-Analyses [Internet]. 2023. Available from: https://github.com/mathurlabstanford/phacking

  56. Mathur M, Braginsky M. multibiasmeta: Sensitivity Analysis for Multiple Biases in Meta-Analyses [Internet]. 2023. Available from: https://github.com/mathurlabstanford/multibiasmeta

  57. Vembye MH, Pustejovsky JE. POMADE: Power for Meta-Analysis of Dependent Effects [Internet]. 2022. Available from: https://cran.r-project.org/package=POMADE

  58. Haddaway NR, Page MJ, Pritchard CC, McGuinness LA, PRISMA2020: An R package and Shiny app for producing PRISMA 2020-compliant flow diagrams, with interactivity for optimised digital transparency and Open Synthesis. Campbell Syst Rev. 2022;8(2):e1230. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1002/cl2.1230.

  59. McGuinness LA, Higgins JPT. Risk-of-bias VISualization (robvis): An R package and Shiny web app for visualizing risk-of-bias assessments. Res Syn Meth. 2021;12:55–61. https://doiorg.publicaciones.saludcastillayleon.es/10.1002/jrsm.1411.

    Google Scholar 

  60. Arvin M, Hoozemans MJM, Burger BJ, Rispens SM, Verschueren SMP, van Dieën JH, et al. Effects of hip abductor muscle fatigue on gait control and hip position sense in healthy older adults. Gait Posture. 2015;42(4):545–9. Available from: https://pubmed.ncbi.nlm.nih.gov/26386676/

  61. Baptista RR, Aires AG, Giacomazzi LP, Vilar TB, Tarouco IL. Is there an effect of stair negotiation on subsequent gait kinetics in elderly women? Res Biomed Eng. 2020;36(1):31–7. Available from: http://link.springer.com/10.1007/s42600-019-00030-1.

  62. Barbieri FA, dos Santos PCR, Simieli L, Orcioli-Silva D, van Dieën JH, Gobbi LTB. Interactions of age and leg muscle fatigue on unobstructed walking and obstacle crossing. Gait Posture. 2014;39(3):985–90.

    PubMed  Google Scholar 

  63. Chen S-H, Chou L-S. Gait balance control after fatigue: Effects of age and cognitive demand. Gait Posture. 2022;95:129–34.

    PubMed  Google Scholar 

  64. Da Rocha ES, Kunzler MR, Bobbert MF, Duysens J, Carpes FP. 30 min of treadmill walking at self-selected speed does not increase gait variability in independent elderly. J Sports Sci. 2018;36(11):1305–11. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1080/02640414.2017.1375139

  65. Drum SN, Faude O, de Fay du Lavallaz E, Allemann R, Nève G, Donath L. Acute effects of walking at moderate normobaric hypoxia on gait and balance performance in healthy community-dwelling seniors: A randomized controlled crossover study. Arch Gerontol Geriatr. 2016;67:74–9.

  66. Egerton T, Brauer SG, Cresswell AG. Fatigue after physical activity in healthy and balance-impaired elderly. J Aging Phys Act. 2009;17(1):89–105.

    PubMed  Google Scholar 

  67. Elhadi MMO, Ma CZ, Wong DWCC, Wan AHPP, Lee WCCC. Comprehensive Gait Analysis of Healthy Older Adults Who Have Undergone Long-Distance Walking. J Aging Phys Act. 2017;25(3):367–77. Available from: https://journals.humankinetics.com/view/journals/japa/25/3/article-p367.xml

  68. Elhadi MMO, Ma CZ-H, Lam WK, Lee WC-C. Biomechanical approach in facilitating long-distance walking of elderly people using footwear modifications. Gait Posture [Internet]. 2018;64(May):101–7. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.gaitpost.2018.05.032

  69. Granacher U, Wolf I, Wehrle A, Bridenbaugh S, Kressig RW. Effects of muscle fatigue on gait characteristics under single and dual-task conditions in young and older adults. J Neuroeng Rehabil. 2010;7(1). Available from: https://pubmed.ncbi.nlm.nih.gov/21062458/

  70. Hamacher DD, Törpel A, Hamacher DD, Schega L. The effect of physical exhaustion on gait stability in young and older individuals. Gait Posture [Internet]. 2016 Jul 1;48:137–9. Available from: https://pubmed.ncbi.nlm.nih.gov/27239774/

  71. Helbostad JL, Leirfall S, Moe-Nilssen R, Sletvold O. Physical Fatigue Affects Gait Characteristics in Older Persons. Journals Gerontol Ser A [Internet]. 2007;62(9):1010–5. Available from: https://academic.oup.com/biomedgerontology/article/62/9/1010/525884

  72. Hurt CP, Rosenblatt N, Crenshaw JR, Grabiner MD. Variation in trunk kinematics influences variation in step width during treadmill walking by older and younger adults. Gait Posture. 2010;31(4):461–4.

    PubMed  Google Scholar 

  73. Ko SU, Hausdorff JM, Ferrucci L. Age-associated differences in the gait pattern changes of older adults during fast-speed and fatigue conditions: Results from the Baltimore longitudinal study of ageing. Age Ageing. 2010;39(6):688–94. Available from: https://pubmed.ncbi.nlm.nih.gov/20833863/

  74. Kushioka J, Sun R, Zhang W, Muaremi A, Leutheuser H, Odonkor CA, Smuck M. Gait Variability to Phenotype Common Orthopedic Gait Impairments Using Wearable Sensors. Sensors. 2022;22(23):9301. https://doiorg.publicaciones.saludcastillayleon.es/10.3390/s22239301.

    PubMed  PubMed Central  Google Scholar 

  75. Morrison S, Colberg SR, Parson HK, Neumann S, Handel R, Vinik EJ, et al. Walking-Induced Fatigue Leads to Increased Falls Risk in Older Adults. J Am Med Dir Assoc. 2016;17(5):402–9.

    PubMed  PubMed Central  Google Scholar 

  76. Odonkor C, Kuwabara A, Tomkins-Lane C, Zhang W, Muaremi A, Leutheuser H, et al. Gait features for discriminating between mobility-limiting musculoskeletal disorders: Lumbar spinal stenosis and knee osteoarthritis. Gait Posture. 2020;80:96–100.

    PubMed  Google Scholar 

  77. Oliveira CF, Vieira ER, Machado Sousa FM, Vilas-Boas JP. Kinematic Changes during Prolonged Fast-Walking in Old and Young Adults. Front Med. 2017;4:207.

    Google Scholar 

  78. Pereira M, Gonçalves M. Effects of Fatigue Induced by Prolonged Gait When Walking on the Elderly. Hum Mov [Internet]. 2011 Jan 1;12(3):242–7. Available from: http://www.degruyter.com/view/j/humo.2011.12.issue-3/v10038-011-0025-7/v10038-011-0025-7.xml

  79. Piche E, Gerus P, Zory R, Jaafar A, Guerin O, Chorin F. Effects of muscular and mental fatigue on spatiotemporal gait parameters in dual task walking in young, non-frail and frail older adults. Aging Clin Exp Res. 2023;(0123456789). Available from: http://www.ncbi.nlm.nih.gov/pubmed/37535312

  80. Qiao Y (Susanna), Harezlak J, Moored KD, Urbanek JK, Boudreau RM, Toto PE, et al. Development of a Novel Accelerometry-Based Performance Fatigability Measure for Older Adults. Med Sci Sport Exerc. 2022;54(10):1782–93. Available from: https://journals.lww.com/10.1249/MSS.0000000000002966

  81. Santos PCR, Gobbi LTB, Orcioli-Silva D, Simieli L, van Dieën JH, Barbieri FA. Effects of leg muscle fatigue on gait in patients with Parkinson’s disease and controls with high and low levels of daily physical activity. Gait Posture. 2016;47:86–91. Available from: https://linkinghub.elsevier.com/retrieve/pii/S096663621630008X

  82. Santos PCR dos, Hortobágyi T, Zijdewind I, Bucken Gobbi LT, Barbieri FA, Lamoth C. Minimal effects of age and prolonged physical and mental exercise on healthy adults’ gait. Gait Posture [Internet]. 2019;74(June):205–11. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.gaitpost.2019.09.017

  83. Santos PCR Dos, Lamoth CJC, Barbieri FA, Zijdewind I, Gobbi LTB, Hortobágyi T, et al. Age-specific modulation of intermuscular beta coherence during gait before and after experimentally induced fatigue. Sci Rep [Internet]. 2020 Dec 1;10(1):1–12. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1038/s41598-020-72839-1

  84. Toebes MJPP, Hoozemans MJMM, Dekker J, van Dieën JH. Effects of unilateral leg muscle fatigue on balance control in perturbed and unperturbed gait in healthy elderly. Gait Posture. 2014;40(1):215–9.

    PubMed  Google Scholar 

  85. Vreede K, Henriksson J, Borg K, Henriksson M. Gait characteristics and influence of fatigue during the 6-minute walk test in patients with post-polio syndrome. J Rehabil Med. 2013;45(9):924–8. Available from: http://www.medicaljournals.se/jrm/content/?doi=10.2340/16501977-1209.

  86. Watanabe K. Relationship Between Toe Clearance Strategy and Regional Regulation of Rectus Femoris Muscle During Swing Phase in Prolonged Walking in Young and Older Adults. Front Physiol [Internet]. 2018;9(SEP):1–8. Available from: https://www.frontiersin.org/article/10.3389/fphys.2018.01274/full.

  87. Wong DWC, Lam WK, Lee WCC. Gait asymmetry and variability in older adults during long-distance walking: Implications for gait instability. Clin Biomech. 2020;72:37–43. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.clinbiomech.2019.11.023

  88. Zhang G, Chen TLW, Wang Y, Tan Q, Hong TTH, Peng Y, et al. Effects of prolonged brisk walking induced lower limb muscle fatigue on the changes of gait parameters in older adults. Gait Posture. 2023;101(January):145–53. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.gaitpost.2023.02.010

  89. Borenstein M. In a meta-analysis, the I-squared statistic does not tell us how much the effect size varies. J Clin Epidemiol [Internet]. 2022;152:281–4. Available from: https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.jclinepi.2022.10.003

  90. Song JL, Hidler J. Biomechanics of overground vs. treadmill walking in healthy individuals. J Appl Physiol. 2008;104(3):747–55.

    Google Scholar 

  91. Monjo F, Terrier R, Forestier N. Muscle fatigue as an investigative tool in motor control: A review with new insights on internal models and posture-movement coordination. Hum Mov Sci. 2015;44:225–33. Available from: https://linkinghub.elsevier.com/retrieve/pii/S0167945715300397

  92. England SA, Granata KP. The influence of gait speed on local dynamic stability of walking. Gait Posture. 2007;25(2):172–8.

    PubMed  Google Scholar 

  93. Hak L, Houdijk H, Steenbrink F, Mert A, van der Wurff P, Beek PJ, et al. Stepping strategies for regulating gait adaptability and stability. J Biomech. 2013;46(5):905–11.

    PubMed  Google Scholar 

  94. Lord S, Galna B, Verghese J, Coleman S, Burn D, Rochester L. Independent Domains of Gait in Older Adults and Associated Motor and Nonmotor Attributes: Validation of a Factor Analysis Approach. Journals Gerontol Ser A Biol Sci Med Sci [Internet]. 2013 Jul 1;68(7):820–7. Available from:https://academic.oup.com/biomedgerontology/article-lookup/doi/10.1093/gerona/gls255.

  95. Lindemann U, Krumpoch S, Becker C, Sieber CC, Freiberger E. The course of gait speed during a 400m walk test of mobility limitations in community-dwelling older adults. Z Gerontol Geriatr. 2021;54(8):768–74.

    PubMed  PubMed Central  Google Scholar 

  96. Glynn NW, Qiao Y (Susanna), Simonsick EM, Schrack JA. Response to “Comment on: Fatigability: A Prognostic Indicator of Phenotypic Aging.” Journals Gerontol Ser A. 2021;76(8):e161–2. Available from: https://academic.oup.com/biomedgerontology/article/76/8/e161/6154281.

  97. Pasma JH, Stijntjes M, Ou SS, Blauw GJ, Meskers CGM, Maier AB. Walking speed in elderly outpatients depends on the assessment method. Age (Omaha). 2014;36(6).

  98. Krumpoch S, Lindemann U, Rappl A, Becker C, Sieber CC, Freiberger E. The effect of different test protocols and walking distances on gait speed in older persons. Aging Clin Exp Res [Internet]. 2021;33(1):141–6. Available from: https://link.springer.com/10.1007/s40520-020-01703-z.

  99. Beauchet O, Allali G, Sekhon H, Verghese J, Guilain S, Steinmetz J-P, et al. Guidelines for Assessment of Gait and Reference Values for Spatiotemporal Gait Parameters in Older Adults: The Biomathics and Canadian Gait Consortiums Initiative. Front Hum Neurosci [Internet]. 2017 Aug 3;11(August). Available from: http://journal.frontiersin.org/article/10.3389/fnhum.2017.00353/full.

  100. Kressig RW, Beauchet O, Gaitrite E, Group N. Guidelines for clinical applications of spatio-temporal. Aging Clin Exp Res. 2005;18(2):174–6.

    Google Scholar 

  101. Marotta L, Scheltinga BL, van Middelaar R, Bramer WM, van Beijnum BJF, Reenalda J, et al. Accelerometer-Based Identification of Fatigue in the Lower Limbs during Cyclical Physical Exercise: A Systematic Review. Sensors. 2022;22(8):1–36.

    Google Scholar 

  102. Heil J, Loffing F, Büsch D. The Influence of Exercise-Induced Fatigue on Inter-Limb Asymmetries: a Systematic Review. Sports Med Open. 2020;6:39. https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s40798-020-00270-x.

    PubMed  PubMed Central  Google Scholar 

  103. Riebe D, Ehrman JK, Liguori G, Magal M, editors. ACSM’s guidelines for exercise testing and prescription. 10th ed. Wolters Kluwer; 2018.

Download references

Acknowledgements

We like to thank all authors that responded to our questions or requests for data. In particular we would like to thank Mina Arvin, Emmanuel Da Rocha, Jorunn Helbostad, Marco Hoozemans, Christopher Hurt, Paolo Cezar Rocha dos Santos, Ruopeng (Robin) Sun, Elodie Piche, Yujia (Susanna) Qiao and Kohei Watanabe. We also would like to thank Esther Verloop for her help setting up the search strategy and Anne de Jong for his help with data management.

Funding

This research was funded by the Dutch Research Council (NWO), file number 023.017.127.

Author information

Authors and Affiliations

  1. Faculty of Health, Sport and Physical Activity, Centre of Expertise Urban Vitality, Amsterdam University of Applied Sciences, Amsterdam, Netherlands

    Paul Benjamin Voorn, Daniël Bossen & Bart Visser

  2. Department of Human Movement Sciences, Faculty of Behavioural and Movement Sciences, Amsterdam Movement Sciences Research Institute, Vrije Universiteit Amsterdam, Amsterdam, Netherlands

    Paul Benjamin Voorn, Remco Oomen, Bart Visser & Mirjam Pijnappels

  3. Department of Experimental and Applied Psychology, Faculty of Behavioural and Movement Sciences, Vrije Universiteit Amsterdam, Amsterdam, Netherlands

    Jacek Buczny

Authors
  1. Paul Benjamin Voorn
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Remco Oomen
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Jacek Buczny
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Daniël Bossen
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Bart Visser
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Mirjam Pijnappels
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

Author Contributions: Conceptualization and design of the study: PV, DB, BV, MP. Search, data extraction and assessment of data: PV, RO, JB, DB, BV. Data analysis: PV, RO, JB, DB, BV, MP. Drafting the manuscript: PV, JB, DB, BV, MP. All authors reviewed and edited the manuscript and approved the final manuscript.

Corresponding authors

Correspondence to Paul Benjamin Voorn or Mirjam Pijnappels.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Voorn, P.B., Oomen, R., Buczny, J. et al. The effect of exercise-induced muscle fatigue on gait parameters among older adults: a systematic review and meta-analysis. Eur Rev Aging Phys Act 22, 4 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s11556-025-00370-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s11556-025-00370-1

Keywords

European Review of Aging and Physical Activity

ISSN: 1861-6909

Contact us