Factors associated with walking adaptability and its association with falling in polio survivors

Open AccessPublished:May 26, 2022DOI:


      • Walking adaptability was assessed with target-stepping and obstacle-avoidance tests
      • Target stepping was determined by leg-muscle strength and balance confidence
      • Obstacle avoidance was solely determined by leg-muscle strength
      • Target stepping was related to falling, while obstacle avoidance not.
      • Foot-placement limitations may increase the fall risk among polio survivors
      Objective To explore factors associated with walking adaptability and associations between walking adaptability and falling in polio survivors.
      Design Cross-sectional study.
      Setting Outpatient expert polio clinic.
      Participants Polio survivors (n=46) who fell in the previous year and/or reported fear of falling.
      Interventions Not applicable.
      Main outcome measures Walking adaptability was assessed on an interactive treadmill and operationalized as variable target-stepping and reactive obstacle-avoidance performance. Further, we collected walking speed and assessed leg muscle strength, balance performance (Berg Balance Scale and Timed-Up-and-Go Test), balance confidence (Activities-specific Balance Confidence scale), ambulation level, orthosis use, fear of falling and number of falls in the previous year.
      Results With walking speed included as covariate, muscle weakness of the most affected leg and balance confidence explained 54% of the variance in variable target-stepping performance. For reactive obstacle-avoidance performance, muscle weakness of the most affected leg and knee extensor strength of the least affected leg explained 32% of the variance. Only target-stepping performance was significantly related to the number of falls reported in the previous year (R2=0.277, p<0.001), and mediated the relation between leg muscle weakness and balance confidence with falling.
      Conclusion Our exploratory study suggests that leg muscle weakness and reduced balance confidence limit walking adaptability in polio survivors. As poorer target-stepping rather than obstacle-avoidance performance was associated with falling, our results indicate that a limited ability to ensure safe foot placement may be a fall risk factor in this group. These findings should be confirmed in a larger sample.

      Key words

      List of abbreviations

      ABC Activities-specific Balance Confidence
      BBS Berg Balance Scale
      CWS Comfortable walking speed
      CoP Centre of Pressure
      FES Falls Efficacy Scale
      FoF Fear of falling
      MRC Medical Research Council
      TUG Timed-Up-and-Go Test
      VE Variable stepping Error
      Aging survivors of polio have a fall rate 2-3 times higher [1-4] than comparable community-dwelling healthy individuals [5], and many falls result in (severe) injuries that impact daily activities, independent living and quality of life [2, 6, 7]. Falls mostly occur during walking [2], an activity of daily life that requires continuous adaptation to task goals and environmental circumstances (walking adaptability). Walking adaptability can be conceptualized into nine domains [8], including obstacle negotiation, walking on uneven surfaces and manoeuvring during walking [8, 9]. Using the C-Mill interactive treadmill (Figure 1), we previously found that polio survivors show poorer walking adaptability as expressed in reduced obstacle-avoidance performance (mimicking obstacle negotiation) and less precise target stepping (mimicking walking on uneven surfaces) when compared to healthy individuals [10].
      Figure 1
      Figure 1The C-Mill interactive treadmill
      The C-Mill interactive treadmill allows safe assessment of walking adaptability by determining variable target-stepping (lower left picture) or reactive obstacle-avoidance performance (lower right picture).
      Together with the ability to generate step movements and maintain balance, walking adaptability is an essential aspect of safe and independent walking [8, 9]. Early identification of limitations in walking adaptability is therefore important, but we first need a better understanding of factors associated with walking adaptability in polio survivors. In neurological disorders such as Parkinson's disease or stroke, balance problems, weakness of the hip abductor and quadriceps muscles, previous falls and impaired executive function are associated with reduced walking adaptability [11, 12]. However, factors related to walking adaptability in the abovementioned groups may not be directly transferable to a neuromuscular disorder such as polio.
      Most falls in polio survivors occur after tripping, slipping or missteps [2, 6], suggesting an association between reduced walking adaptability and falling. Although not confirmed in polio survivors, a similar association has been found in community-dwelling elderly [13] and in people with Parkinson's disease [14]. Previously identified factors associated with risk of falls in polio survivors include intrinsic problems such as leg muscle weakness [15] (specifically of the most affected leg [2, 4, 16]), fear of falling [2, 16, 17], self-reported balance problems [2] and reduced walking speed [16]. Nevertheless, extrinsic factors such as environmental circumstances might also be involved [18] but have yet to be studied.
      Our goals in the present study were therefore to 1) explore factors associated with walking adaptability (target-stepping performance and obstacle-avoidance performance) in polio survivors, and 2) explore associations between walking adaptability and falling.


       Study design

      Data for this cross-sectional study originate from a 2-year prospective cohort study on walking adaptability in polio survivors, with measurements taken at baseline, two weeks, 1 year and two years. Here we used data collected at the two-week point to account for learning effects [19]. The study protocol was previously approved by the medical ethics committee of the Academic Medical Center (AMC, Amsterdam, the Netherlands). Study reporting was in accordance with the ‘Strengthening Reporting of Observational Studies in Epidemiology’ recommendations [20].


      Polio survivors (n=48) were recruited from the outpatient clinic of the department of Rehabilitation Medicine of the Amsterdam UMC, location AMC, from December 2016 to August 2018. Inclusion criteria were: aged between 18 and 80 years, able to walk indoors without assistive devices (e.g. crutch, cane or walker), and ≥ 1 fall reported in the previous year and/or fear of falling. A diagnosis of any other medical condition that increases fall risk was an exclusion criterion. Participants provided written informed consent prior to study enrolment.


      Data used in this study included walking-adaptability outcomes, leg muscle strength, balance performance, balance confidence, walking ability, fear of falling and fall frequency.

       Walking adaptability

      Participants performed walking-adaptability tests on the C-Mill interactive treadmill a as described previously [10]. After a familiarization protocol (4-5 minutes) and determination of comfortable treadmill walking speed (CWS, in m/s), participants first performed three target-stepping tasks, followed by two obstacle-avoidance tasks. For both tasks, the most challenging conditions were used in the current study because of superior reproducibility compared to the easier conditions [19]. All tests were performed at fixed CWS and a safety harness without weight bearing was used to prevent falling. Participants used their customary orthosis during testing.
      Variable target stepping was assessed while following a sequence of rectangular step targets, projected with 30% inter-target variation relative to the participants’ own gait pattern, for two minutes. Participants were asked to centre their feet in the step targets, which were 3 cm larger than their right foot. Target-stepping performance was calculated over goal-directed steps, i.e. with the midstance CoP located within the target boundaries, defined as the Variable stepping Error (VE, in mm). The VE was calculated as the standard deviation over the distances from the middle of the step target to the midstance CoP using MATLAB 2019ab. A smaller VE represents better target-stepping performance. For reactive obstacle avoidance, 12 obstacles (30 cm long and as wide as the treadmill belt) were projected onto the treadmill belt at the anticipated foot contact location. Successful avoidance was scored when both feet were positioned outside the obstacle boundaries without handrail use. Obstacle-avoidance success rate (%) was calculated when at least 10 obstacles were projected onto the treadmill, and higher percentages represent better performance.

       Leg muscle strength

      Muscle strength was assessed manually according to the Medical Research Council (MRC) scale (range 0-5) [21] for eight muscle groups per leg. We calculated left and right MRC sum-scores (range 0-40, note: in case of an ankle arthrodesis, the dorsal and plantar MRC of that side were set at zero [22]) and we defined a most and least affected leg (i.e. lowest and highest MRC sum-score, respectively). We calculated MRC asymmetry-scores by taking the absolute value of the difference in left and right MRC sum-scores [22].
      For the left and right knee extensors, we measured isometric strength on a fixed dynamometerc. Participants were seated with the back of the chair positioned in 85° and with the knee in 60° flexion. Maximal strength was defined as the peak torque (in Nm) of three maximal voluntary isometric contractions of 5 seconds, interspersed with periods of 30 seconds rest.

       Balance performance and balance confidence

      Balance performance was assessed with the Berg Balance Scale (BBS) and the Timed-Up-and-Go test (TUG). The BBS-score (range 0 to 56, higher scores indicate better balance performance) is summed from 14 activity items (range 0 to 4), indicating whether a participant is able to perform the activity (independently) [23]. The TUG-score represents the time (in seconds) needed to rise from a chair, walk three meters, turn around and sit back in the chair [24, 25]. A lower TUG-score indicates better balance performance. Balance confidence was measured with the Activities-specific Balance Confidence (ABC) Scale, summed from 16 items (range 0-100%) [26]. Participants used their customary orthosis during balance testing.

       Walking ability and fear of falling

      We assessed patients’ functional ambulation level with a 3-category self-report classification (i.e. household, limited community or full community [27]) and we inventoried orthosis use (i.e. none, ankle-foot orthosis or knee-ankle-foot orthosis). Fear of falling was assessed with the Falls Efficacy Scale (FES-I) [28], a valid and reliable 6-item questionnaire rated on a 4-point scale. The FES-score was calculated by summing the six items (range: 6 to 24). A FES score >10 indicates high fear of falling [29].

       Fall frequency

      Fall frequency was retrospectively assessed using a previously described questionnaire [2]. Participants were asked how often they had fallen in the previous year, with the following answer options: never, once, twice, three times, four times, more often.

       Statistical analysis

      Data were analysed using SPSS softwared. Considering the relative small sample size and the number of inferential analyses on the same data, we interpreted the results as explorative.
      Multiple linear regression models were developed with variable target-stepping performance and reactive obstacle-avoidance performance as dependent variables. To decide whether a variable should be included in the multiple regression model, separate univariate models were tested for independent variables known to be related to gait performance and/or falling in polio survivors. These included the MRC sum-score of most and least affected legs, MRC asymmetry-score, isometric knee extension strength of most and least affected legs, BBS-score, TUG-score, ABC-score, CWS, age, functional ambulation level, orthosis use and FES-score. When a variable was correlated with walking adaptability (i.e. p<0.1, which was chosen to make sure possible relevant variables were included in the explorative analysis), it was included in the final model using backward elimination. Pearson or Spearman's correlations coefficients were determined to assess multicollinearity (i.e. r>0.80), in which case the variable with the lowest correlation with walking adaptability was omitted. Because walking speed influences the variable stepping error, regression models for variable target-stepping performance were adjusted [30] by including CWS in the first block with method enter.
      To explore the association between walking adaptability and falling, we performed a linear regression analysis with fall frequency as dependent variable and variable target-stepping and reactive obstacle-avoidance performance as independent variables. We also performed a mediation analysis (PROCESS version 3.5 for SPSS [31]) to test to what extent walking-adaptability performance (M) mediated the association between known fall risk factors in polio survivors (X) and the fall frequency (Y, Figure 2). To perform the mediation analysis, the following conditions had to be met:
      • 1
        The fall risk factor was significantly associated with falling (i.e. the total effect)
      • 2
        The fall risk factor was significantly associated with walking adaptability (path a)
      • 3
        Walking adaptability was significantly associated with falling (path b)
      Figure 2
      Figure 2Mediation model for the association of known fall risk factors associated with walking adaptability and number of falls in the previous year
      The total effect, direct effect (path c) and indirect effects (path a and path b) of fall risk factors in polio survivors (X) on the number of falls reported in the previous year (Y). Walking-adaptability performance was added to the model as Mediator (M).
      Target-stepping and obstacle-avoidance performance were analysed in separate models. We evaluated the effect of the mediator by comparing the direct path (path c) with the total effect. If the direct path became non-significant after including a mediator, this indicates that the effect of X on Y is (partly) determined by the mediator. CWS was included in the mediation model as covariate with target-stepping performance as mediator.


      Two participants were excluded due to invalid foot placement detection on the C-Mill (see [10] for explanation). Consequently, data on 46 participants were analysed (Table 1). Missing data included one target-stepping trial and two obstacle-avoidance trials amongst 3 participants, resulting in available target-stepping data for 45 participants and obstacle-avoidance data for 44 participants. MRC-scores were missing for one participant.
      Table 1Participants’ characteristics and walking-adaptability outcomes
      Socio-demographicsPolio survivors (n=46)
      Gender (females)32 (69.6%)
      Age (years)63.2 ± 8.8
      Body Mass Index (kg/m2)27.5 ± 4.3
      Diagnosed with post-polio syndrome (yes/no)39/7
      Clinical characteristics
      MRC-sum score most-affected leg *26.0 [5]
      MRC-sum score least-affected leg *38.0 [5]
      MRC-asymmetry score *14 [14.5]
      Knee extension strength most-affected leg (in Nm)35.3 ± 38.2
      Knee extension strength least-affected leg (in Nm)96.6 ± 42.9
      TUG-score (sec)11.9 ± 3.0
      BBS-score52.8 ± 3.3
      ABC-score65.4 ± 20.3
      FES-score13.6 ± 3.5
      Functional ambulation levelHousehold7 (15.2%)
      Limited community28 (60.9%)
      Full community11 (23.9%)
      Orthosis use

      None22 (47.8%)
      AFO11 (23.9%)
      KAFO13 (28.3%)
      Walking-adaptability outcomes
      Variable target-stepping performance, VE (mm) *42.1 ± 9.5
      Reactive obstacle-avoidance success rate (%) **64 ± 26
      Fall frequency
      Self-reported fall frequency in the previous year (number of participants (%))

      None6 (13.0%)
      One10 (21.7%)
      Two6 (13.0%)
      Three7 (15.2%)
      Four4 (8.7%)
      Five or more13 (28.3%)
      Socio-demographics and walking-adaptability outcomes are shown as mean±SD, MRC-sum score (range 0-40) is shown as median [Interquartile Range (IQR)] and other clinical characteristics as number (percentage).
      Abbreviations: MRC: Medical Research Council, TUG: Timed-Up-and Go, BBS: Berg Balance Scale, ABC: Activities Specific Balance Scale, FES: Falls Efficacy Scale, AFO: Ankle-Foot Orthosis, KAFO: Knee-Ankle-Foot Orthosis.
      * based on 45 participants
      ** based on 44 participants

       Factors associated with walking adaptability

      Univariate analysis revealed a significant association between target-stepping performance and CWS (R2=0.263, Table 2), indicating a higher VE amongst faster walkers. When adding independent variables to the model, with CWS as a covariate, variable target-stepping performance was significantly associated with the MRC sum-score of the most affected leg, MRC asymmetry-score, BBS-score, ABC-score and FES-score (Table 2). Due to multicollinearity between the MRC sum-score of the most affected leg and the MRC asymmetry-score (r=0.949, p=0.001), MRC asymmetry-scores were omitted from the multiple regression model. After backward elimination of the other variables, the MRC sum-score of the most affected leg and the ABC-score, with CWS included as covariate, together explained 54% of variance in the VE (Table 2).
      Table 2Regression models for variable target-stepping performance
      Independent variablesUnivariate regressionMultiple regression
      β95% CI of βp-valueCovariateβ95% CI of βp-valueR2β95% CI of βp-value
      CWS0.513[0.249, 0.777]<0.001----0.2630.781*[0.534, 1.027]<0.001
      FES-score0.223[-0.039, 0.486]0.094CWS0.553[0.290, 0.816]<0.0010.311---
      BBS-score-0.270[-0.574, 0.034]0.080CWS0.657[0.353, 0.961]<0.0010.316---
      ABC-score-0.337[-0.595, -0.079]0.012CWS0.602[0.344, 0.860]<0.0010.374-0.321*[-0.549, -0.093]0.007
      MRC-asymmetry score0.400[0.151, 0.650]0.002CWS0.628[0.378, 0.878]<0.0010.419---
      MRC-sum score (most-affected leg)-0.452[-0.711, -0.193]0.001CWS0.707[0.448, 0.966]<0.0010.440-0.417*[-0.659, -0.173]0.001
      Results of the regression analysis for target-stepping performance with independent variables and CWS as covariate are shown. Abbreviations: 95% CI: 95% Confidence Interval, CWS: comfortable walking speed, FES: Falls Efficacy Scale, BBS: Berg Balance Scale, ABC: Activities Specific Balance Confidence, MRC: Medical Research Council.
      *For the multiple regression model: R2=0.540
      Univariate analysis revealed that reactive obstacle-avoidance performance was associated with the MRC sum-score of the most affected leg, MRC asymmetry-score, isometric knee extension strength of the most affected leg, isometric knee extension strength of the least affected leg, BBS-score, TUG-score and the ABC-score. After correcting for multicollinearity, all variables except the MRC asymmetry-score were added to the multiple regression model. The MRC sum-score of the most affected leg and the isometric knee extension strength of the least affected leg explained 32% of the variance in obstacle-avoidance performance (Table 3).
      Table 3Regression models for obstacle-avoidance performance
      Independent variablesUnivariate regressionMultiple regression
      β95% CI of βp-valueR2β95% CI of βp-value
      ABC-score0.299[-0.006, 0.604]0.0540.089---
      MRC-asymmetry score-0.356[-0.651, -0.061]0.0190.127---
      BBS-score0.384[0.097, 0.672]0.0100.148---
      MRC-sum score (most-affected leg)0.425[0.140, 0.711]0.0040.1810.318[0.043, 0.593]0.024
      Knee-extension (most-affected leg)0.449[0.171, 0.727]0.0020.202---
      CWS0.457[0.180, 0.734]0.0020.209---
      TUG-score-0.472[-0.750, -0.194]0.0010.223---
      Knee-extension (least-affected leg)0.486[0.214, 0.758]0.0010.2370.383[0.108, 0.659]0.008
      Abbreviations: 95% CI: 95% Confidence Interval, ABC: Activities Specific Balance Confidence, MRC: Medical Research Council, BBS: Berg Balance Scale, CWS: comfortable walking speed, TUG: Timed-Up-and-Go Test.
      *For the multiple regression model: R2=0.316

       Relationship between walking adaptability and falling

      The fall frequency was significantly associated with target-stepping performance (R2=0.277, p<0.001), but not with obstacle-avoidance performance (R2=0.0, p=0.931).
      The ABC-score and MRC sum-score of the most affected leg met all conditions needed to perform mediation analysis with variable target-stepping performance as mediator. Once target-stepping performance was added as mediator with CWS included as covariate, the direct effect of the ABC-score and the MRC sum-score of the most affected leg on the number of falls became non-significant, indicating a significant mediation effect (Table 4).
      Table 4Results of the mediation analysis
      Fall-risk factor (X)Indirect effectDirect effectTotal effect
      Path aPath bPath cCoefficientp-value
      A (SE)95% CI of Ap-valueB (SE)95% CI of Bp-valueC (SE)95% CI of Cp-value95% CI of Coefficient
      ABC-score-0.151 (0.057)[-0.266,

      0.0120.110 (0.032)[0.044, 0.175]0.002-0.024 (0.013)[-0.050, 0.002]0.068-0.040 (0.013)[-0.067,

      MRC-sum score (most-affected leg)-0.406 (0.115)[-0.639,

      0.0010.113 (0.034)[0.044, 0.183]0.002-0.30 (0.029)[-0.089, 0.028]0.301-0.076 (0.028)[-0.133,

      Mediation effects (A/B/C/Coefficient) of fall-risk factor (X) on the number of falls reported in the previous year (Y), with target-stepping performance as mediator (M) and CWS as covariate. For the mediation model, see Figure 2.
      Abbreviations: 95% CI = 95% Confidence Interval, ABC = Activities-specific Balance Confidence scale; MRC = Medical Research Council; SE = standard error


      In polio survivors who had experienced falls and/or fear of falling, walking adaptability seemed to be mainly determined by leg muscle strength (for both reactive obstacle-avoidance and target-stepping performance) and balance confidence (for target-stepping performance alone). However, despite the known role of leg muscle weakness in falls, participants’ performance in reactive obstacle-avoidance was not associated with falling, whereas variable target-stepping performance explained 28% of variance in the number of falls reported in the previous year. Furthermore, the association between leg muscle weakness and balance confidence with falling was mediated by target-stepping performance, suggesting that leg muscle weakness leads to a reduced ability to ensure safe foot placement, consequently increasing fall risk in polio survivors. Even though findings were interpreted as explorative to prevent type I errors, they might aid identification of polio survivors with walking-adaptability limitations and those at risk of falling.
      Since it is easier to follow step targets when walking more slowly [30], we added CWS as a covariate in regression analyses for target-stepping performance. Once CWS was included as a covariate, reduced variable target-stepping performance was associated with leg muscle weakness, balance performance, balance confidence and increased fear of falling in polio survivors. In previous work in elderly [32] and Parkinson's disease [12], target stepping was associated with (reactive) balance performance, executive function and reaction time. In these two studies, participants had to hit suddenly appearing targets, which might explain the relevance of response factors in completing this task successfully. In our study, we calculated target-stepping performance over a trajectory of step targets for both legs. Muscle weakness of the most affected leg may limit target-stepping performance either through an inability to aim the foot at the target or by causing balance problems when serving as stance leg while the other foot is aimed at the target. We previously found that the least affected leg shows reduced target-stepping performance compared to the most affected leg [10], which supports the idea that muscle weakness in the most affected leg is the restrictive factor in target stepping while supporting body weight as the stance leg.
      Reactive obstacle avoidance was significantly associated with leg muscle strength, balance performance, balance confidence and walking speed. Similarly, leg muscle strength (including the hip, knee and ankle muscles) and dynamic balance performance were associated with obstacle-avoidance performance in people with Parkinson's disease [33], while people with stroke have difficulties in controlling balance while crossing obstacles [34]. Based on our multivariate model, leg muscle strength appears to be the best predictor of reactive obstacle-avoidance performance. For successful 2D obstacle avoidance, step length needs to be increased to generate an avoidance manoeuvre, an undertaking that requires sufficient leg muscle strength [32]. Furthermore, as gait stability decreases with increasing step length [35], leg muscle strength and balance control are essential to maintain stability [36, 37].
      While reactive obstacle-avoidance performance was primarily affected by leg muscle weakness, which is a known risk factor for falling in polio survivors [2, 4, 15, 16], surprisingly we found no association between reactive obstacle-avoidance performance and falling. However, our obstacle-avoidance tests focused on avoiding trips rather than on the recovery from a trip. Although tripping is a frequent cause of falls among polio survivors [2, 6], it is not yet known whether failure to avoid an obstacle leads to tripping (i.e. poor walking adaptability) or, alternatively, unsuccessful recovery from a trip is the main problem. Our results suggest that the latter possibility might be well worth pursuing in future studies. Alternatively, avoiding projections of 2D obstacles on a treadmill while wearing a safety harness might not sufficiently represent situations in daily life, since no toe clearance is required and mistakes do not result in trips [38]. This might perhaps explain why obstacle avoidance was not associated with falling. By contrast, variable target-stepping performance was associated with the number of falls reported in the previous year. Target stepping involves foot placement control of every single step and thus requires high levels of gait control. Due to polio sequelae, a reduced ability to activate the quadriceps muscles [39] might impair gait control and consequently affect fall risk. Furthermore, we identified reduced target-stepping performance as a mediator of the relationship between leg muscle weakness/balance confidence and falling. Leg muscle weakness and self-reported balance problems [2] are known risk factors for falling in polio survivors. This is concordant with our results, which also indicate that these two factors limit target-stepping performance, thus helping further illuminate fall risk. This knowledge could be used to identify polio survivors at increased risk of falling, as previously shown for stroke survivors and people with Parkinson's disease, where the inclusion of walking-adaptability tests in fall risk assessment led to better identification of prospective fallers [40]. Studies of walking-adaptability training have also shown promising results in reducing fall risks in community-dwelling elderly [41]. This approach therefore warrants further investigation in polio survivors.

       Study limitations

      Our participants reported at least one fall in the previous year and/or a fear of falling, reducing within-group contrasts. Further, falls were assessed retrospectively and results may thus have been influenced by recall bias, which could be avoided with a prospective fall registration. The relative small sample size only allowed us to do explorative analysis. The level of uncertainty in our findings was indicated by the wide 95% confidence intervals for all coefficients in the regression analyses. Nevertheless, our cross-sectional study design laid groundwork on factors associated with walking adaptability in polio survivors, which should be pursued and strengthened in a longitudinal study with a larger sample.


      Leg muscle weakness and reduced balance confidence seem to limit walking adaptability in individuals with prior polio who have experienced falls and/or a fear of falling. Explorative analyses indicate that reduced target-stepping performance is associated with a higher fall frequency and mediates the association of falling with leg muscle weakness and balance confidence, while reactive obstacle-avoidance performance is not associated with falling. This suggests that limitations in the ability to ensure safe foot placement due to reduced leg muscle strength might increase fall risk in polio survivors, which should be confirmed in a larger sample.


      a. The C-Mill, Motek, Amsterdam, the Netherlands b. Matlab, The Mathworks, version R2019a, Natick, MA, USA c. Biodex system 4, Biodex Medical Systems, Shirley, NY, USA d. IBM SPSS Statistics, version 26; SPSS Inc., Chicago, IL, USA
      Acknowledgements We would like to thank all polio survivors for their voluntary participation in this study, and Melvyn Roerdink for his help in the data analysis.
      Conflict of interest None.
      Funding We could purchase the C-Mill for the department of Rehabilitation Medicine of the Amsterdam UMC with a grant from the Amsterdam Movement Sciences research institute. No other grants were received.
      Author contribution FN and MB initiated this study. JT performed the data collection and data analysis. Statistical analysis were performed by JT, MB and MJ. JT prepared the manuscript together with MB and MJ, all authors contributed by revising the manuscript and approved the final version for submission.


      [1] Matsuda PN, Verrall AM, Finlayson ML, Molton IR, Jensen MP. Falls among adults aging with disability. Archives of physical medicine and rehabilitation. 2015;96(3):464-471.
      [2] Bickerstaffe A, Beelen A, Nollet F. Circumstances and consequences of falls in polio survivors. Journal of rehabilitation medicine. 2010;42(10):908-915.
      [3] Nam KY, Lee S, Yang EJ, Kim K, Jung SH, Jang SN, et al. Falls in Korean Polio Survivors: Incidence, Consequences, and Risk Factors. Journal of Korean medical science. 2016;31(2):301-309.
      [4] Imoto D, Sawada K, Horii M, Hayashi K, Yokota M, Toda F, et al. Factors associated with falls in Japanese polio survivors. Disability and rehabilitation. 2019:1-5.
      [5] Stalenhoef PA, Diederiks JP, de Witte LP, Schiricke KH, Crebolder HF. Impact of gait problems and falls on functioning in independent living persons of 55 years and over: a community survey. Patient education and counseling. 1999;36(1):23-31.
      [6] Brogardh C, Lexell J, Hammarlund CS. Experiences of falls and strategies to manage the consequences of falls in persons with late effects of polio: A qualitative study. Journal of rehabilitation medicine. 2017;49(8):652-658.
      [7] Legters K, Verbus NB, Kitchen S, Tomecsko J, Urban N. Fear of falling, balance confidence and health-related quality of life in individuals with postpolio syndrome. Physiotherapy theory and practice. 2006;22(3):127-135.
      [8] Balasubramanian CK, Clark DJ, Fox EJ. Walking adaptability after a stroke and its assessment in clinical settings. Stroke research and treatment. 2014;2014:591013.
      [9] Shumway-Cook AEPaA. Dimensions of Mobility: Defining the Complexity and Difficulty Associated With Community Mobility. Journal of Aging and Physical Activity. 1999;7:7-19.
      [10] Tuijtelaars J, Roerdink M, Raijmakers B, Nollet F, Brehm MA. Polio survivors have poorer walking adaptability than healthy individuals. Gait & posture. 2021;87:143-148.
      [11] Hollands KL, Pelton TA, van der Veen S, Alharbi S, Hollands MA. A novel and simple test of gait adaptability predicts gold standard measures of functional mobility in stroke survivors. Gait & posture. 2016;43:170-175.
      [12] Caetano MJD, Lord SR, Allen NE, Song J, Paul SS, Canning CG, et al. Executive Functioning, Muscle Power and Reactive Balance Are Major Contributors to Gait Adaptability in People With Parkinson's Disease. Frontiers in aging neuroscience. 2019;11:154.
      [13] Caetano MJD, Lord SR, Brodie MA, Schoene D, Pelicioni PHS, Sturnieks DL, et al. Executive functioning, concern about falling and quadriceps strength mediate the relationship between impaired gait adaptability and fall risk in older people. Gait & posture. 2018;59:188-192.
      [14] Caetano MJD, Lord SR, Allen NE, Brodie MA, Song J, Paul SS, et al. Stepping reaction time and gait adaptability are significantly impaired in people with Parkinson's disease: Implications for fall risk. Parkinsonism Relat Disord. 2018;47:32-38.
      [15] Lord SR, Allen GM, Williams P, Gandevia SC. Risk of falling: predictors based on reduced strength in persons previously affected by polio. Archives of physical medicine and rehabilitation. 2002;83(6):757-763.
      [16] Brogardh C, Flansbjer UB, Lexell J. Determinants of Falls and Fear of Falling in Ambulatory Persons With Late Effects of Polio. PM & R: the journal of injury, function, and rehabilitation. 2017;9(5):455-463.
      [17] Brogardh C, Lexell J. Falls, fear of falling, self-reported impairments, and walking limitations in persons with late effects of polio. PM & R: the journal of injury, function, and rehabilitation. 2014;6(10):900-907.
      [18] Pasquetti P, Apicella L, Mangone G. Pathogenesis and treatment of falls in elderly. Clinical cases in mineral and bone metabolism: the official journal of the Italian Society of Osteoporosis, Mineral Metabolism, and Skeletal Diseases. 2014;11(3):222-225.
      [19] Tuijtelaars J, Brehm M-A, Nollet F, Roerdink M. Construct Validity And Reproducibility of C-Mill Walking-Adaptability Assessment in Polio Survivors. Archives of physical medicine and rehabilitation. 2021;102(10):e47-e48.
      [20] Vandenbroucke JP, von Elm E, Altman DG, Gøtzsche PC, Mulrow CD, Pocock SJ, et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration. Annals of internal medicine. 2007;147(8):W163-194.
      [21] Medical Research Council. Aids to examination of the peripheral nervous system1976.
      [22] Nollet F, Beelen A, Prins MH, de Visser M, Sargeant AJ, Lankhorst GJ, et al. Disability and functional assessment in former polio patients with and without postpolio syndrome. Archives of physical medicine and rehabilitation. 1999;80(2):136-143.
      [23] Berg KO W-DS, Williams JI, Gayton D. Measuring balance in the elderly: preliminary development of an instrument. Physiotherapy Canada. 1989;41:304-311.
      [24] Mathias S, Nayak US, Isaacs B. Balance in elderly patients: the "get-up and go" test. Archives of physical medicine and rehabilitation. 1986;67(6):387-389.
      [25] Flansbjer UB, Lexell J. Reliability of gait performance tests in individuals with late effects of polio. PM & R: the journal of injury, function, and rehabilitation. 2010;2(2):125-131; quiz 121 p following 167.
      [26] Powell LE, Myers AM. The Activities-specific Balance Confidence (ABC) Scale. The journals of gerontology Series A, Biological sciences and medical sciences. 1995;50a(1):M28-34.
      [27] Brehm MA, Ploeger HE, Nollet F. Self-reported functional ambulation is related to physical mobility status in polio survivors; a cross-sectional observational study. Annals of physical and rehabilitation medicine. 2020.
      [28] Kempen GI, Todd CJ, Van Haastregt JC, Zijlstra GA, Beyer N, Freiberger E, et al. Cross-cultural validation of the Falls Efficacy Scale International (FES-I) in older people: results from Germany, the Netherlands and the UK were satisfactory. Disability and rehabilitation. 2007;29(2):155-162.
      [29] Delbaere K, Close JC, Mikolaizak AS, Sachdev PS, Brodaty H, Lord SR. The Falls Efficacy Scale International (FES-I). A comprehensive longitudinal validation study. Age and ageing. 2010;39(2):210-216.
      [30] Roerdink M, Geerse DJ, Peper CLE. 'Haste makes waste': The tradeoff between walking speed and target-stepping accuracy. Gait & posture. 2021;85:110-116.
      [31] Hayes AF. Introduction to Mediation, Moderation, and Conditional Process Analysis. A Regression-Based Approach. New York: Guilford Publications; 2018.
      [32] Caetano MJD, Menant JC, Schoene D, Pelicioni PHS, Sturnieks DL, Lord SR. Sensorimotor and Cognitive Predictors of Impaired Gait Adaptability in Older People. The journals of gerontology Series A, Biological sciences and medical sciences. 2017;72(9):1257-1263.
      [33] Liao YY, Yang YR, Wu YR, Wang RY. Factors influencing obstacle crossing performance in patients with Parkinson's disease. PLoS One. 2014;9(1):e84245.
      [34] Said CM, Goldie PA, Patla AE, Culham E, Sparrow WA, Morris ME. Balance during obstacle crossing following stroke. Gait & posture. 2008;27(1):23-30.
      [35] Hak L, Houdijk H, Steenbrink F, Mert A, van der Wurff P, Beek PJ, et al. Stepping strategies for regulating gait adaptability and stability. Journal of biomechanics. 2013;46(5):905-911.
      [36] van Dieën JH, Pijnappels M, Bobbert MF. Age-related intrinsic limitations in preventing a trip and regaining balance after a trip. Safety science. 2005;43(7):437-453.
      [37] Wolfson L, Judge J, Whipple R, King M. Strength is a major factor in balance, gait, and the occurrence of falls. The journals of gerontology Series A, Biological sciences and medical sciences. 1995;50 Spec No:64-67.
      [38] Punt M, Bruijn SM, Wittink H, van de Port IG, Wubbels G, van Dieën JH. Virtual obstacle crossing: Reliability and differences in stroke survivors who prospectively experienced falls or no falls. Gait & posture. 2017;58:533-538.
      [39] Beelen A, Nollet F, de Visser M, de Jong BA, Lankhorst GJ, Sargeant AJ. Quadriceps muscle strength and voluntary activation after polio. Muscle Nerve. 2003;28(2):218-226.
      [40] Geerse DJ, Roerdink M, Marinus J, van Hilten JJ. Walking adaptability for targeted fall-risk assessments. Gait & posture. 2019;70:203-210.
      [41] Nørgaard JE, Jorgensen MG, Ryg J, Andreasen J, Danielsen MB, Steiner DK, et al. Effects of gait adaptability training on falls and fall-related fractures in older adults: a systematic review and meta-analysis. Age and ageing. 2021;50(6):1914-1924.