Gait and Posture 21 (2005) 311–317

The heel-contact gait pattern of habitual toe walkers

P. Crennaa,∗, E. Fedrizzib, E. Andreuccib, C. Frigoc, R. Bonob

a Laboratory for Movement Analysis in Children (LAMB) P. & L. Mariani, Institute of Human Physiology I,School of Medicine, University of Milan, via Mangiagalli 32, I-20133 Milan, Italy

b Department of Developmental Neurology, Neurological Institute “C. Besta”, Milan, Italyc Department of Bioengineering, Polytechnic of Milan, Milan, Italy

Received 15 September 2003; accepted 25 March 2004

Abstract

We used kinematic, kinetic and EMG analysis to compare the spontaneous heel-contact gait patterns of 13 children classified as habitualtoe walkers (HTWs) and age-matched controls. In the HTWs, the incidence of spontaneous heel-contact strides during a single recordingsession ranged from 15% to 92%, with no correlation with age, passive ankle joint excursion, walking speed and trial order. Hallmarks ofthe heel-contact strides were premature heel-rise, reversal of the second rocker, relative shortening of the loading response and anticipationand enhancement of the electromyographic (EMG) activity normally observed in the triceps surae (TS) during the first half of the stancephase. This variant of the locomotor program is different from the walking patterns observed in normally developing toddlers and childrenwith cerebral palsy (CP). It does not necessarily reflect a functional adaptation to changes in the rheological properties of the muscle–tendoncomplex.© 2004 Elsevier B.V. All rights reserved.

Keywords: Habitual toe walking; Idiopathic toe walking; Ontogeny of gait; Cerebral palsy

1. Introduction

A gait pattern marked by a forefoot or tip-toe weight-bearing attitude during the stance phase of the stride cyclecan be observed in children during early independent walk-ing but within a few months this is normally replaced by thetypical mature support pattern, the body weight being pro-gressively loaded onto the heel, lateral sole, forefoot and toes[1,2]. Some children, however, keep walking on their toes,with no clinical signs of sensory-motor impairment or mus-cular disease, and this is termed idiopathic toe walking[3,4].

In a small proportion of children toe walking is persis-tent although they can usually produce a heel-contact gaitif asked to do so[5–7]. A primary increase in the passivestiffness of the calf muscles has been proposed as a cause[5,8,9]. In most cases, however, toe walking is facultative andhas been described as habitual toe walking (HTW)[10]. InHTW the range of passive ankle dorsiflexion is only slightlylimited or even normal[4,11,12]. Different etiologies have

∗ Corresponding author. Tel.:+39-02-503-15428;fax: +39-02-50315430.

E-mail address: paolo.crenna@unimi.it (P. Crenna).

been suggested for HTW, including clinically undetectableneural impairments and/or anomalous persistence of the im-mature digitigrade gait, with possible secondary changes inthe passive muscle–tendon properties[3,13,14].

HTW is not associated with any one single locomotor pat-tern and its definition largely relies on clinical exclusion cri-teria [3,4]. A specific kinematic changes such as increasedfoot plantar-flexion and reduced knee flexion at touchdown,or greater excursion of the ankle joint angle have been ob-served in HTW as compared to the equinus gait of childrenwith cerebral palsy (CP)[14,15]. Similarly, the electromyo-graphic (EMG) profile in HTW, which includes abnormalco-contraction of calf and pretibial muscles in stance andactivation of triceps surae (TS) in late swing[16,17], wasalso found in diplegic children[18,19]and even in normallydeveloping children during early independent walking[20].Several studies, moreover, have reported that the kinematicand EMG features of HTW can be reproduced in normalchildren, when asked to walk on their toes[10,15,21].

Conclusions about the lack of specificity of HTW arebased mainly on analysis of selected stride cycles with noconsideration for the potentially different gait patterns ofindividual children. With the exception of a brief mention

0966-6362/$ – see front matter © 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.gaitpost.2004.03.005

312 P. Crenna et al. / Gait and Posture 21 (2005) 311–317

by Griffin et al. [10], the heel-contact gait of HTWs hasnever been investigated and so the present study set out toinvestigate the heel-contact gait in HTW.

2. Methods

2.1. Subjects

Thirteen children (mean age 6.6±2.2 years, mean weight25.5 ± 9 kg, mean height 118± 15 cm) participated in thestudy, with their parents’ informed consent and the approvalof the local Ethical Committee. Two children were siblings(numbers 11 and 12). All were referred for gait analysis be-tween 1997 and 2000, and had been diagnosed with HTWby a pediatric neurologist. All had normal muscle tone, deeptendon reflexes, muscle strength and sensation, normal brainand spinal cord imaging, muscle enzymes, calf muscle EMGand sural nerve somato-sensory evoked potentials[3,14].None developed any neurological disorders over the subse-quent 3–6 years after gait analysis.

Bilateral goniometric evaluation of passive ankle dor-siflexion, by one experienced physiotherapist before therecording session (prone position, knee extended, measure-ment of end-range resistance while ensuring no foot prona-tion), indicated variable limitation in 10 children. To assessa possible relationship between passive muscle–tendon stiff-ness and walking patterns, four ranges of ankle dorsiflexionwere established: level 0 (reference), > 20◦; level 1, from>10◦ to 20◦; level 2, from >0◦ (neutral) to 10◦ and level 3,=<0◦ (Table 1).

2.2. Set-up

Recordings were made in a gait laboratory (SAFLo,Milan) equipped with a 9 m linear walkway. Lower limb

Table 1

Subject Sex Age (years) Limitation of passive ankle dorsiflexiona Rate of occurrence of heel-contact walking(percentage of accepted strides)b

Right side Left side

1 F 7 2 2 162 F 4 1 1 153 F 9 3 3 434 M 5 3 3 505 M 6 3 3 836 M 11 3 3 617 M 7 0 0 928 M 6 1 1 379 M 4 2 3 75

10 M 10 0 0 4511 M 8 2 3 7612 F 4 0 0 5713 M 5 1 1 33

a Passive ankle dorsiflexion (knee extended): level 0, >20◦; level 1, from >10◦ to 20◦; level 2, from >0◦ (neutral) to 10◦ and level 3,=<0◦.b Acceptance criteria: steady state, straight-line progression, only one foot on the platform and in its entirety, no lost markers, artefact-free EMGs.

kinematics were obtained by a motion analyser (ELITE,BTS, Italy) employing four 50 Hz TV-cameras which de-tected the 3D coordinates of retro-reflective markers (10 mmin diameter) glued onto the posterior superior iliac spines,lower vertex of the sacrum, lateral femoral condyles, lateralmalleoli and lateral aspect of the fifth metatarsal heads,according to the SAFLo protocol[22]. The accuracy ofdetection of the marker coordinates was about 1 mm in acalibrated volume 3 m long, 2.5 m high, 1.2 m wide. Anthro-pometric measurements were used to calculate hip and kneejoint centers, according to a model described in[23]. Thefoot/floor ground reaction forces were recorded by a forceplatform (Kistler, Switzerland) embedded in the floor at themiddle of the walkway and not visible to the subject. Tele-metric EMGs (Telemg, BTS, Italy) were collected by surfacepre-amplified bipolar electrodes (diameter 20 mm, inter-axisdistance 25 mm) from the soleus (SOL), gastrocnemius me-dialis (GAM), gastrocnemius lateralis (GAL) and tibialisanterior (TA) muscles of both limbs, following standard pro-cedures for sensor placing and cross-talk minimization[24].

2.3. Protocol

Children walked barefoot and practiced for a short period.During the experimental session, they were asked to “walk”and “walk fast”, in a random sequence, with no constraintson the gait speed and style. Data were captured over at least30 trials (4 s each). The starting position was adjusted toensure that right and left limbs landed on the platform ina comparable number of trials. On average, 23 trials persession met the requisites for further analysis (steady state,straight-line progression, only one foot on the platform andin its entirety, no lost markers, artefact-free EMGs). For eachtrial force-plate data were available from one stride of onelimb, while kinematics and EMG data were available fromone or two strides of both limbs.

P. Crenna et al. / Gait and Posture 21 (2005) 311–317 313

Fig. 1. Kinematics of the heel-contact gait in a control child and a representative habitual toe walker (HTW). Each panel shows, from top to bottom,the profiles of the knee, ankle, leg/vertical and foot/horizontal angles, over a normalized stride cycle. For each subject, five strides at different velocitiesare superimposed (control: 76, 78, 85, 121, 155% height/s, mean 103% height/s; HTW: 82, 92, 96, 112, 138% height/s, mean 104% height/s, subjectnumber 8). The thick lines at the bottom indicate the mean duration of the stance phase± S.D. Premature heel-rise and reversal of the second rocker inthe HTW are marked by arrows on the profiles of the foot/horizontal angle and ankle joint angle, respectively.

2.4. Data processing

Kinematic, force-plate and EMG data were time-normalized with respect to the gait cycle. The longitudinalaxes of thigh, shank, and foot were defined as the linesconnecting the hip and knee, the knee and ankle, and theankle and mid-forefoot centers, respectively. The flex-ion/extension angles at the ankle, knee and hip joint werecalculated according to the SAFLo model[22] (Fig. 1, left).The angles of the longitudinal axis of the leg to the verticalaxis (positive: forward), and of the longitudinal axis of thefoot to the horizontal plane were measured. In the normalheel-contact gait, the foot-horizontal angle stays withinthe upright standing range throughout the foot-flat supportperiod, until a large negative wave begins at mid stance,coinciding with the heel-rise. On this negative wave, weidentified the start of the heel-rise as the moment when thefoot-horizontal curve had moved more than 3◦ from thestanding condition. The same threshold was used to mea-sure the onset of the late stance push-off plantar-flexion,on the ankle joint angle profile. Spatio-temporal repre-sentations of the evolution of the ground reaction vectorssampled every 20 ms during the stance phase (vector dia-grams, VD[25,26]) were also generated from the platformdata. The anterior–posterior length of the support underthe foot sole was computed from the forward displacementof the center of foot pressure (COP) and was normalizedto the foot length. A gait pattern in which foot kinemat-

ics and VD indicated a ground contact made by the heeland a maximum forward displacement of the COP ex-ceeding 80% of the foot length was classified as heel-toewalking.

The EMG recordings were high-pass filtered (10 Hz), am-plified (1000×) and full-wave rectified. In the triceps surae(TS) the EMG components normally occupying the firsthalf (TSa) and the second half (TSb) of the single-supportphase were analysed independently, with a separationpoint set at the zero-crossing of the anterior–posteriorground reaction force. In the antagonist TA muscle, thetwo EMG bursts recruited around ground contact (TAa)and toe-off (TAb) were evaluated independently. By meansof an interactive program (Matlab), the amplitude of theEMG components was measured by one of the authors(PC), as the area under the envelope, and the timing wasdetermined with reference to the initial ground contact(time 0).

Kinematic, force-plate and EMG parameters from all theheel-contact strides by each HTW were first averaged to rep-resent that individual, and then the averages were used forgroup analysis. Comparisons were made with normative datafrom 13 age-matched children from the authors’ database(30 trials each). Statistical analysis was made by one–wayanalysis of covariance (ANCOVA, fixed effects), with sub-ject group as independent variable, parameters of interest asdependent variables and walking speed as covariate. Statis-tical significance was assigned toP-values<0.05.

314 P. Crenna et al. / Gait and Posture 21 (2005) 311–317

3. Results

Within the spectrum of velocity covered in the “walking”and “walking fast” trials (73–130% height/s, mean 97±13%,S.D., 13 subjects), all the HTWs spontaneously adopted bothtoe/forefoot and heel-contact walking, in variable propor-tions. The incidence of heel-contact strides ranged from 15%to 92% (Table 1) and was not significantly correlated withthe children’s age, passive ankle joint excursion, walkingspeed or trial order. Within the HTW group, no significantdifferences were observed in the spatio-temporal dimensionsof the heel-contact strides compared with toe/forefoot con-tact strides. The latter had the typical kinematic, kinetic andEMG features described in the literature (seeSection 1), soanalysis focused on the sole strides with a heel-toe groundcontact pattern.

3.1. Kinematics

The leg/foot position at touchdown (Fig. 1) showed asmall increase in knee flexion (23± 3◦ HTWs versus 16±5◦ controls, mean± S.D., 13 subjects each group;P <

0.01), and a more vertical tibia alignment (leg/vertical angle12 ± 4◦ versus 16± 3◦) and downward inclination of thefoot (foot/horizontal angle−12± 7◦ versus−9± 4◦), bothbelow statistical significance. Likewise, the ankle joint angleat contact was not significantly modified (65± 6◦ versus64± 3◦).

Fig. 2. Vector diagrams and pathways of the center of foot pressure (COP) during heel-contact gait in a control child (A–C) and four HTWs (D–G:numbers 10, 3, 7 and 1). Numbers on top of each diagram indicate walking speed. Progression from left to right. Downward displacement of COP is inthe medio-lateral direction. BW, body weight. Faster advancement of COP during weight-acceptance in the HTWs as compared to controls is indicatedby the lower density of vectors and their points of application in left wing of the vector diagrams.

The early-mid stance phase was marked by premature riseof the heel from the ground (at 20±6% of stride versus 27±3%;P < 0.05;Fig. 1), which was not significantly correlatedwith age, plantar-flexor stiffness, trial order or knee flexionangle at initial contact. Moreover, in contrast to the controls,whose heel-rise onset was negatively correlated with thewalking speed (r = −0.67; P < 0.05), no such trend wasseen in the HTWs. A further kinematic finding in all subjectswas an abnormal switch from dorsal- to plantar-flexion ofthe foot, i.e. reversal of the second rocker (70± 27% of theheel-contact strides, onset at 20±5% of stride;Fig. 1). Thiswas followed by slower ankle plantar-flexion, until the mainpush-off wave started in late stance (onset at 48± 3% ofstride in HTWs versus 45± 5% in controls, n.s.).

3.2. Ground reaction forces

In the HTWs the VDs indicated obvious abnormalities in97% of the strides, particularly during the weight-acceptancephase (Fig. 2). On average, the backward-oriented vectorsoccupied a larger proportion of the foot length than in nor-mal controls (60± 15% versus 31± 9%, P < 0.001). Inaddition, the mean distance between their points of appli-cation (which represents the velocity of COP progressionduring the post-contact deceleration phase) was significantlygreater in HTWs (6.55 ± 1.8% versus 2.61 ± 1.2% footlength,P < 0.001). In parallel, the deceleration phase, mea-sured on the backward-directed component of the horizontal

P. Crenna et al. / Gait and Posture 21 (2005) 311–317 315

force, was significantly shorter (21± 3% versus 29± 2%of stride,P < 0.001). In 11% of the strides, a short-lastingbackward displacement of the COP pathway was also no-ticed in early stance (reversal of the second rocker was ab-sent in these strides).

3.3. Electromyography

There were substantial changes in the EMG activity re-cruited on the triceps surae during the first half of the stancephase (TSa component), which was analysed in further de-tail for the SOL muscle (Fig. 3). In the control children,TSa-SOL began on average at 14± 4% of the stride andattained an amplitude (area) as large as 43± 18% of themain propulsive SOL burst (TSb-SOL). The onset latency ofTSa-SOL was positively correlated with the walking speed(r = 0.49; P < 0.05), while its relative amplitude was neg-atively correlated (r = −0.60; P < 0.01). This referenceEMG profile was classified as type I-SOL pattern.

In 87% of the heel-toe strides of HTWs, TSa-SOL startedearly (6.5 ± 4.5% of the stride,P < 0.05, peak between13% and 16%), was increased in amplitude (area 91± 31%TSb,P < 0.005) and was frequently separated from TSb bya period of EMG silence (amplitude within the same range

Fig. 3. EMG patterns in the leg muscles of HTWs during heel-contact strides. The upper panels give examples of the typical EMG profiles. The lowerpanel shows the most common combination (Type II TS/Type I TA, five strides superimposed, subject number 7), with the average ankle joint angle(hatched area: mean control value± S.D.).

observed in the swing phase). Analysis of speed-dependenteffects on the pooled data from the 13 HTWs brought tolight a tendency to changes in amplitude, but not in timing.This figure was classified as type II-SOL pattern. In 13% ofthe heel-contact strides of HTWs, TSa-SOL started beforecontact with the ground (onset−10±4% of stride) and waslarger (area 85± 63% TSb), thereby offering a third profile,classified as type III-SOL. This was similar to the one ob-served in the same subjects during the toe/forefoot contactstrides which, however, had significantly greater SOL ac-tivity (area 180± 52% of TSb,P < 0.001). The behaviourdescribed for TSa-SOL was also detected in TSa-GAM and,less consistently, in TSa-GAL, in both of which the time ofonset was not significantly different from SOL.

The antagonist tibialis anterior had a normal TAa compo-nent in 88% of the heel-contact strides by HTWs (type I-TApattern). Less frequently, TAa was recruited prematurely; itwas hardly separated from TAb and showed no or only lit-tle activation in early stance (type II-TA pattern; incidence7%). Sometimes TAa was delayed toward the early stance,with no or little pre-contact activation (type III-TA pattern;incidence 5%). In conclusion, the EMG pattern typically as-sociated with the heel-contact strides of HTWs (P < 0.005)was type II-TS/type I-TA (Fig. 3, lower panel).

316 P. Crenna et al. / Gait and Posture 21 (2005) 311–317

4. Discussion

4.1. Incidence of heel-contact walking in HTWs

During any single recording session, the HTWs sponta-neously adopted both toe/forefoot and heel-contact gait. Therate of occurrence of the two patterns did not correlate withthe children’s age, or the static, passive ankle joint excursion,which ranged from a normal to mildly/moderately-limiteddorsiflexion. In our children, therefore, the rheologicalproperties of the plantar-flexor muscle–tendon complexwere not likely to be a primary cause of the different lo-comotor phenotypes. Along the same line, Hicks et al.[14]reported that seven HTWs (four with static equinus contrac-tures) “were able to vary their ankle motion from a toe-toepattern to a heel-toe pattern, even without verbal cueing”.Context-contingent control mechanisms might possibly playa role. In this respect, the fact that there was no correlationbetween the incidence of heel-contact strides and eitherthe walking speed or trial order indicates that “executive”aspects of locomotor control, as well as short-term adaptivephenomena, are not consistently involved.

4.2. Distinction from the heel-contact gait ofnormal children

The most striking changes marking the heel-contact gaitof HTWs were seen at 10–30% of the stride cycle. DynamicEMG studies have shown that during this phase the forwardrotation of the shank and the simultaneous progression ofthe COP are normally controlled by a lengthening contrac-tion of the TS (the TSa component), with EMG silence onthe antagonists TA[27,28]. In the heel-contact strides ofHTWs, TSa proved to be most often prematurely recruited,with an increase in relative amplitude, and frequently fol-lowed by a definite silent period, while the reciprocalEMG silence was usually preserved. The prematurely aug-mented plantar-flexor torque resulting from these changesmight well be responsible for the rapid progression of theCOP toward the metatarsal heads and the shortening ofthe weight-acceptance phase. Moreover, the enhanced TSain the bi-articular knee-flexor gastrocnemii was congruentwith a premature rise of the heel and early plantar-flexionof the foot. Indeed, TSa activity attained a peak between13% and 16% of the stride, which, taking into accountthe electro-mechanical delay[29], is consistent with theachievement of maximum mechanical effect at around 20%of the stride. The increased braking of the forward rotationof the shank, produced by the augmented TSa component,may produce knee hyperextension. The absence of hyper-extension in the children tested indicates a protective actionof the gastrocnemii and possibly of the hamstrings, but thelatter was not analysed in this study.

In their study of six HTWs Griffin et al.[10] mentionedthat during heel-toe contact walking the calf muscles wererecruited in early-stance, “with the gastrocnemius becom-

ing active sooner than usual”. However, they also reportedco-activation of the antagonist muscle (our uncommon typeIII-TA pattern) and a longer first rocker. The partial discrep-ancy with the findings of this study might be because Grif-fin and colleagues studied voluntary and not spontaneousheel-contact walking. The children in their study had to beasked to perform this task and two of them “complained ofcalf pain” when doing so. They also walked in specially de-signed boots with metal contacts and not barefoot.

4.3. Distinction from immature heel-contact walking

In view of the proposed link between HTW and imma-ture locomotion[3,13,14], possible similarities should beconsidered between the heel-contact gait of HTWs and thattypical of growing children during their transition from anearly independent to an adult walking pattern[2,20]. Thefew available kinematic and EMG reports[1,2,20,30]indi-cate no real similarities with the patterns described here forthe HTW. According to Okamoto and Gato[2] the activityof the lateral gastrocnemius in the first half of the stance(the TSa component) is hardly detectable in stride cycleswith weak plantar-flexion before touchdown by the end ofthe second year. Moreover, when TSa activity is enhanced,the discharge of the antagonist muscles and the ground con-tact leg/foot position are clearly different from those docu-mented here in the HTWs (see Fig. 6 in[30]).

Although further information is certainly necessary, thepresent evidence suggests that the heel-contact gait ofHTWs is not consistently related to immature locomotorprogrammes. EMG studies under static conditions havein fact shown that muscle synergies reportedly consideredas “primitive” (e.g. co-activation of ankle and knee exten-sors during voluntary knee extension) are absent in HTWs,whereas they are seen in children with mild diplegia[17].

4.4. Distinction from the heel-contact gait ofchildren with CP

Many children with CP lack a heel-strike. The equinus,crouched, stiff-legged, hip-hiking and recurvatum knee pat-terns, and all the related mixed forms, fall within this group[31–34]. Among the gait patterns that may involve heel con-tact, the “mild knee” walking pattern—with a reduced rangeof motion in the lower limb joints but preservation of themain kinematic profiles[34]—can be easily distinguishedfrom HTWs. The “jump knee” pattern (increased knee flex-ion in early stance rapidly corrected in mid-late stance)comprises a possibly normal ankle attitude at touchdown,post-contact increment of TS activity and premature ankleplantar-flexion during second rocker, as seen in the HTWs[34]. However, distinguishing features of CP children in-clude a major increase in knee flexion at touchdown (mean40◦), immediate onset of ankle dorsiflexion after groundcontact, augmented plantar-flexion in the late stance/earlyswing phase, major increase of the vertical reaction during

P. Crenna et al. / Gait and Posture 21 (2005) 311–317 317

weight-acceptance (mean 150% of the push-off wave) andpre-contact activation of TS (peak at around 5% of the stride)[33]. In the light of these considerations, therefore, analy-sis of heel-contact gait might be useful for differentiation ofHTW and CP, particularly in spastic diplegia, which is notfeasible from the evaluation of toe/forefoot contact walking[3,4,14,17].

5. Conclusions

The heel-contact gait in HTWs appears specific and dif-fers from that seen in age-matched controls, toddlers andfrom patterns most commonly described in CP. Examina-tion of heel-rise timing, ankle joint angle, VD pattern andspatio-temporal profiles of TSa and TAa EMG compo-nents might provide a complementary diagnostic tool forHTWs.

Acknowledgements

This study was supported by Ministry of Education(F.I.R.S.T.) and by the P. and L. Mariani Foundation. Thetechnical assistance of Mauro Recalcati is gratefully ac-knowledged.

References

[1] Forssberg H. Ontogeny of human locomotor control. I. Infant step-ping, supported locomotion and transition to independent locomo-tion. Exp Brain Res 1985;57:480–93.

[2] Okamoto T, Goto Y. Human infant pre-independent and independentwalking. In: Kondo S, editor. Primate morphophysiology, locomotoranalyses and human bipedalism. Tokio: University of Tokio Press;1985. p. 25–45.

[3] Sala DA, Shulman LH, Kennedy RF, Grant AD, Chu ML. Idio-pathic toe walking: a review. Dev Med Child Neurol 1999;41:846–8.

[4] Taussig G, Delouvée E. La marche en equin idiopatique del’enfant. Diagnostic and evolution spontanée. Ann Readapt Med Phys2001;44:333–9.

[5] Hall JE, Salter RB, Bhalla SH. Congenital short tendo calcaneus. JBone Joint Surg 1967;49B:695–7.

[6] Sobel E, Caselli MA, Velez Z. Effect of persistent toe walking onankle equinus—analysis of 60 idiopathic toe walkers. J Am PodiatrMed Assoc 1997;87:17–22.

[7] Kogan M, Smith J. Simplified approach to idiopathic toe-walking. JPediatr Orthop 2001;21:790–1.

[8] Levine MS. Congenital short tendo calcaneus: report of a family.Am J Des Child 1973;125:858–9.

[9] Katz MM, Mubarak SJ. Herditary tendo Achilles contractures. JPediatr Orthop 1984;4:711–4.

[10] Griffin PP, Wheelhouse WW, Shiavi R, Bass W. Habitual toe-walkers.A clinical and electromyographic gait analysis. J Bone Joint Surg1977;59B:97–101.

[11] Caselli MA, Rzonca EC, Lue BY. Habitual toe walking. Evaluationand approach to treatment. Clin Pediatr Med Surg 1988;5:547–59.

[12] Shulman LH, Sala DA, Chu MLY, Mc Caul PR, Sandler BJ. De-velopmental implications of idiopathic toe walking. J Pediatrics1997;130:541–6.

[13] Cahuzac JP, Carriere JP, Pascal JF. Le toe walking syndrome. Mono-graphie de Podologie 8. Paris: Masson; 1987.

[14] Hicks R, Durinick N, Gage J. Differentiation of idiopatic toe-walkingand cerebral palsy. J Pediatr Orthop 1988;8:160–3.

[15] Kelly IP, Jenkinson A, Stephens M, O’Brien T. The kinematic patternsof toe walkers. J Pediatr Orthop 1997;17:478–80.

[16] Papariello SG, Skinner SR. Dynamic electromyography analysis ofhabitual toe-walkers. J Pediatr Orthop 1985;5:171–5.

[17] Policy JF, Tornburn L, Rinsky LA, Rose J. Electromyographic test todifferentiate mild diplegic cerebral palsy and idiopatic toe walking.J Pediatr Orthop 2001;21:784–9.

[18] Kalen V, Adler N, Bleck EE. Electromyography of idiopathic toe-walking. J Pediatr Orthop 1986;6:31–3.

[19] Leonard CT, Hirshfeld H, Forssberg H. The development of inde-pendent walking in children with Cerebral Palsy. Dev Med ChildNeurol 1991;33:567–77.

[20] Sutherland DH, Olshen RA, Biden EN, Wyatt MP. The developmentof mature walking. London: Mac Keith Press, Oxford: BlackwellScientific Publications Ltd., Philadelphia: JB Lippincott Co.; 1988.

[21] Davids JR, Foti T, Dabelstein J, Bagley A. Voluntary (normal) versusobligatory (cerebral palsy) toe-walking in children. A kinematic. JPediatr Orthop 1999;19:461–9.

[22] Frigo C, Rabuffetti M, Kerrigan CD, Deming CL, Pedotti A. Func-tionally oriented and clinically feasible quantitative gait analysismethod. Med Biol Engin Comp 1998;36:179–85.

[23] Frigo C, Rabuffetti M. Multifactorial estimation of hip and kneejoint centres for clinical application of gait analysis. Gait Posture1998;8:91–102.

[24] Hermens HJ, Freriks B. The state of the art of sensors and sen-sor placement procedures for surface EMG: a proposal for sensorplacement procedures. Roessingh Research and Development. TheNetherlands; 1997. p. 125–46.

[25] Pedotti A. Simple equipment used in clinical practice for evaluationof locomotion. IEEE Trans BME 1977;24:456–61.

[26] Crenna P, Frigo C. Monitoring gait by a vector diagram technique inspastic patients. In: Delwaide PJ, Young RR, editors. Clinical neu-rophysiology in spasticity. Amsterdam: Elsevier Science Publishers;1985. p. 109–24.

[27] Perry J. Gait analysis. Normal and pathological function. New York:Slack; 1992.

[28] Crenna P, Do MC, Brénière Y. Motor programmes for the terminationof gait in humans: organisation and velocity-dependent adaptation.J Physiol (Lond) 2001;537:1059–72.

[29] Cavanagh PR, Komi PV. Electromechanical delay in human skeletalmuscle under concentric and eccentric contractions. Eur J ApplPhysiol 1979;42:159–63.

[30] Okamoto T, Okamoto K, Andrew PD. Electromyographyc develop-mental changes in one individual from newborn stepping to maturewalking. Gait Posture 2003;17:18–27.

[31] Gage J. Gait analysis in Cerebral Palsy. London: Mac Keith Press;1991.

[32] Sutherland DH, Davids JR. Common gait abnormalities of the kneein cerebral palsy. Clin Orthop 1993;288:139–47.

[33] Kerrigan DC, Sheffler LR. Spastic paretic gait: an approach to eval-uation and treatment. Clin Rev Phys Rehab Med 1995; 253–68.

[34] Lin CJ, Guo LY, Su FC, Chou YL, Cerng RJ. Common abnormalkinetic patterns of the knee in gait in spastic diplegia of cerebralpalsy. Gait Posture 2000;11:224–32.