Secondary Logo

Gait Training With Visual Feedback and Proprioceptive Input to Reduce Gait Asymmetry in Adults With Cerebral Palsy: A Case Series

Levin, Ilana PT, DPT; Lewek, Michael D. PT, PhD; Feasel, Jeff MS; Thorpe, Deborah E. PT, PhD

doi: 10.1097/PEP.0000000000000362

Purpose: The purpose of this case series was to investigate the feasibility of using visual feedback on gait asymmetry during gait retraining and whether this leads to reduced asymmetry, improvement in gait speed, cost of walking, and dynamic balance in ambulant adults with cerebral palsy (CP).

Methods: Five adults with CP, who were ambulatory and had step length or stance time asymmetry, trained for 18 sessions on a split-belt treadmill with concurrent visual feedback from a virtual environment. Training also included overground gait training to encourage transfer of learning.

Results: All participants reduced gait asymmetry and improved on outcomes at posttest and follow-up.

Conclusions: Outcome measures and training protocols were feasible in this sample of convenience of adults with CP who were ambulatory and who did not have visual impairment. The adults with CP in this study demonstrated individual improvements in gait and balance following training.

This paper examines whether providing visual feedback about gait asymmetry in the form of virtual reality during gait retraining improves gait speed and dynamic balance in adults with cerebral palsy.

Department of Allied Health Sciences, Division of Physical Therapy, Program in Human Movement Science (Drs Levin, Thorpe, and Lewek), and Department of Computer Science (Mr Feasel), University of North Carolina-Chapel Hill, Chapel Hill.

Correspondence: Ilana Levin, PT, DPT, Department of Allied Health Sciences, Division of Physical Therapy, University of North Carolina-Chapel Hill, 3043 Bondurant Hall, CB 7135, Chapel Hill, NC 27599 (

Grant Support: This study was supported by a grant from the UNC-CH University Research Council.

The authors declare no conflicts of interest.

Back to Top | Article Outline


The brain lesion that causes cerebral palsy (CP) is nonprogressive.1 However, walking often begins to decline in adults with CP in their third and fourth decades of life and coincides with an increase in musculoskeletal pain, fatigue, and deficits in balance.2 Reduced balance is the most frequent complaint associated with the decline in walking.3 The decline in walking imposes a more sedentary lifestyle, increasing risk of falling, negatively affecting health status, employment, quality of life, and may result in increased disability.4 Life expectancy for most adults with CP is within 6 years of the general population, and an early decline in walking may cause a decline in health and function typically appearing later in life in the general population.2

Compared with typical peers, gait of adults with CP is often slower, less stable, less mechanically efficient,5 has a higher energy cost of walking,6 and is often asymmetrical in step length and/or stance time.7 Intervention research in adults with CP is sparse and insufficient in its ability to guide evidence-based decision making regarding which deficits to target to prevent or ameliorate the deterioration in gait and balance.8,9

Reducing gait asymmetry in adults poststroke has been studied using split-belt treadmill (TM)-based gait training.10 In adults with CP, the effects of single-belt TM-based gait training on various gait abnormalities have been studied, but TM training to modify gait asymmetry has not been studied.11 During TM walking, a congruent optic flow, using a virtual environment (VE), contributes to the control of gait speed and step length and increases joint angles toward overground values.12 Combining visual feedback from a VE with TM-based gait training was previously used to reduce gait asymmetry in individuals poststroke.12 However, people with chronic gait deviations may not perceive the deviations as requiring correction. Therefore, providing augmented feedback through VE, about the deviations, during gait retraining can potentially reduce these deviations.13 VE-based feedback augments the feedback on gait abnormalities and converts information that is usually accessible only by an internal focus of attention, such as attention to gait kinematics and asymmetry, to accessible information. Using an external focus of attention is advantageous when learning a complex task.14 Coupling VE-based visual feedback with proprioceptive feedback from a computerized TM was found to promote desirable modifications of various spatiotemporal gait parameters in older adults and adults poststroke.15 Adding overground transfer practice following TM gait training with error augmentations was shown to assist reducing step length asymmetry overground.16

Although many adults with CP exhibit gait asymmetry, it is not known whether they can reduce their gait asymmetry following training. Furthermore, it is unknown whether reducing gait asymmetry positively affects other mobility-related outcomes.12 The goals of this study were to assess the feasibility and efficacy of training adults with cerebral palsy to reduce gait asymmetry using concurrent visual and proprioceptive feedback followed by overground gait training to enhance transfer of learning.

Back to Top | Article Outline



Seven participants were enrolled. All signed an informed consent approved by the Institutional Review Board of the University of North Carolina at Chapel Hill. Inclusion criteria included (1) a diagnosis of CP, (2) age 18 to 65 years, (3) ability to ambulate at least 10 m on a flat surface without an assistive device, and (4) step length or stance time asymmetry ratios above cutoff values during comfortable gait speed (1.08 for step length asymmetry [SLA] and 1.05 for stance time asymmetry [STA]).12 Exclusion criteria included (1) severe osteoporosis, cardiorespiratory, or metabolic diseases, (2) persistent/chronic pain, (3) uncontrolled seizures, (4) significant expressive communication impairment, and (5) inability to understand simple instructions in English.

Back to Top | Article Outline

Training Protocol

Participants were required to complete 18 sessions, training 2 to 3 times a week. Each session included 22 minutes of gait training on the Integrated Virtual Environment Rehabilitation Treadmill (IVERT) system.17 The IVERT system (Figure 1) provides visual input while the participant is walking on a split-belt TM (Bertec Corp, Worthington, Ohio) in a “user-controlled” mode. In this mode, the average speed of the TM belts matches the participant's walking speed, and the difference in speed between the belts is proportional to either the STA or SLA. This speed difference between the belts offers proprioceptive feedback to the user. A VE is projected on 3 screens surrounding the TM and produces an optic flow that provides feedback on gait speed, such that the scenery moves toward the user congruently with the speed of walking. The IVERT also provides augmented visual feedback regarding gait asymmetry by sliding the projected scenery sideways along an arc, for which the radius is proportional to real-time STA or SLA. The visual feedback affords real-time error correction. When the user reduces gait asymmetry, the VE simulates walking on a straighter path and the difference in speeds between the belts is reduced.

Fig. 1

Fig. 1

All participants were full weight bearing and secured in an overhead safety harness. Training sessions began with a 2-minute warm-up walking in a fixed speed mode, with no feedback on gait asymmetry, followed by two 10-minute trials (with 5-minute rest between trials) walking in a user-controlled speed mode with visual feedback on asymmetry. All trainings were guided by a licensed physical therapist. Concurrent verbal feedback during training on the IVERT was minimal to allow participants to focus their attention on walking and processing the visual and proprioceptive feedback. Gait training on the IVERT was individualized and started with block practice advancing to variable practice of the following activities as training progressed: (1) accommodation to user-controlled mode, (2) exploring strategies to reduce asymmetry (modifying segmental coordination), (3) increasing duration of symmetrical walking, (4) increasing speed while maintaining gait symmetry, (5) incorporating countermovements of upper extremities/reducing upper extremity posturing, and (6) volitionally overcorrecting and then returning to gait symmetry.

IVERT training was followed by 10 to 15 minutes of overground gait training to facilitate skill transfer.18 Overground training included (1) whole-task practice—walking at various speeds and with speed modulations while implementing strategies used on the IVERT to reduce gait asymmetry, (2) part-task practice—weight shifts and dynamic control exercises to refine and expand movement repertoire, and (3) return to whole-task practice.

Back to Top | Article Outline

Outcome Measures

Participants were evaluated at (1) baseline (pretest); (2) postintervention, within 1 week after the last session (posttest); and (3) follow-up, 4 to 6 weeks after completing the intervention. The following outcomes were assessed:

  1. Gait speed and spatiotemporal asymmetry were measured using a 14-ft (4.3-m) GAITRite Walkway System (CIR Systems, Inc, Havertown, Pennsylvania). The GAITRite mat has good psychometrics in adults following stroke.19 Using a similar protocol, participants performed 3 valid passes overground at a comfortable gait speed (CGS) and 3 at a fast gait speed (FGS). For each speed, SLA and STA were calculated by dividing the larger value of one side by the respective smaller value on the contralateral side.
  2. Functional walking capacity was assessed using the 6-minute walk test (6MWT). This is a submaximal test of “functional walking capacity” with reliability in adults with CP. In adults with CP, the distance walked during the test is considered to be influenced by neuromuscular and musculoskeletal components rather than cardiopulmonary fitness.20
  3. Oxygen cost of walking (CoW) during the 6MWT.6 Data were collected using a portable metabolic cart (K4 b2; COSMED USA Inc, Chicago, Illinois) during baseline resting and throughout the 6MWT and processed as previously published.21
  4. Dynamic balance was measured using the Four-Square Step Test (FSST).22 The FSST is a timed test that requires participants to step rapidly in a multidirectional sequence over 4 walking canes placed in a cross-configuration. The FSST has good psychometrics in older adults and people poststroke.23 In our study, participants performed 1 practice trial and 2 timed trials; the faster time was used for analysis.
  5. Physical activity was measured using an ankle-worn StepWatch (SW) activity monitor. SW monitors were previously used to assess step counts of youth and adolescents with CP.24 Informal interviews were conducted to account for activities and life events during the monitored period.
Back to Top | Article Outline

Data Analysis

Results were analyzed as a case series. All changes were evaluated with each participant as his or her own control. We compared changes in outcomes with previously published norms for the population developing typically and with minimal detectable change (MDC) values for comparable patient populations. MDC is the amount of change necessary to suggest that the change in performance on an outcome measure exceeds variations and errors in measurement.19

MDCs for gait speed and gait asymmetry were based on values established in adults poststroke. CGS: MDCCGS = 0.2 m/s, MDCSLA = 0.15, MDCSTA = 0.09. FGS: MDCFGS = 0.22 m/s, MDCSLA = 0.2, MDCSTA = 0.1.19 Normative values for asymmetry ratios were based on values of adults typically developing: SLA ≤ 1.08, STA ≤ 1.05.19 MDC for 6MWT was previously established for adults with CP walking without assistive devices when comparing the first and second trials (56 m).25 Costs of walking MDC values were previously established as a percentage change from baseline,26 and a conservative MDC of 13.6%, previously established for adults with polio, was adopted for current analysis. For FSST, a cut point of 15 seconds or more, which identifies older adult who fall and people with acute stroke at risk of falling, has been published27; however, MDC values were not available for this test. Percentage change from baseline was used to normalize measures and reflect individual changes. Valid daily step monitoring data were computed from days with 8 or more hours during which evidence of use was observed. Average and maximum daily steps were derived from the valid daily (24 hours) SW data for each participant.

Back to Top | Article Outline



From our sample of convenience, 5 of the 7 adults with CP completed the study (Table 1). One participant could not follow the training protocol because of difficulties mastering the user-controlled speed mode and difficulties using the visual feedback. A second participant disclosed chronic hip and knee pain after the first training session. No adverse events occurred during the intervention or testing sessions. In 2 testing sessions, the COSMED portable metabolic cart failed to collect reliable data. For all other tests, in all sessions, data collection was completed.



Back to Top | Article Outline

Results by Participant

S1, a 69-year-old man with hemiplegic CP, reduced his SLA at CGS at posttest to within the normative range (Figure 2A) and increased CGS (Figure 2B) by more than the respective MDC. At follow-up, his SLA at CGS remained within the normative range. His SLA at FGS was within the normative range at pretest and remained within the normative range at posttest and follow-up (Figure 2C). He increased the 6MWT distance by a change larger than the MDC at follow-up (Figure 3A). He reduced CoW (consumed less milliliter oxygen per kilogram per meter) at posttest and reduced it further at follow-up (Figure 3B). Finally, he reduced FSST time by 16% at posttest and by another 2% at follow-up (Figure 3C).

Fig. 2

Fig. 2

Fig. 3

Fig. 3

S2, a 25-year-old man with diplegic CP with mild ataxia, did not reduce his SLA at CGS (Figure 2A) or change his CGS (Figure 2B). He reduced his SLA at FGS to within the normative range at posttest (Figure 2C) and had no change in FGS (Figure 2D). At follow-up, he remained in the normative range of asymmetry at FGS. He did not change his 6MWT distance throughout (Figure 3A); however, he demonstrated an increase in CoW at posttest. He reported having recovered from a recent respiratory illness contracted immediately after completing the intervention that may explain the increase in CoW. A COSMED portable metabolic cart malfunction led to missing data for his follow-up CoW (Figure 3B). He reduced FSST time by 17% at posttest and by another 6% at follow-up (Figure 4).

Fig. 4

Fig. 4

S3, a 24-year-old man with hemiplegic CP, reduced SLA at CGS to within the normative range at posttest and stayed in the normative range at follow-up (Figure 2A). He had initial SLA within the normative range at FGS, which was maintained throughout the study (Figure 2C). There was no change in his 6MWT distance throughout the study (Figure 3A), and CoW did not change at follow-up (Figure 3B). His posttest CoW data were missing because of a COSMED portable metabolic cart malfunction. He reduced FSST time by 22% at posttest but increased FSST time from posttest by 16% at follow-up (not reaching pretest value; Figure 3C).

S4, a 27-year-old man with diplegic CP, reduced his SLA at CGS (Figure 2A) and CGS at posttest (Figure 2B) by more than the respective MDC values. However, at follow-up, he increased his asymmetry at CGS (not reaching pretest values; Figure 2A), while maintaining the increase in CGS (Figure 2B). He did not change asymmetry at FGS, nor changed FGS at posttest, but at follow-up, he reduced asymmetry at FGS to within the normative range (Figure 2C) while maintaining FGS (Figure 2D). He increased 6MWT distance at posttest by more than MDC and maintained the improvement at follow-up (Figure 3A). He demonstrated an increase in CoW and an increase in walking speed (Figure 3B). On the FSST, at pretest he was above the cutoff score for “at risk for falling” (≥15 seconds). He reduced FSST time by 49% at posttest and at follow-up maintained most of the gain, increasing his time from posttest by 8% (Figure 3C).

S5, a 38-year-old woman with diplegic CP, reduced her STA at CGS to within the normative range at posttest and maintained within the normative range value at follow-up (Figure 2A). She did not change CGS (Figure 2B). She reduced STA at FGS to within the normative range and increased her FGS at posttest. At follow-up, she maintained within the normative range of STA at FGS (Figure 2C) as well as the increase in FGS (Figure 2D). She increased 6MWT distance at posttest by more than the MDC value and maintained the achieved improvement at follow-up (Figure 3A). She increased CoW as well as her average walking speed during the 6MWT (Figure 3B). She reduced FSST time by 33% at posttest and maintained the improvement at follow-up (Figure 3C).

Back to Top | Article Outline

StepWatch Results

Participants wore the StepWatch between 3 and 8 days (mean 5.5 days; Figure 4). The largest within participant changes in the average number of strides and in the number of strides walked during the most active day were mainly explained by activities that were not routine for each participant. For example, S1 went golfing during his posttest, and he was recovering from a medical complication during the follow-up week. S2 started a new job, that required increased walking, just before follow-up after being unemployed. S4 got sick during wear time postintervention, and S5 was sick during wear time at follow-up. We acknowledge that the presence of atypical activities and events during the monitoring periods made the validity of the data questionable; hence, no further analysis was indicated.

Back to Top | Article Outline


The intervention and testing protocols were found to be feasible for 5 of the 7 adults with CP. All 5 adhered to the schedule, expressed motivation and interest throughout the training period, and completed the training protocol without adverse effects. Prescreening should be added in future studies for possible visual processing impairments.

In this study, adults with CP demonstrated the ability to acquire a more symmetrical gait pattern following training. The age range in this study was broad and we were aware of age-related changes when learning to modify gait.28 However, our goal was to explore whether adults with CP of similar ambulatory abilities were able to learn to modify their gait. Of the 5 participants completing the study, 4 participants qualified based on SLA ratio and 1 participant qualified on STA ratio. All 5 participants reduced asymmetry at CGS either at posttest (S1, S3, S4, and S5) or at follow-up (S2). Three of the 5 participants reduced asymmetry at FGS at posttest (S2 and S5) or at follow-up (S4). The 2 participants (S1 and S3) who did not change asymmetry at FGS were already within the normative range at FGS at pretest. All participants improved their FSST time at posttest and at follow-up, either maintained the improvement or declined minimally, but did not return to pretest values. Changes in the secondary outcomes, CGS, FGS, 6MWT distance, CoW, and daily average number of strides were variable. Three participants (S1, S4, and S5) demonstrated improvements on combinations of 2 or more of the secondary outcome measures at posttest, and most of the improvements were maintained or surpassed at follow-up. However, 2 participants (S2 and S3) improved on only 1 secondary outcome, dynamic balance (FSST). S2 had close to normative values for walking speeds, which may have produced a ceiling effect. S3 exhibited some difficulties with visual-perceptual processing and required more verbal cues during training to attend and respond to the visual feedback provided by the VE system, implying possible cognitive processing deficits. Another explanation for why S2's reduction in FGS asymmetry and S3's reduction in CGS asymmetry did not result in respective improvement in gait speed is that these participants may have directed attention to their gait pattern during the overground walkway testing, with a speed-accuracy trade-off that affected their gait speed.29

Three participants (S1, S4, and S5) improved 6MWT distance at posttest and continued to improve at follow-up, yet only S4 had a reduction in FGS asymmetry between post-test and follow-up. This may imply that these participants had optimized the altered gait pattern by follow-up, making it more efficient even without changing parameters affecting asymmetry. This improvement in efficiency is supported by mild reduction in CoW between posttest and follow-up demonstrated by all 3 participants who increased their walking speed during the 6MWT.

All participants improved their balance, as demonstrated by FSST time, and supported by anecdotal participant reports. S4 (who was at high risk of falling) and S5 (who was at the threshold of an increased risk for falling) exhibited the largest reduction in FSST time to values below the risk threshold. These participants also demonstrated the largest pretest asymmetry and the largest reduction in asymmetry.

We were not able to identify changes in the amount of daily walking that could be attributed to the intervention. Considering the variability in daily walking, the average daily strides and the most active stride day were not sufficiently robust to measure effects of the intervention. It is also unknown whether the number of days the SWs were used was sufficient for a reliable estimate of regular levels of walking in adults with CP. However, participants shared anecdotes of positive effects on ease of walking and perceived improvement in balance during their routine daily walking. Considering that life events may cause high variability in daily walking, stride counting may not be a good measure to reflect changes in walking-related function. Evaluating changes in step intensity24 and using an accelerometer that provides data about various types of movements, not just walking, may be more reliable measures to account for changes in levels of physical activity. Longer monitoring periods would account for differences between weekdays and weekends and lessen the effects of atypical events. A larger sample is needed to conduct a group statistical analysis of the amount of daily strides and rate of stepping intensity.

Projected VE was a feasible tool to provide real-time augmented feedback during gait retraining for adults with CP. The constant availability of the VE feedback afforded switching attention between internal and external feedback as participants were engaged in “problem solving” to optimize gait symmetry. The technology-based rehabilitation environment was engaging and participants demonstrated high attendance and compliance.30

The limitations of this case series were as follows: (1) the intervention included overground transfer training, which limits the attribution of changes in outcomes to the feedback during treadmill walking; (2) the small number of participants; (3) no control group; (4) use of outcome tools with unknown psychometrics for adults with CP; and (5) lack of screening for visual acuity and visual-perceptual processing skills.

Back to Top | Article Outline


In this small case series, TM-based gait retraining in adults with CP, using real-time visual feedback on asymmetry plus overground transfer-of-skills training, was found to be feasible for adults with no significant visual processing limitations. The participants demonstrated the ability to change spatiotemporal gait parameters to decrease gait asymmetry. The reduction in asymmetry may have contributed to the positive gains in other outcomes including short- and long-distance gait speeds, gait efficiency, and dynamic balance. We cannot determine whether these positive changes were the direct result of the reduction in gait asymmetry or the result of other possible benefits from the multifaceted intervention.

Future research should address the question of whether targeting gait asymmetry is the most effective intervention to achieve changes in walking speeds, dynamic balance, and daily walking, and whether participant characteristics can predict responders and nonresponders. Future research should also evaluate outcomes related to community ambulation and balance, participation, and quality of life. Finally, to study gait and balance effectively in adults with CP, there is a need to establish the psychometrics of measures of gait and balance specifically for this population.

Back to Top | Article Outline


1. Rosenbaum P, Paneth N, Leviton A, et al A report: the definition and classification of cerebral palsy April 2006. Dev Med Child Neurol Suppl. 2007;109:8–14.
2. Haak P, Lenski M, Hidecker MJC, Li M, Paneth N. Cerebral palsy and aging. Dev Med Child Neurol. 2009;51(suppl 4):16–23. doi:10.1111/j.1469-8749.2009.03428.x.
3. Opheim A, Jahnsen R, Olsson E, Stanghelle JK. Walking function, pain, and fatigue in adults with cerebral palsy: a 7-year follow-up study. Dev Med Child Neurol. 2009;51(5):381–388. doi:10.1111/j.1469-8749.2008.03250.x.
4. Morgan P, McGinley J. Gait function and decline in adults with cerebral palsy: a systematic review. Disabil Rehabil. 2014;36(1):1–9. doi:10.3109/09638288.2013.775359.
5. Bar-Haim S, Al-Jarrah MD, Nammourah I, Harries N. Mechanical efficiency and balance in adolescents and young adults with cerebral palsy. Gait Posture. 2013;38(4):668–673. doi:10.1016/j.gaitpost.2013.02.018.
6. Kamp FA, Lennon N, Holmes L, Dallmeijer AJ, Henley J, Miller F. Energy cost of walking in children with spastic cerebral palsy: relationship with age, body composition and mobility capacity. Gait Posture. 2014;40(1):209–214. doi:10.1016/j.gaitpost.2014.03.187.
7. Böhm H, Döderlein L. Gait asymmetries in children with cerebral palsy: do they deteriorate with running? Gait Posture. 2012;35(2):322–327. doi:10.1016/j.gaitpost.2011.10.003.
8. Morgan P, Murphy A, Opheim A, Pogrebnoy D, Kravtsov S, McGinley J. The safety and feasibility of an intervention to improve balance dysfunction in ambulant adults with cerebral palsy: a pilot randomized controlled trial. Clin Rehabil. 2014;29(9):907–919. doi:10.1177/0269215514556299.
9. Jeglinsky I, Surakka J, Carlberg EB, Autti-Rämö I. Evidence on physiotherapeutic interventions for adults with cerebral palsy is sparse. A systematic review. Clin Rehabil. 2010;24(9):771–788. doi:10.1177/0269215510367969.
10. Helm EE, Reisman DS. The split-belt walking paradigm: exploring motor learning and spatiotemporal asymmetry poststroke. Phys Med Rehabil Clin N Am. 2015;26(4):703–713. doi:10.1016/j.pmr.2015.06.010.
11. Kim O-Y, Shin Y-K, Yoon YK, Ko EJ, Cho S-R. The effect of treadmill exercise on gait efficiency during overground walking in adults with cerebral palsy. Ann Rehabil Med. 2015;39(1):25–31. doi:10.5535/arm.2015.39.1.25.
12. Lewek MD, Feasel J, Wentz E, Brooks FP, Whitton MC. Use of visual and proprioceptive feedback to improve gait speed and spatiotemporal symmetry following chronic stroke: a case series. Phys Ther. 2012;92(5):748–756. doi:10.2522/ptj.20110206.
13. Wutzke CJ, Faldowski RA, Lewek MD. Individuals poststroke do not perceive their spatiotemporal gait asymmetries as abnormal. Phys Ther. 2015;95(9):1244–1253. doi:10.2522/ptj.20140482.
14. Wulf G, Höß M, Prinz W. Instructions for motor learning: differential effects of internal versus external focus of attention. J Mot Behav. 1998;30(2):169–179. doi:10.1080/00222899809601334.
15. Lamontagne A, Fung J, McFadyen BJ, Faubert J. Modulation of walking speed by changing optic flow in persons with stroke. J Neuroengineering Rehabil. 2007;4:22. doi:10.1186/1743-0003-4-22.
16. Reisman DS, McLean H, Keller J, Danks KA, Bastian AJ. Repeated split-belt treadmill training improves poststroke step length asymmetry. Neurorehabil Neural Repair. 2013;27(5):460–468. doi:10.1177/1545968312474118.
17. Feasel J, Whitton MC, Kassler L, Brooks FP, Lewek MD. The integrated virtual environment rehabilitation treadmill system. IEEE Trans Neural Syst Rehabil Eng Publ IEEE Eng Med Biol Soc. 2011;19(3):290–297. doi:10.1109/TNSRE.2011.2120623.
18. de Mello Monteiro CB, Massetti T, da Silva TD, et al Transfer of motor learning from virtual to natural environments in individuals with cerebral palsy. Res Dev Disabil. 2014;35(10):2430–2437. doi:10.1016/j.ridd.2014.06.006.
19. Lewek MD, Randall EP. Reliability of spatiotemporal asymmetry during overground walking for individuals following chronic stroke. J Neurol Phys Ther. 2011;35(3):116–121. doi:10.1097/NPT.0b013e318227fe70.
20. Slaman J, Dallmeijer A, Stam H, et al The six-minute walk test cannot predict peak cardiopulmonary fitness in ambulatory adolescents and young adults with cerebral palsy. Arch Phys Med Rehabil. 2013;94(11):2227–2233. doi:10.1016/j.apmr.2013.05.023.
21. Maltais DB, Robitaille N-M, Dumas F, Boucher N, Richards CL. Measuring steady-state oxygen uptake during the 6-min walk test in adults with cerebral palsy: feasibility and construct validity. Int J Rehabil. 2012;35(2):181–183. doi:10.1097/MRR.0b013e3283527792.
    22. Dite W, Temple VA. A clinical test of stepping and change of direction to identify multiple falling older adults. Arch Phys Med Rehabil. 2002;83(11):1566–1571.
    23. Goh EY, Chua SY, Hong S-J, Ng SS. Reliability and concurrent validity of Four Square Step Test scores in participants with chronic stroke: a pilot study. Arch Phys Med Rehabil. 2013;94(7):1306–1311. doi:10.1016/j.apmr.2013.01.027.
    24. Bjornson KF, Zhou C, Stevenson R, Christakis D, Song K. Walking activity patterns in youth with cerebral palsy and youth developing typically. Disabil Rehabil. 2014;36(15):1279–1284. doi:10.3109/09638288.2013.845254.
    25. Andersson C, Asztalos L, Mattsson E. Six-minute walk test in adults with cerebral palsy. A study of reliability. Clin Rehabil. 2006;20(6):488–495.
    26. Darter BJ, Rodriguez KM, Wilken JM. Test-retest reliability and minimum detectable change using the K4b2: oxygen consumption, gait efficiency, and heart rate for healthy adults during submaximal walking. Res Q Exerc Sport. 2013;84(2):223–231. doi:10.1080/02701367.2013.784720.
    27. Blennerhassett JM, Jayalath VM. The Four Square Step Test is a feasible and valid clinical test of dynamic standing balance for use in ambulant people poststroke. Arch Phys Med Rehabil. 2008;89(11):2156–2161. doi:10.1016/j.apmr.2008.05.012.
    28. Malone LA, Bastian AJ. Age-related forgetting in locomotor adaptation. Neurobiol Learn Mem. 2016;128:1–6. doi:10.1016/j.nlm.2015.11.003.
    29. Schmidt R, Lee TD. Motor Control and Learning: A Behavioral Emphasis. 4th ed. Champaign, IL: Human Kinetics; 2005.
    30. Bonnechère B, Jansen B, Omelina L, et al Can serious games be incorporated with conventional treatment of children with cerebral palsy? A review. Res Dev Disabil. 2014;35(8):1899–1913. doi:10.1016/j.ridd.2014.04.016.

    cerebral palsy; gait asymmetry; gait training; virtual environment

    Copyright © 2017 Wolters Kluwer Health, Inc. and Section on Pediatrics of the American Physical Therapy Association. All rights reserved.