Toe walking gait is characterized by forefoot weight bearing only, rather than by heel-toe weight progression. As many as 24% of toddlers engage in toe walking, but this usually resolves spontaneously.1 , 2 When this gait persists beyond 2 to 3 years of age in a child without evidence of neurologic impairment, the condition is diagnosed as idiopathic toe walking (ITW).2 Prevalence and etiology of ITW are unknown, but family history or premature birth is often associated.1 , 3 , 4 Consensus about medical management has not been established. Some advocate referral for intervention by 3 years of age, whereas others view the condition as benign and consider intervention unnecessary.2 , 5
In mature gait, the weight-bearing point of contact begins at the heel, transfers forward beneath the ankle during midstance, and terminates with forefoot push-off. Midstance support surface contact is maintained by superimposed body segment organization.6 With toe walking, normal loading and midstance transfer of body weight are absent.
Normal heel-toe progression requires at least 10° of ankle dorsiflexion (DF) range of motion.6 Limited DF has been widely reported with ITW.3 , 4 , 7 , 8 In children who are typically developing (TD), DF decreases from an average of 54° at birth to 41° at 2 years of age.9 For those over 2 years normative data are sparse, but Cusick10 indicated 20° to 30° DF is usual by 4 to 7 years.10 In those with ITW, limitations in DF often progress with advancing age and contribute to other musculoskeletal abnormalities such as excessive pelvic tilt, genu valgum, genu recurvatum, or external tibial torsion.4 , 7 , 11 Undifferentiated leg pain has also been reported.4 , 12
Interventions to address persistent toe walking have been directed toward increasing DF in children aged 4 to 13 years, with the expectation that gait changes would follow the gains. Methods used to increase DF have included serial casting, ankle foot orthoses, botulinum toxin, surgical Achilles tendon lengthening, and passive stretching.3 , 8 , 11 – 13 Two to 13 years after initial orthopedic assessment, Eastwood and colleagues13 reported outcomes for 49 children with ITW who received no intervention. Parent ratings (analog scale of percentage of toe walking) indicated that half had improved and half were unchanged. Physician evaluation of wet footprints indicated that 12% achieved heel-toe gait.13 In the same study, the long-term effects of casting were evaluated in 41 children. Parents rated gait unchanged in 49% and improved or normal in 51%. Physicians rated 78% as toe walking, and 22% normal.13
In 8 children with ITW, Brouwer and associates8 reported DF gains using serial casting. Before treatment, 5 of the children consistently demonstrated toe walking and 3 fluctuated between toe and foot-flat contact. Videotape observational gait analysis immediately after casting showed that all could achieve heel strike. Seven of the children retained this ability 6 weeks later.8 Using the normalcy index (NI), McMulkin and associates11 compared gait in 14 patients before and after surgical lengthening of gastrocsoleus. The NI provides an overall rating relative to normal gait based on analysis of 16 kinematic and kinetic variables. Preoperatively, the NI was significantly different from normal in individuals with ITW. Following surgery, the NI showed statistically significant improvement from initial scores but remained significantly different from normal values.11
Hirsch and Wagner12 retrospectively reviewed conservative treatment of 14 children. Passive stretching was applied to calf muscles by a physical therapist and in 5 cases augmented with serial casting or ankle foot orthoses, and a home program was conducted by parents (unspecified frequency or episode length). One patient showed an immediate intervention effect on gait. Seven years after intervention, 73% of participants did not demonstrate toe walking when observed unobtrusively.12
Mechanisms underlying persistent toe walking may relate to motor immaturity. The propulsion strategy in toddlers with TD differs from that seen in mature gait. Movements are induced by whole-body displacements with stepping for balance recovery. Lower extremity antagonist coactivation is postulated to compensate for limited muscular strength, and push-off is absent.14 , 15 During the development of gait, initial heel contact may appear by 18 to 24 months, but typically heel-strike with active dorsiflexion does not occur before 2 years.15 Gradual transformation to the use of plantar flexion for propulsion is not complete until 4 to 5 years of age.16 During the period of emerging heel-strike, trunk alignment changes from slight forward inclination to a more erect posture, and reciprocal activity between lower extremity antagonists gradually replaces the coactivation pattern.15 Reciprocal activation and a more upright trunk posture are postulated to play a pivotal role in attaining heel-toe gait. Delayed attainment of these attributes may underlie persistent toe walking, but literature to support this postulate has not been found.
Motor control has been defined as regulation of mechanisms essential to movement.17 Motor control interventions are directed at changing movement capacity, in contrast to orthopedic interventions that primarily focus on attaining normal DF range of motion. Clinical observation of the gait of preschool children with ITW reveals toddler-like gait attributes: constant running rather than walking, poor ability to modulate pace, and jumping or hopping rather than controlled stepping. Play behaviors may include avoidance of transition movements (such as stooping) that require eccentric lower extremity muscle control. These gait and play characteristics are postulated to reflect a lag in reciprocal dorsiflexion–plantar flexion development and failure to achieve full upright trunk posture. With a forwardly displaced trunk, the weight-bearing point of contact is confined to the forefoot, and heel-toe weight transfer is absent.
In a study of the natural history of DF in children with ITW aged 1 to 15 years, Sobel and associates4 proposed that ITW gait results from a long period of time walking on toes rather than DF limitations from birth.4 From a motor control perspective, limited reciprocal dorsiflexion–plantar flexion and upright postural control are viewed as the primary impairment, and gastrocsoleus muscle contracture resulting from the lengthy period of toe walking is viewed as a significant secondary problem.
The intervention protocol used in the present study was developed through clinical experience. With pre–school-aged children, motor control–based interventions were designed to influence gait attributes and simultaneously prevent or reverse the muscle contracture. Emphasis was placed on activities and habits intended to influence muscle activation and posture deficiencies, with the objective of expanding the child's ability to manage the body center of mass over the feet. Stance control gains were expected to facilitate spontaneous heel-toe progression and thus avoid or reverse the musculoskeletal consequences of toe walking. A published report using a strategy to actively change gait in children with ITW involved 2 cases and employed daily auditory feedback treatment using a pressure-sensitive heel-switch for periods of 3 and 6 months. Those authors reported DF and gait improvements that were maintained 12 months after treatment.18 No other studies were found documenting ITW outcomes after using motor control–based interventions.
The primary purposes of this study were (1) to develop a physical therapy motor control intervention protocol and (2) to evaluate the effects of this protocol on gait, motor skill development; and ankle mobility in 5 children aged 2½ to 6 years with ITW. The expectation was that improved lower extremity muscle control and trunk posture would shift gait toward a heel-toe pattern.
A convenience sample of 5 children was enrolled nonconcurrently in the study. Each met the following eligibility criteria: (1) age 30 to 72 months; (2) parent estimate of toe-walking frequency to be 50% or more; (3) neutral (0°) or greater passive DF; (4) motor skill development no lower than 1.5 SD below the mean on gross motor subtests of the Peabody Developmental Motor Scales, second edition (PDMS-2)19; and (5) available for 16 weeks of research activity. Each child was reported to have attained independent walking by 12 months of age. See Table 1 for participant characteristics. Parents signed an informed consent, approved by Rocky Mountain University of Health Professions institutional review board.
Instrumentation and Data Collection
Descriptive data were collected at a preintervention eligibility session using a parent questionnaire that was developed from literature review of characteristics believed influential in ITW (Appendix A available online at http://links.lww.com/PPT/A18). Following the preintervention eligibility session, 2 data collection schedules were employed to accommodate different types of measurements.
The first schedule used a variable baseline case series structure with multiple gait measurements (single system AB method and 2 postintervention measurement sessions). Gait measures were collected 5 or 6 times during the baseline phase, weekly during intervention, and at 2 follow-up sessions.
The second schedule was used for motor skill development and DF measures. These parameters were assessed at the preintervention eligibility session and 2 postintervention sessions. Figure 1 shows the order of activities for all participants incorporating both schedules. Participants enrolled as they became available and each completed all activities. Data collection occurred at a private pediatric physical therapy practice in Issaquah, Washington, and at a neurodevelopmental center in Seattle, Washington.
Except for gait laboratory measures, the reliability and validity of gait measures were not included in reports of interventions for ITW.8 , 11 – 13 , 18 For the present study, we sought a method that would overcome the propensity of children with ITW to alter their usual gait pattern in the clinic setting.3 , 4 , 12 To capture attributes of spontaneous gait with young children, a method using in-shoe sensors was developed, which could be used in a natural environment.
A portable in-shoe gait event detector (GED) (see Figure 2) was developed by Robert Price, MSME (University of Washington) and evaluated with 8 children with ITW and 6 with TD. The GED was used and evaluated within a controlled 130-m natural environment. A set of trials included walking outside on a level sidewalk (1) to a destination to get a “treasure box key” and (2) return to clinic for a prize. The test-retest reliability intraclass correlation coefficient for the GED was 0.85. Simultaneous comparisons of heel-strike frequency with the GED and outdoor video resulted in 70% agreement. For these trials, the camera was transported to follow the child, starting at a distance of 12 ft and then a variable distance as the trial progressed. Known groups comparison between children with TD and ITW was significantly different for both GED and video measures (Mann-Whitney U test: P < .001 and P < .002, respectively). Validity of GED detection of heel-strike frequency was judged as moderate, but technical difficulties became apparent with individual participants during this intervention study.
In addition, parents' perceptions of heel strikes in daily living at baseline and from the previous week were recorded using a visual analog scale (VAS) (Appendix A, available online at http://links.lww.com/PPT/A18). The VAS was used to judge whether families observed the presence or absence of spontaneous heel-toe gait at home. Overall the Spearman ρ correlation was 0.34 between all GED and VAS measures taken during this study.
PDMS-219 stationary, locomotor, and object manipulation subtests were administered to obtain the gross motor quotient (GMQ): testing 3 times within 4 months was more frequent than the usual 6-month intervals between PDMS-219 administrations. Standard procedures were followed to determine the range of items administered. The first author provided intervention. An independent pediatric physical therapist unfamiliar with the child administered follow-up PDMS-219 examinations. Before the study, the first author established 89% agreement with items on the University of Washington PDMS-219 training module. The independent rater then established 84% agreement with the investigator during simultaneously scored testing of a 4-year-old child with TD.
The first author measured DF with a goniometer for initial and follow-up examinations. The child was positioned in prone with the knee supported in neutral extension and the rearfoot stabilized in subtalar neutral. The mean of 3 measurements was used in the analysis.
The primary objective of motor control intervention was to facilitate a more erect standing and walking posture to secure the ground reaction force relative to the ankle axis.20 In standing, additional treatment objectives were facilitated with a plantigrade foot, such as neutral calcaneous alignment and neutral knee extension. Intervention was composed of 2 one-hour sessions per week over 9 weeks, provided by the first author. Each child's play choices were adapted to therapy goals. See Appendix B (available online at http://links.lww.com/PPT/A19) for photographic examples, detailed strategies, and home program sequence. At the first intervention session, a written explanation about ITW was given to parents and they were invited to be present during sessions. At each session, the opportunity was given to discuss gait concerns. Home assignments were based on the child's performance during the sessions, but home activity compliance was not documented.
An independent pediatric physical therapist rater scored 3 intervention videotapes of each child for procedural reliability. Videotaping occurred between sessions 4 to 6, 10 to 12, and 16 to 18, with the sessions chosen when concurrent activities in the clinic were at a minimum. The camera was set on a tripod to span the room and ran unattended for the duration of the session. The camera was not obscured from the therapist or the child. The independent rater scored procedural reliability using a checklist of activity items that were documented as present or omitted. A mean procedural reliability score of 87% was obtained for all sessions.
For the GED gait measures baseline and treatment phase comparisons were carried out using a visual display of celeration lines (Figure 3) to determine whether change had occurred.21 Any upward trend toward the right was considered a change toward heel-toe gait. Consistent measures (3 of 4 in series) at 80% or greater frequency of heel-toe progression were set as clinically meaningful change. Spearman rank correlations between GED and parental VAS measures of frequency of heel-toe gait VAS were calculated for individual cases.
Because of the small sample size, nonparametric statistics (Friedman analysis of variance) were used for comparisons of DF and PDMS-2 19 subscale data for 3 examinations (preintervention, immediate postintervention, 30-day follow-up). The alpha level was 0.05. No change was predicted for the PDMS-219 GMQ scores. Post hoc analyses employed the Wilcoxon signed-ranks tests for each sample pair, with a Bonferroni corrected alpha value less than 0.017. A change of 5° or greater in DF was considered clinically meaningful.
Individual case GED heel-strike frequency charts are shown in Figures 3a and 3b. GED toe sensor problems eliminated measures of one leg for participants 1, 2, and 5. Gait descriptions and parental VAS data are presented in the description of each case.
Ankle DF for each case is shown in Figure 4. Group comparison showed a statistically significant difference (P = .001). Post hoc comparison differences were significant between preintervention and both follow-up examinations (P = .007 and .005, respectively), but not significant between follow-up examinations (P = .497).
Individual subject and group mean scores for the PDMS-2 19 subscales are presented in Figure 5. Group GMQ score comparisons showed a significant difference (P = .022). However, post hoc comparisons failed to meet critical values for preintervention versus the 2 follow-up sessions, (P = .043 and .043, respectively) or between the follow-up examinations (P = .785).
Case Outcomes and Descriptions
In the baseline phase (Figure 3a), the GED celeration trend line sloped steeply upward and displayed lack of stability. In the intervention phase the trend line sloped upward to a level above baseline range, showing a heel-strike frequency gain. The immediate postintervention score was consistent with the trend, but 30 days later the score dropped to baseline range, showing reversal when intervention ceased. During baseline and the first 3 weeks of intervention the VAS scores were 10%; they then increased to 30% for the remaining sessions. A moderate correlation was found between VAS and GED scores (rs = 0.63). At the immediate follow-up, right DF showed a clinically meaningful gain of 6° (see Figure 4) compared to the eligibility examination. No additional intervention took place after concluding the study, and 8 months later the parent indicated return of the toe-walking gait. The child was then fitted for custom foot orthotics, which she used for activities of daily living. Within 6 weeks of orthotic intervention, the parent reported that foot alignment concerns, toe-walking gait, and leg pain had resolved.
In the baseline phase (Figure 3a), the right GED celeration line was nearly level. The intervention line sloped slightly upward with the highest heel toe frequency at 46%, indicating a gain but not a predominant heel-toe gait. Alignment of the immediate follow-up measure with the gradual slope supports a possible treatment effect. The parent rating of heel-strike frequency was consistently 10% until the last follow-up measure at 15%. Negligible correlation was found between GED and VAS scores (rs = 0.09). Ankle DF measures, all lower than 5°, showed no clinically meaningful changes (Figure 4). Ankle clonus was not elicited at the qualifying assessment, but after treatment was under way, clonus was observed at various sessions. Achilles tendon lengthening took place a year after the study.
In the baseline phase (Figure 3a), heel-strike frequency was stable on the left and variable on the right. With intervention, both celeration lines sloped slightly downward. Follow-up measures appeared consistent with a pattern of measurement instability, indicating that intervention did not affect gait. Parental VAS ratings were nearly all 20% heel-strike and correlated negatively with the GED (rs = −0.35). Ankle DF gain was less than 5° on the right. A gain of 6° DF on the left was present at immediate follow-up, but measurement 30 days later did not support the change (Figure 4). Following the study, at the parent's request the child received 6 additional intervention sessions in the clinic. At discharge 6 months after the study, the parent indicated that the child's gait was “better” while wearing shoes with standard foot orthosis (Pattibob model shoe inserts, Cascade DAFO, Inc, Ferndale, WA), but that toe walking persisted when barefoot.
In the baseline phase (Figure 3b), heel-strike frequency was stable with near level celeration lines for both lower limbs. With intervention the left celeration line sloped slightly upward and the right celeration line remained level. Alignment of the immediate follow-up measures with the gradual celeration line slope for the left leg supports a slight treatment effect but not a change to a predominant heel-toe gait. During baseline and the first 3 weeks of intervention, parental VAS scores were 10% and then increased to 20% by the conclusion of intervention with follow-up at 25% and 50%. Only this subject showed VAS ratings generally higher than the GED but the correlation between VAS data and the GED was low (rs = 0.33). Ankle dorsiflexion range-of-motion measures, all between 5° and 10°, did not show clinically meaningful changes (Figure 4). Because the child had greater difficulty managing right leg stance, intervention efforts were biased toward right leg control. An informal parent report a year following the study indicated continued problems with right heel contact during gait.
In the baseline phase (Figure 3b), the right leg celeration line was level. With intervention, the celeration line sloped slightly upward, with one instance of 100% heel-toe gait. The immediate follow-up data point did not provide strong support for the upward trend, but the 30-day follow-up measurement was consistent with retention of a treatment effect. During baseline and the intervention phases, parental VAS scores ranged from 20% to 40%. Follow-up VAS scores were 15% and 10%. Correlation between the VAS and GED was negative (rs = −0.42). The parent indicated that VAS ratings varied because of increased toe walking with fatigue and footwear choices (different shoes; or barefoot, especially at home). Preintervention left and right DF measures were 12° and 9°, respectively. Following intervention, the left ankle achieved range within normal limits (20° or greater) and the right ankle measured 15° at the 30-day follow-up examination (Figure 4). A parent reported that the child had indicated one episode of leg pain believed to be associated with therapy activities. Throughout the study, the investigator observed small lower leg contusions, apparently caused by bumping into furnishings. By 6 weeks of the intervention phase, parents reported that the child took initiative to practice gait at home, and when “heel-toe” was mentioned she stopped toe walking. An informal parent report a year following the study indicated gait mostly with heels down, but occasional toe walking associated with fatigue occurred when barefoot.
In contrast to approaches that directly address DF limitations with ITW, motor control intervention was based on the premise that toe-walking gait in children older than 3 years is due to motor control deficiency. However, results from this case series failed to show that the intervention protocol was sufficient to induce the expected gait shift. Parental VAS measures may not accurately quantify the extent of toe walking but were assumed to reflect whether toe walking was generally present or absent in daily life. No parent indicated a change to spontaneous heel-toe gait during the time period of the study. Similarly, GED measures did not indicate that participants achieved a consistent heel-toe pattern on the outdoor walking path. However, each parent anecdotally reported to the investigator that the child was stronger and/or responded more readily to requests to walk with heels down. Other outcome measures showed gains. Ankle DF measures improved at follow-up examinations, but only case 5 attained DF of 15° or greater. Although group contrasts with the PDMS-219 failed to meet statistical significance, each participant showed gains in follow-up GMQ scores compared to the initial examination.
Motor Control Intervention for ITW
Eastwood and colleagues13 indicated that with no intervention only 12% of children with ITW eventually acquired normal gait. In the review of passive ankle stretching, Hirsch and Wagner12 reported few immediate gait changes. When assessed at least 7 years afterward, gait appeared normal in 73% of participants.12 Similar to the retrospective report by Hirsch and Wagner,12 the motor control intervention provided with this study may have contributed to informally reported gait improvements with cases 1, 3, and 5 during the first year afterward.
Several factors may account for absence of the expected gait outcome with this study. First, measurement limitations were present in both the GED validation study and this intervention study. The GED validation study indicated that when outdoors children were less likely to alter gait, and the GED demonstrated moderate validity in that setting. Repeated measures in the baseline phase of this study suggest that in some children with ITW heel-strike frequency may not be a stable attribute. Variability could be due to volitional gait alteration or performance changes due to other factors such as fatigue. Unexpected toe sensor registration problems meant that for 3 participants the measurements for both extremities were unavailable. In the construction of the GED first metatarsal head weight-bearing was presumed, but while wearing shoes some children with ITW did not consistently trigger the sensor. Also, the correlation between GED measures and VAS measures was very low. The VAS was used to document toe walking versus heel-strike gait observations at home. In this intervention study, no change to spontaneous heel-toe gait was reported with the VAS, and therefore GED inaccuracy does not appear to be an explanation for absence of the expected outcome.
A second possible explanation of the failure to achieve a change in gait is that the protocol may have been inadequate to address the problem. With gait practice using auditory feedback, Conrad and Bleck18 indicated that gains in accumulated heel-strikes endured and reversal of DF limitation was maintained.18 Differences between this motor control protocol and their study were shorter intervention length (9 weeks vs 3-6 months) and lower intensity (twice weekly with a short home program vs daily). In this study, the preset intervention time-span (18 sessions in 9 weeks) may have been too short or predetermined activities may not have allowed enough flexibility to address specific needs. For example, discussion with parents about footwear and optional shoe inserts was included but the direction to limit barefoot walking was not given.
A third possibility for the absence of gait change is that the underlying etiology and presentation may be more complex and variable than originally assumed. Attributes observed but not quantified in this case series included complaints of pain, activity level (low stamina to high locomotor activity), tactile defensiveness, weakness, frequent stumbles or falls, asymmetry, sporadic unsustained ankle clonus, lower leg contusions, and possible abnormal musculoskeletal alignment of the foot.
The results of this case series are not generalizable because of limited GED baseline stability, a small convenience sample, use of 2 previously untested gait measurement systems, one investigator for all intervention and most outcome measures, and no system to monitor home program adherence. Participants were limited to children meeting specific eligibility criteria and were not representative of all children presenting for physical therapy intervention for toe walking. Ankle range-of-motion gains were reported, but reliability of goniometric measures with young children and for those with ITW were not obtained.
Summary and Implications
This case series was a starting point to determine whether a motor control protocol could be expected to change toe-walking gait in pre–school-aged children. The intervention was shown to reduce DF limitations, and it may advance motor skills and possibly (by parent anecdotal report) improve voluntary heel-toe gait. However, neither the parent VAS nor the GED measurement system demonstrated the expected outcome of a shift to spontaneous heel-toe gait during the study.
Additional treatment modalities, such as measures to reduce sensory reactivity, serial casting to address initial DF limitations, and orthotics for intrinsic foot instability should be examined for efficacy in conjunction with, or in lieu of, motor control intervention. Alternative measurement strategies are needed that are more sensitive to changes in the gait pattern, such as a pedometer to compare the number of steps over a set distance within the natural environment or the GAITRite system to measure the patterns of foot contact.
We thank the children and parents who participated in this study, Northwest Pediatric Therapies and Boyer Neurodevelopmental Clinic for gracious sharing of clinic space, Torey Gilbertson, DPT, for assistance with developmental testing and procedural reliability, and Anne Shumway-Cook PT, PhD, FAPTA, for mentorship during previous stages of this project.