Can Strength Training PredictablyImprove Gait Kinematics? A PilotStudy on the Effects of Hip and KneeExtensor Strengthening on Lower-Extremity Alignment in Cerebral PalsyDiane L. Damiano, Allison S. Arnold, Katherine M. Steele, Scott L. Delp

Background. Computer simulations have demonstrated that excessive hip andknee flexion during gait, as frequently seen in ambulatory children with cerebralpalsy (CP), can reduce the ability of muscles to provide antigravity support andincrease the tendency of hip muscles to internally rotate the thigh. These findingssuggest that therapies for improving upright posture during gait also may reduceexcessive internal rotation.

Objective. The goal of this study was to determine whether strength training candiminish the degree of crouched, internally rotated gait in children with spasticdiplegic CP.

Design. This was a pilot prospective clinical trial.

Methods. Eight children with CP participated in an 8-week progressive resistanceexercise program, with 3-dimensional gait analysis and isokinetic testing performedbefore and after the program. Secondary measures included passive range of motion,the Ashworth Scale, and the PedsQL CP Module. To identify factors that may haveinfluenced outcome, individual and subgroup data were examined for patterns ofchange within and across variables.

Results. Strength (force-generating capacity) increased significantly in the left hipextensors, with smaller, nonsignificant mean increases in the other 3 extensor musclegroups, yet kinematic and functional outcomes were inconsistent. The first reportedsubject-specific computer simulations of crouch gait were created for one child whoshowed substantial benefit to examine the factors that may have contributed to thisoutcome.

Limitations. The sample was small, with wide variability in outcomes.

Conclusions. Strength training may improve walking function and alignment insome patients for whom weakness is a major contributor to their gait deficits.However, in other patients, it may produce no change or even undesired outcomes.Given the variability of outcomes in this and other strengthening studies in CP,analytical approaches to determine the sources of variability are needed to betteridentify those individuals who are most likely to benefit from strengthening.

D.L. Damiano, PT, PhD, is Chief,Functional and Applied Biome-chanics Section, RehabilitationMedicine Department, NIH Clini-cal Center, Bldg 10, Room1–1469, Bethesda, MD 20892(USA). Address all correspondenceto Dr Damiano at: damianod@cc.nih.gov.

A.S. Arnold, PhD, is Research As-sociate, Concord Field Station,Harvard University, Bedford,Massachusetts.

K.M. Steele, MS, is a doctoral stu-dent in the Department of Me-chanical Engineering, StanfordUniversity School of Engineering,Stanford, California.

S.L. Delp, PhD, is Professor of Bio-engineering and Mechanical Engi-neering, Stanford UniversitySchools of Engineering andMedicine.

[Damiano DL, Arnold AS, SteeleKM, Delp SL. Can strength train-ing predictably improve gait kine-matics? A pilot study on the effectsof hip and knee extensor strength-ening on lower-extremity align-ment in cerebral palsy. Phys Ther.2010;90:269–279.]

© 2010 American Physical TherapyAssociation

Perry Issue:Gait Rehab

Post a Rapid Response orfind The Bottom Line:www.ptjournal.org

February 2010 Volume 90 Number 2 Physical Therapy f 269

The long-term goal of this re-search collaboration is to de-termine whether targeted

strengthening of the hip and kneeextensors, alone or in combinationwith other treatments, can effec-tively diminish the excessive kneeflexion, hip flexion, and hip internalrotation commonly observed in chil-dren with spastic diplegia who walkwith a crouched gait pattern. Cere-bral palsy (CP) is the most prevalentphysical disability originating inchildhood, with the largest propor-tion of this patient population havingspastic diplegia,1 characterized by in-volvement primarily in the lower ex-tremities. Nearly all children withspastic diplegia will ambulate, al-though often at a later age and withgreater hip and knee flexion thanchildren without neuromotor dis-abilities.2 Much therapeutic effort isdirected at promoting and maintain-ing upright ambulation in these chil-

dren, including regular physical ther-apy and bracing, as well as moreinvasive methods of spasticity reduc-tion (eg, intramuscular injections, or-thopedic surgery). Unfortunately,the outcomes of strength trainingand of many of the other interven-tions for improving walking in theseindividuals remain inconsistent.

There are several reasons to hypoth-esize that strengthening the hip andknee extensors might improve thegait kinematics of children with spas-tic diplegia. Perry’s pioneering workdemonstrated electromyographic(EMG) activity of the hip and kneeextensors during the early stancephase of normal gait,3 and computer-based simulations of walking dynam-ics have revealed that the gluteusmaximus and vastus muscles play animportant role in early stance by sup-porting body weight and controllinghip and knee extension.4 (For a re-view of principles associated withhuman locomotion, refer to Kuo andDonelan in this issue.5) Simulationsanalyzing the roles of these musclesduring crouch gait have revealedthat the capacity of gluteus maximusand other muscles to extend the hipand knee may be diminished sub-stantially by hip and knee flexion.6

The presence of weakness in chil-dren with spastic diplegia is nowwell documented, with even themost functional individuals demon-strating substantial generalized mus-cle weakness.7,8 Together, these ob-servations suggest that the crouchedpostures of some children with CPmay be exacerbated by lower-extremity extensor weakness. Mus-culoskeletal modeling and anatomi-cal studies also have shown thatexcessive hip flexion alters the bal-ance of the muscles that rotate thehip. With excessive hip flexion, themoment arms (ie, the lever arms ormechanical advantage) of some hipexternal rotators are diminished andthe moment arms of hip internal ro-tators are increased, potentially con-

tributing to excessive hip internal ro-tation.9 For example, the capacity ofthe gluteus maximus muscle to gen-erate external rotation of the hip iscompromised when the hip isflexed. This finding suggests that im-provement of excessive hip flexionin people with crouch gait might im-prove hip rotation.

Despite this rationale, the potentialfor strength training to improve thegait mechanics of people with spas-tic diplegia remains unclear.Strength training was briefly a part ofearly physical therapy managementof CP and has experienced a resur-gence over the past 10 to 15 years.This resurgence was preceded by de-cades during which resistance train-ing was contraindicated for peoplewith CP because of clinical concernsthat it would exacerbate spasticity.Some studies have demonstratedthat short-term progressive resis-tance exercise programs can safelyincrease strength (force-generatingcapacity) in targeted muscles in peo-ple with CP without increasing spas-ticity.10,11 Positive effects on gait andgross motor function also have beenreported,12 although less consis-tently than gains in strength.13 A re-cent meta-analysis, however, failedto provide evidence that strengthtraining is effective in improvingstrength, increasing gait speed, orproducing a clinically meaningfulchange in gross motor function inpeople with CP.14 The potential forstrength training to improve the gaitkinematics of people with CP hasbeen evaluated in only a few studies,and findings have been equivocal.Most of these studies averaged dataover a small number of subjects (ie,11–21 subjects) and revealed onlymodest mean changes in the subjects’joint angles during walking.15–19

The variability in outcomes reportedin previous studies may be due to arange of factors, including method-ological factors that may have lim-

Available WithThis Article atptjournal.apta.org

• eTable 1: Participants Orderedby Decreasing Change in KneeExtensor Torque

• eTable 2: Mean Results and PValues for Selected Gait MeasuresBefore and After Strengthening

• eTable 3: Mean Changes inPedsQL Cerebral Palsy ModuleChild Questionnaire Scores Beforeand After Strengthening

• Video: In honor of Dr JacquelinPerry, view art by patients fromRancho Los Amigos NationalRehabilitation Center.

• Podcast: “Stepping Forward WithGait Rehabilitation” symposiumrecorded at APTA CombinedSections Meeting, San Diego.

• Audio Abstracts Podcast

This article was published ahead ofprint on December 18, 2009, atptjournal.apta.org.

Strength Training and Gait Kinematics

270 f Physical Therapy Volume 90 Number 2 February 2010

ited the effectiveness of the training(eg, insufficient muscle loads, insuf-ficient training durations) or, insome cases, impairments other thanweakness that may have limited thesubjects’ functional gains followingstrengthening (eg, difficulties withbalance). We believe new ap-proaches are needed to rigorouslyevaluate the biomechanical effects ofresistance exercise programs, tohelp explain the reasons for incon-sistent outcomes across subjects,and to identify individuals who aremost likely to benefit fromstrengthening.

This article presents the results froma pilot strength training study inwhich we used a combination ofphysical examination, gait analysis,and state-of-the-art computer simula-tion to evaluate the effects of an8-week progressive resistance exer-cise program on participants’ hipand knee angles (kinematics) duringwalking. We hypothesized thatstrengthening the hip and knee ex-tensor muscles would decrease theexaggerated hip and knee flexionand hip internal rotation of childrenwith spastic diplegia during thestance phase of gait. Secondary hy-potheses were that strengtheningwould improve temporal spatial gaitparameters, such as walking speedand stride length, and lead to im-proved physical functioning and re-lated quality of life. Our main objec-tive was to assess whether strengthtraining could improve lower-extremity alignment during walkingand, if so, use simulation to providebiomechanical insights into the rea-sons for these improvements.

MethodParticipantsThe goal of this pilot study was toenroll 10 children with spastic diple-gia, based on a power analysis usingdata from 2 previous studies,15,17

within the age range of 5 to 17 yearsand diagnosed at Gross Motor Func-

tional Classification System (GMFCS)levels I to III. The children had to bemore than 1 year postsurgery and 6months post–botulinum toxin injec-tions in the lower extremity. Addi-tional inclusion criteria, based onphysical examination, included bilat-eral passive hip extension to neutralwith the other hip flexed to 90 de-grees, passive knee extension within5 degrees of full extension while po-sitioned supine, and passive hip ex-ternal rotation of at least 20 degreesas assessed in a prone position withhip extended and knees flexed to 90degrees. These criteria were chosenin an effort to exclude childrenwhose gait deviations were predom-inantly constrained by musculoskel-etal contracture. Previous orthope-dic surgery or neurosurgery werenot considered a reason for exclu-sion, with the exception of previousrotational osteotomies.

Children were screened visually todetermine whether they exhibited acrouched internal rotation gait pat-tern, and these kinematic criteriawere confirmed by gait analysis. Inparticular, children had to satisfy thefollowing requirements bilaterallyprior to participating in the strength-ening portion of the study: (1) hipand knee flexion greater than 1 stan-dard deviation above the mean nor-mative value at initial contact; (2)excessive adduction and internal ro-tation of the hip at mid stance; and(3) less than 20 degrees of ankle dor-siflexion at mid stance, because ex-cessive weakness of the plantar flex-ors or overlengthening of the calfmuscles alone may contribute tocrouch gait.

Participants were recruited from theneurology and neurosurgery cere-bral palsy clinics and the physicaltherapy clinic at St. Louis Children’sHospital associated with WashingtonUniversity. Informed consent and pe-diatric assent were obtained from all

participants prior to the initialassessment.

Eight children (3 male, 5 female) metthe inclusion criteria and completedthe 8-week strengthening programduring the study period (eTab. 1available at ptjournal.apta.org). Thechildren ranged in age from 5.5 to13.4 years. Five children were classi-fied at GMFCS level III and used anassistive device to walk, 2 childrenwere classified at GMFCS level II, and1 child was classified at GMFCS level I.

Strength Training ProgramEach child participated one-on-onewith a therapist in a community-based physical therapy program, at-tending three 1-hour sessions perweek for 8 weeks. Each therapist-guided session consisted of a pro-gressive resistance exercise programusing free weights or weight ma-chines that targeted the gluteusmaximus and quadriceps muscles bi-laterally, with at least 1 day of restbetween sessions (examples shownin Fig. 1). The program used a com-bination of open- and closed-chainexercises for maximum transfer toboth the stance and swing phases ofgait, and these exercises were de-signed to work the muscles in themost extended portion of the range.Examples include the use of cuffweights attached to the distal thighduring prone hip extension exer-cises with the knee kept flexedthroughout the motion, the use of aresisted leg press, and the use of aweight machine that resisted kneeextension in a reclined sittingposition.

The intensity and difficulty of theprogram were adjusted individually.The amount of resistance appliedwas based on the number of repeti-tions each child could perform be-fore fatiguing; the target was 8 to 10repetitions, with the total number ofrepetitions consistent across partici-pants (30 repetitions per muscle

Strength Training and Gait Kinematics

February 2010 Volume 90 Number 2 Physical Therapy f 271

group). Passive stretching exercisesat the hip and knee and a 5-minutewalk at a relaxed pace on a treadmillor indoor track (depending on thechild’s ability and preference) wereperformed for warm-up before andcool-down after the strengtheningexercises.

Functional AssessmentEach child received an assessmentconsisting of the following: (1) aphysical examination; (2) a 3-dimensional (3-D) gait analysis; (3)an isokinetic strength assessment, asmeasured by maximum hip and kneeextensor concentric torque at 30°/s;and (4) a validated self-report mea-sure of physical functioning and re-lated quality of life, as quantified bythe parent-proxy version of thePedsQL 3.0 Cerebral Palsy Module.20

The exercise program was startedwithin a week after the initial assess-ment, and an identical assessmentwas conducted within a week ofcompleting the program. The physi-cal examination was performed toassess range of motion at the hip,knee, and ankle (to ensure that thechild met the inclusion and exclu-sion criteria) and to assess spasticityin the hamstring and quadricepsmuscles. Spasticity was measured us-ing the Ashworth Scale (rangingfrom 0 to 4, with a score of 1 indi-

cating normal tone [resistance topassive stretch] and a score of 0 in-dicating less-than-normal tone). The3-D gait analysis was performed todetermine the child’s lower-extremity joint angles during walk-ing, with data from 5 trials collectedat both freely selected and “as fast aspossible without running” speeds.Fifteen retroreflective markers wereplaced on the skin overlying specificanatomic locations on the pelvis andbilateral lower extremities. The 3-Dlocations of these markers weretracked using an 8-camera Vicon sys-tem,* and the data were processedusing Plug-In-Gait.† The isokineticstrength assessment enabled com-parison of each child’s isokineticpeak torque before and afterstrengthening. The peak torque datawere expressed as the actual valueand also were divided by bodyweight and multiplied by 100 to fa-cilitate comparisons across childrenof varying sizes. Paired t tests (P�.05)were used to evaluate changes in se-lected assessment measures for thechildren before and after strengthen-ing. Due to the small sample and the

pilot nature of this study, no correc-tion for multiple tests was applied.

Subject-Specific Simulations ofCrouch GaitAdditional data were collected onchildren who were independent am-bulators with the aim of creatingsubject-specific computer simula-tions of each child’s crouched gait.To enable dynamic simulations,these data must include “clean” con-secutive forceplate strikes during thegait analysis. These data were avail-able, from both prestrengtheningand poststrengthening training as-sessments, for one individual in thissample. In this case, the gait assess-ments included full-body 3-D kine-matics, ground reaction forces andmoments from forceplates, and sur-face EMG recordings from the tibialisanterior, gastrocnemius, rectus fem-oris, medial hamstring, adductor lon-gus, vastus medialis, gluteus medius,and gluteus maximus muscles.

Computer simulations of the stancephase were generated that repro-duced this child’s 3-D gait kinematicsand kinetics before and after strength-ening. This process provided one ofthe first subject-specific simulations ofan individual with a crouched, internalrotation gait pattern and is the firstsuch model to evaluate changes in the

* Vicon, 7388 S Revere Pkwy, Suite 901, Cen-tennial, CO 80112.† Oxford Metrics, 14 Minns Business Park,West Way, Oxford OX2 0JB, United Kingdom.

Figure 1.Two examples of hip extensor strengthening exercises performed during training using weight machines. Other exercises also wereperformed (see text).

Strength Training and Gait Kinematics

272 f Physical Therapy Volume 90 Number 2 February 2010

actions of muscles as a result ofstrength training. To generate the sim-ulations, we first created a computermodel of the child that included 3-Drepresentations of the bones, joints,and muscles scaled to the child’s an-thropometric dimensions.

Next, we used OpenSim biomechan-ics software21 to estimate the muscleactivation patterns and muscleforces that, when applied to themodel, produced joint angles andground reaction forces that corre-sponded closely to the experimentalmeasurements of the child’s gait ki-nematics and kinetics. This softwareuses an optimization algorithm,called Computed Muscle Control,22

to determine the excitation patternsfor each of the 92 muscle compart-ments in the model. We used thechild’s measured EMG data to verifythat the simulated muscle excita-tions reflected the child’s activationpatterns. We generated 2 simula-tions: one that reproduced thechild’s gait dynamics beforestrengthening and another that re-produced the gait dynamics afterstrengthening.

Lastly, we analyzed the simulationsto identify factors that enabled thisindividual to walk more upright afterstrengthening. In particular, we per-formed a perturbation analysis4 to as-sess the capacity of the gluteus maxi-mus, vastus, and other muscles tosupport the body and extend the hipand knee during the stance phase.This analysis makes small adjust-ments (ie, perturbations) to the forcein each muscle and measures the re-sulting changes in the hip and kneeangles to determine the role thateach muscle plays in extending orflexing the joints throughout themovement.

Figure 2.Change in isokinetic peak torque divided by body weight and multiplied by 100 foreach participant for the (A) right and left hip extensors and (B) right and left kneeextensors. A positive change indicates an increase in strength. (C) Change in minimumknee flexion angle during stance. A positive change indicates an increase in knee flexion(ie, greater crouch).

Strength Training and Gait Kinematics

February 2010 Volume 90 Number 2 Physical Therapy f 273

ResultsEffects of Strength Training onPeak Isokinetic TorqueMost of the participants’ hip andknee extensor muscles were stron-ger after training, as measured by thechange in isokinetic peak torque.The mean percentage gains instrength (ie, change in peak torquedivided by initial peak torque value)ranged from 18.0% for the right kneeextensors to 79.4% for the left hipextensors (Table); however, the per-centage gains varied widely acrossparticipants. The magnitude andconsistency of the strength changesgenerally were greater at the hipthan at the knee (eTab. 2 available atptjournal.apta.org). Four childrenachieved appreciable increases inpeak torque in at least 3 of the 4muscle groups (Fig. 2). The other 4

children responded less dramaticallyto the strengthening program, show-ing either a negligible change or asmall decrease in peak torque in oneor both limbs and joints.

Effects of Strength Training onGait KinematicsSome, but not all, of the childrenwalked with improved hip and kneeextension during stance followingthe exercise program (Fig. 2). Thechanges in hip extension generallywere correlated to the changes inknee extension (r�.75, P�.05). Sev-eral of the children walked with im-proved hip rotation; however, wedid not detect a correlation betweenthe changes in hip rotation and thechanges in either hip or kneeextension.

The 2 children who achieved thegreatest gains in knee extensorstrength (participant 1, classified atGMFCS level I, and participant 2,classified at GMFCS level III) alsoshowed the greatest improvementsin knee extension during the stancephase. However, for the other 6 chil-dren, knee flexion appeared toworsen with increasing strengthgains, so no general conclusionscould be made (Fig. 3).

Effects of Strength Training onGait Temporal-SpatialParametersGait speed, stride length, and ca-dence were not significantlychanged, on average, in either theself-selected or fast speed conditionsfollowing the strengthening program(eTab. 2 available at ptjournal.apta.org). However, changes in all 3measures varied across subjects; forexample, self-selected walking speedimproved 18% for participant 7, butdiminished 25% for participant 5. Nosignificant correlations were foundbetween change in strength andchange in these temporal-spatialparameters.

Effects of Strength Training onPerceived Physical Function andRelated Quality of LifeA secondary outcome measure wasthe parent-proxy version of thePedsQL 3.0 Cerebral Palsy Module.

Table.Mean (SD) Before and After Strengthening, Mean Difference and Standard Error (SE) of the Difference, 95% Confidence Interval(CI) of the Difference, Percent Change, and P Values for Isokinetic Strength Tests (N�8)a

BeforeStrengthening

Mean (SD)

AfterStrengthening

Mean (SD)Mean

Difference (SE)95% CI of the

DifferencePercentChange P

Peak concentric torque at30°/s (ft-lb)

Right hip extensors 12.7 (6.9) 17.2 (7.0) 4.5 (2.8) �2.21 to 11.0 35.4 .16

Left hip extensors 10.7 (4.9) 19.2 (8.6) 8.5 (2.5) 2.70 to 14.4 79.4 .01

Right knee extensors 13.9 (5.5) 16.4 (7.0) 2.6 (1.2) �0.20 to 5.34 18.0 .06

Left knee extensors 13.1 (6.3) 16.8 (9.3) 3.7 (1.6) �0.22 to 7.56 28.2 .06

a Values significant at P�.05 shown in bold.

Figure 3.Scatterplot of change in minimum knee flexion angle (average for right and left legs)against (A) Ashworth Scale score and (B) change in knee extensor strength. Positivecorrelation between change in minimum knee flexion angle during stance and Ash-worth Scale score of the hamstring muscles. Note that the Ashworth Scale score was notcollected for one participant.

Strength Training and Gait Kinematics

274 f Physical Therapy Volume 90 Number 2 February 2010

These data showed trends toward asmall amount of improvement (eTab.3 available at ptjournal.apta.org), butnone of the changes were statisti-cally significant when averagedacross the participants.

Secondary Measures ofImpairmentMean passive knee extension was�2 degrees at the start of the pro-gram. A small, but significant, de-crease (increased tightness) in pas-sive knee extension was found as aresult of the training (P�.05 for bothright and left sides), with a mean

change of about 3 degrees. The pop-liteal angle also worsened slightly,but the change was not significanton either side. No significant corre-lation was found between changes inpassive and active knee extension.Spasticity did not change signifi-cantly before and after the training,but the degree of spasticity was di-rectly related to the change in kneeposition (r�.55, P�.05). Figure 3shows the initial Ashworth Scalescore for each leg plotted against thechange in minimum knee flexion, in-dicating that the greater the spastic-

ity, the worse the response tostrengthening.

Analysis of the Subject-SpecificSimulationsParticipant 1 responded well to thestrengthening program, demonstrat-ing notable improvements in knee ex-tension, hip extension, and hip rota-tion throughout the gait cycle (Fig. 4).Analysis of the subject-specificcomputer simulations and the associ-ated clinical data provided insightsinto why this child’s gait improved.

Figure 4.Joint angles (with standard deviation bands), obtained from gait analysis, from the left side of one participant who responded wellto the strengthening program (one for whom computer simulation was created). Note that the participant’s excessive knee flexion,hip flexion, and internal hip rotation were diminished after strengthening. Pre�before strengthening, Post�after strengthening,Flex�flexion, Ext�extension, Dorsi�dorsiflexion, Plantar�plantar flexion.

Strength Training and Gait Kinematics

February 2010 Volume 90 Number 2 Physical Therapy f 275

Prior to the strengthening program,participant 1 had lower than normalstrength of the hip and knee exten-sors, with the left side slightlyweaker than the right side, and hewalked in a moderate crouch gaitpattern. As reported in previousstudies, this posture puts greaterthan normal demands on the vastusmuscles to support the body weightand diminishes the potential of thegluteus maximus and vastus musclesto extend the hip and knee.4,6,23 Al-though this child walked with exces-sive knee flexion, analysis of thecomputer simulation revealed thathis hamstring muscles were operat-ing at muscle-tendon lengths suffi-cient for normal walking,24 so a sur-gical lengthening of the hamstringmuscles was not needed to releaseexcessively tight muscles. He hadnormal range of motion at the hipand knee, with popliteal angles of126 and 132 degrees on the right andleft sides, respectively. He had un-dergone a selective dorsal rhizotomyat age 5 years and exhibited goodselective control, as demonstratedby the ability to dorsiflex bilaterally

without evidence of hip flexion.Thus, for this participant, an increasein strength was sufficient to improvehis gait, and the resulting functionalgains were not limited by other con-founding impairments.

Our analysis of the child’s gait dy-namics supports this explanation. Af-ter the strengthening program, hedemonstrated a substantial increase(from 20% to 85% of peak isometrictorque) in the strength of his hip andknee extensors, with the left extrem-ity values now exceeding those ofthe right extremity. The simulationssuggested that the force generatedby the left gluteus maximus muscleduring early stance was increased,restoring an important mechanismthat promotes hip and knee exten-sion in normal gait. Furthermore, thecapacity for the gluteus maximusand vastus muscles to accelerate thehip and knee into extension was in-creased (ie, angular acceleration perunit muscle force) (Fig. 5). Thus, af-ter strengthening, the child’s exces-sive knee flexion was diminished,which reduced the demands on the

vastus muscles and increased the ca-pacity of the gluteus maximus, vas-tus, and other muscles to supportthe body weight and to produce hipand knee extension. The decrease inhip flexion in stance also shifted themoment arms of the gluteus maxi-mus, gluteus medius, and other hipmuscles toward external rotation;this may have contributed to this in-dividual’s small improvement in hiprotation.

DiscussionThe goal of our pilot strengtheningprogram, guided by previous studiesand insights from musculoskeletalmodels, was to improve the chil-dren’s lower-extremity alignmentduring walking. Unfortunately, theprogram did not produce systematicdecreases in excessive hip flexion orknee flexion during the stancephase. Furthermore, because stance-phase hip extension was improvedin only a few children, we were un-able to test our hypothesis that im-proving hip extension would de-crease hip internal rotation. Wemade an effort to exclude children

Figure 5.(Left) The 3-dimensional musculoskeletal model in poses corresponding to key events in the gait cycle. The musculoskeletal modelused to create a simulation of one of the participants has 10 body segments and 92 muscle compartments. (Right) Capacity of thegluteus maximus and vastus muscles to accelerate the hip and knee into extension before (Pre) and after (Post) strengthening for theparticipant on whom the simulation was performed.

Strength Training and Gait Kinematics

276 f Physical Therapy Volume 90 Number 2 February 2010

whose gait deviations may have beencaused primarily by other factors,such as weak or over-lengthenedplantar flexors, muscle contractures,or bone deformities. Nonetheless, re-sponses to the strength trainingwere variable, similar to previousstudies.15,16,18 The large variability inresponse, combined with the smallsample size, likely affected our abil-ity to detect significant mean groupresponses to strengthening. Al-though the conclusions we can drawfrom analysis of only 8 participantsare limited, examination of these re-sults is informative for clinical prac-tice and for guiding future research.

Consistent with previous studies, weshowed that gains in hip and kneeextensor strength are possible in am-bulatory children with CP, althoughsome children achieved larger gains,and responded differently to thosegains, than others. It is difficult toexplain why some children did notmake appreciable strength gains,given the fairly long (8-week) dura-tion and the design of this program,whereby a therapist closely moni-tored each individual and progres-sively increased the amount of resis-tance used during training. Possibleexplanations include neurologicalfactors, such as a primary agonist in-sufficiency that was not amenable totraining, or pre-existing muscle adap-tations that may have limited the ca-pacity of some muscles to change inresponse to loading. At present,these explanations remain hypothet-ical and warrant further investiga-tion. The children’s strengthchanges were less consistent andsmaller at the knee than at the hip,which may be related to increases inhamstring muscle strength. Greaterstrength in the hamstring muscleswould contribute to increased exten-sion torque at the hip, but could bea potential source of greater antago-nist restraint when producing exten-sion torque at the knee.

Previous studies have shown that re-sistance training, in some cases, canproduce positive effects on grossmotor functional abilities, includingself-selected and maximum walkingspeed.10,13 In the present study, al-though there was a nonsignificanttrend toward improvement on thequestionnaire that measured the chil-dren’s perception of physical perfor-mance and related quality of life, wedid not detect a systematic improve-ment in walking speed or other gaitparameters. This finding is not sur-prising, given the limited response tostrength training for some of the chil-dren, as we suspect that improve-ments in strength may be needed tostimulate changes in these functionalmeasures.

Only a few studies have evaluatedthe effects of increased strength ongait kinematics in people with CP,and findings have been equivocal.For example, Damiano and col-leagues15 examined the effectivenessof a 6-week isotonic quadriceps mus-cle strengthening program for im-proving crouch gait in 14 childrenwith spastic diplegia. Knee extensorstrength improved significantly, asdid stride length at free and fastspeeds. However, the only signifi-cant change in the gait kinematicswas a more extended knee position,on average, at initial contact. Six ofthe children also achieved greaterknee extension at mid stance, but 2children developed a slightly wors-ened crouch gait (ie, knee flexionincreased approximately 5°) (unpub-lished observation).

In a subsequent strength study,Damiano and Abel17 targeted the 2weakest muscles per individual andfound significant changes in walkingspeed and function in a sample of 11children with hemiplegia or diplegia,but no consistent improvement ingait kinematics, which was not sur-prising given the variability in thetarget muscles across subjects. Un-

ger and colleagues16 evaluated theeffects of an 8-week, generalizedlower-extremity strengthening pro-gram for 21 adolescents with CP andreported a significant improvementin their gait kinematics of 5.1 de-grees at mid stance only by summingup changes at the hip, knee, and an-kle. Eek and colleagues18 trained the4 weakest muscles per subject in 16independently ambulatory childrenwith CP and observed improvementsin stride length and hip and anklekinetics, but no systematic change inthe gait kinematics. Lee and col-leagues19 conducted a generalizedfunctional lower-extremity trainingprogram in 16 children, ages 4 to 12years, and found no changes in thegait kinematics.

Previous studies10,12,15–19 generallyhave reported mean changes in gaitmeasures and have provided little in-sight into subject-specific factorsthat may have influenced the func-tional or kinematic outcomes. Onepotential precipitating factor oftenmentioned clinically is growth. Aschildren grow, their strength mustkeep pace, or else their posture iscompromised, as often is apparent innormal adolescence, suggesting thatcrouch in individuals with CP maybe exacerbated by growth. How-ever, Wren and colleagues25 recentlyevaluated the influence of age,among other factors, on differenttypes of malalignment in nearly1,000 patients with CP and foundthat rotational malalignment was as-sociated with advancing age, but in-terestingly, crouch did not emerge asbeing significantly affected by age. Itis likely that the natural history ofcrouch is altered by surgical proce-dures or botulinum toxin injectionson the hamstrings and other mus-cles, which may have affected theiranalyses.

Given the variability of responses ob-served here, the reasons why somechildren in the present study walked

Strength Training and Gait Kinematics

February 2010 Volume 90 Number 2 Physical Therapy f 277

with exaggerated knee flexion orinternal rotation following the pro-gram remain unclear. In the sub-group analysis comparing indepen-dent ambulators with those whoused an assistive device, the childrenwho were more functional showed asmall improvement in comparisonwith an appreciable worsening inthe children who required an assis-tive device to walk; however, thisdifference did not reach significance,and the subgroup sample sizes weresmall. We did find decreased passiveknee extension as a result of the pro-gram, as well as a relationship be-tween greater hamstring musclespasticity and poorer kinematic out-comes. We believe that unavoidablestrengthening of the hamstring mus-cles during the hip extensor trainingmay be responsible for the decreasein active and passive knee extension,despite the greater extensorstrength, with those children withgreater hamstring muscle spasticityat greater risk. More stringent mus-cle length or spasticity criteria, aswell as closer monitoring of ham-string muscle length during the inter-vention, may be warranted in futurestudies. Interestingly, the child whohad the most positive outcome wasan independent ambulator and hadundergone a selective dorsal rhizot-omy several years prior to the study.

In their review of the role of strengthtraining for improving gait in ambu-latory children and adolescents withCP, Mockford and Caulton13 notedthat the wide range of gait-relatedoutcomes within and across studiesoccurs because individuals haveshown such a wide variety of abnor-malities and compensations; there-fore, they concluded that no generalconclusions can be drawn. Indeed,the heterogeneity of the motor dis-order in people with CP makes itdifficult to predict outcomes of anyintervention in this population. Onestudy in the adult orthopedic litera-ture supports the premise that an

8-week strengthening program—inthis case, targeting the hip abductorsand lateral rotators in 15 female run-ners who were healthy—can alterkinematics in a clinically significantand predictable way.26 Based on ourown observations and data from theliterature, we remain cautiously op-timistic about the potential forstrength training to improve walkingin some patients. One hypothesisthat emerges from these results isthat strengthening may be more ben-eficial for children with milder in-volvement, which seems reasonablebecause those patients are likely tohave less spasticity and fewer motorcontrol deficits that could limit theirresponse to strengthening. It also ispossible that programs need to be oflonger duration or coupled withother interventions to address all ofthe factors that potentially contrib-ute to excessive hip flexion, kneeflexion, and hip internal rotation dur-ing walking.

ConclusionThe effect of strength training ongait kinematics in people with CPremains unpredictable at the level ofthe individual patient, and the cur-rent approach of mean group analy-sis of heterogeneous samples willnot resolve this dilemma. We con-tend that studies with larger samplesthat can more adequately examinepatient and intervention factors thatcould influence outcome, large-scaleregression analyses, or subject-specific computer simulations ofwalking are needed to help explainthe variability in outcomes and toidentify individuals who are mostlikely to benefit from strengthening.More research also is needed to de-termine how strength training mightcomplement or interact with othertreatments, especially those inter-ventions that tend to further weakenpatients. Subject-specific musculo-skeletal models and simulations ofwalking are currently labor- andtechnology-intensive efforts, but

they hold promise for identifying theimpairments that impede locomotorfunction, thereby refining and im-proving treatment approaches in CP.

Dr Damiano was the principal investigatorand project director of the clinical trial. DrArnold and Dr Delp provided concept/idea/research design and project management.Dr Arnold, Ms Steele, and Dr Delp providedwriting and data analysis. Ms Steele pro-vided consultation (including review ofmanuscript before submission).

The authors acknowledge the United Cere-bral Palsy Research & Education Foundationfor funding this project. This research alsowas supported, in part, by the intramuralprogram at the National Institutes of Health.The authors thank Julie Anderson for her as-sistance in all aspects of the study and TomNuzum and Dave Reddy for conducting thetraining sessions for all participants.

Human subjects approval was obtained fromthe Washington University Human StudiesCommittee prior to study initiation.

This article was received February 23, 2009,and was accepted September 21, 2009.

DOI: 10.2522/ptj.20090062

References1 Hirtz D, Thurman DJ, Gwinn-Hardy K,

et al. How common are the “common”neurologic disorders? Neurology. 2007;68:326–337.

2 Sussman M, ed. The Diplegic Child Evalu-ation and Management. Rosemont, IL:American Academy of Orthopaedic Sur-geons; 1992.

3 Perry J Determinants of muscle function inthe spastic lower extremity. Clin OrthopRelat Res. 1993;(288):10–26.

4 Arnold AS, Anderson FC, Pandy MG, DelpSL. Muscular contributions to hip andknee extension during the single limbstance phase of normal gait: a frameworkfor investigating the causes of crouch gait.J Biomech. 2005;38:2181–2189.

5 Kuo AD, Donelan JM. Dynamic principlesof gait and their clinical implications. PhysTher. 2010;90:157–174.

6 Hicks JL, Schwartz MH, Arnold AS, DelpSL. Crouched postures reduce the capac-ity of muscles to extend the hip and kneeduring the single-limb stance phase of gait.J Biomech. 2008;41:960–967.

7 Damiano DL, Vaughan, CL, Abel MF. Mus-cle response to heavy resistance exercisein spastic cerebral palsy. Dev Med ChildNeurol. 1995;37:731–739.

8 Wiley ME, Damiano DL. Lower extremitystrength profiles in spastic cerebral palsy.Dev Med Child Neurol. 1998;40:100–107.

Strength Training and Gait Kinematics

278 f Physical Therapy Volume 90 Number 2 February 2010

9 Delp SL, Hess WE, Hungerford DS, JonesLC. Variation of rotation moment armswith hip flexion. J Biomech. 1999;32:493–501.

10 Andersson C, Grooten W, Hellsten M,et al. Adults with cerebral palsy: walkingability after progressive strength training.Dev Med Child Neurol. 2003;45:220–228.

11 Fowler EG, Ho TW, Nwigwe AI, Dorey FJ.The effect of quadriceps femoris musclestrengthening exercises on spasticity inchildren with cerebral palsy. Phys Ther.2001;81:1215–1223.

12 Dodd KJ, Taylor NF, Damiano DL. A sys-tematic review of the effectiveness ofstrength-training programs for peoplewith cerebral palsy. Arch Phys Med Reha-bil. 2002;83:1157–1164.

13 Mockford M, Caulton JM. Systematic re-view of progressive strength training inchildren and adolescents with cerebralpalsy who are ambulatory. Pediatr PhysTher. 2008;20:318–333.

14 Scianni A, Butler JM, Ada L, Teixeira-Salmela LF. Muscle strengthening is noteffective in children and adolescents withcerebral palsy: a systematic review. Aust JPhysiother. 2009;55:81–87.

15 Damiano DL, Kelly LE, Vaughn CL. Effectsof quadriceps femoris muscle strengthen-ing on crouch gait in children with spasticdiplegia. Phys Ther. 1995;75:658–667.

16 Unger M, Faure M, Frieg A. Strength train-ing in adolescent learners with cerebralpalsy. Clin Rehabil. 2006;20:469–477.

17 Damiano DL, Abel MF. Functional out-comes of strength training in spastic cere-bral palsy. Arch Phys Med Rehabil. 1998;79:119–125.

18 Eek MN, Tranberg R, Zugner R, et al. Mus-cle strength training to improve gait func-tion in children with cerebral palsy. DevMed Child Neurol. 2008;50:759–764.

19 Lee JH, Sung IY, Yoo JY. Therapeutic ef-fects of strengthening on gait function incerebral palsy. Disabil Rehab. 2008;30:1439–1444.

20 Varni JW, Burwinkle TM, Berrin SJ, et al.The PedsQL in pediatric cerebral palsy:reliability, validity, and sensitivity of theGeneric Core Scales and Cerebral PalsyModule. Dev Med Child Neurol. 2006;48:442–449.

21 Delp SL, Anderson FC, Arnold AS, et al.OpenSim: open-source software to createand analyze dynamic simulations of move-ment. IEEE Trans Biomed Eng. 2007;55:1940–1950.

22 Thelen DG, Anderson FC, Delp SL. Gener-ating dynamic simulations of movementusing computed muscle control. J Bio-mech. 2003;36:321–328.

23 Liu MQ, Anderson FC, Pandy MG, Delp SL.Muscles that support the body also modu-late forward progression during walking.J Biomech. 2006;39:2623–2630.

24 Arnold AS, Asakawa DJ, Delp SL. Do thehamstrings and the adductors contributeto excessive internal rotation of the hip inpersons with cerebral palsy? Gait Posture.2000;11:181–190.

25 Wren TA, Rethlefsen S, Kay RM. Preva-lence of specific gait abnormalities in chil-dren with cerebral palsy: influence of ce-rebral palsy subtype, age, and previoussurgery. J Pediatr Orthop. 2005;25:79–83.

26 Snyder KR, Earl JE, O’Connor KM, Eber-sole KT. Resistance training is accompa-nied by increases in hip strength andchanges in lower extremity biomechanicsduring running. Clin Biomech (Bristol,Avon). 2009;24:26–34.

Strength Training and Gait Kinematics

February 2010 Volume 90 Number 2 Physical Therapy f 279