The research protocol was reviewed and approved by our Institutional Research Review Committee. A database was utilized to identify all children with cerebral palsy between the ages of five and nineteen years who had comprehensive, quantitative gait analyses performed, both prior to and following ankle plantar flexor surgery, at our institution between January 2001 and June 2008. None of the fifty-three subjects had orthopaedic surgery prior to the initial motion analysis laboratory study, and none had received botulinum toxin injections within six months of the initial motion analysis laboratory study. The surgical procedure for each subject was a gastrocnemius-soleus recession, as initially described by Strayer, or a tendo Achillis lengthening, as initially described by White, based on the magnitude of the fixed muscle contracture, as either a single-event, single-level surgery or as part of single-event, multilevel surgery following the initial motion analysis laboratory study7-9. The specific surgical procedure at the ankle and all concomitant procedures are presented in a table in the Appendix.
The original records associated with the preoperative and postoperative motion study for each child were reviewed to determine the subject's age, sex, physical topographic classification of cerebral palsy (hemiplegia or diplegia), and functional level as determined by the Gross Motor Function Classification System (GMFCS)10. The affected side was included for the children with hemiplegic cerebral palsy. A single side was selected at random for inclusion for each subject with diplegic cerebral palsy. Data obtained from the preoperative and postoperative physical examination record in the Motion Analysis Laboratory for each patient included ankle dorsiflexion passive range of motion with the knee flexed and extended, ankle dorsiflexor selective control, ankle dorsiflexor strength, ankle plantar flexor strength, confusion test results, and ankle clonus. All goniometric measurements were made by specifically trained laboratory staff using the Shriners Motion Analysis Laboratory Network (SMALnet) standard clinical examination protocol with measures recorded to the nearest 5° increment11-13. Five clinicians performed the goniometric examinations in the Motion Analysis Laboratory during the time period included in the study. Regular quality control practices (e.g., having the physical examination periodically performed by two examiners together) were utilized to maximize interobserver reliability. Available evidence suggests that the reliability of measurements improves with the experience of the testers and the use of a standard protocol14. Ankle range-of-motion measurements were expressed in degrees. Ankle dorsiflexor selective control was graded on a 4-point scale from the SMALnet protocol (a higher score indicates better selective motor control)11,15. Ankle dorsiflexor strength was graded on an eight-level scale from the SMALnet protocol (a higher score indicates greater strength)11. Ankle plantar flexor strength was graded on a distinct eight-level scale based on the ability to perform a specific number of toe rises11,16.
The confusion test was graded as positive (obligatory ankle dorsiflexion occurring when flexing the ipsilateral hip against resistance in the seated position while the knee is flexed) or negative (resisted hip flexion occurring without associated ankle dorsiflexion in the seated position)17. A positive confusion test indicates the persistent presence of a primitive reflex connection or motor synergy between hip flexion and ankle dorsiflexion. Ankle clonus was graded as present or absent (following rapid manual dorsiflexion of the ankle with the subject in the seated position, with the hip and knee flexed to 90°)11-13.
Bilateral three-dimensional kinematic and kinetic data were collected with use of a twelve-camera motion measurement system (Vicon 512, Oxford Metrics Group, Oxford, England) and two force platforms (Advanced Mechanical Technology, Watertown, Massachusetts). Subjects were instrumented with passive reflective markers consistent with the Newington model for gait analysis18. Subjects made several passes through the laboratory measurement volume walking at a self-selected speed. Swing phase gait kinematic parameters examined in this study included peak and mean ankle dorsiflexion in the swing phase and were expressed in degrees. Active ankle dorsiflexor function in the swing phase was graded as present when there was either a dorsiflexion wave of >5° or the ankle was maintained at ≥10° above the maximum ankle plantar flexion motion, as determined by the physical examination, for the first half of the swing phase (Figs. 1-A and 1-B). Bilateral dynamic electromyographic data of the tibialis anterior and gastrocnemius muscles were collected with use of surface electrodes (M-100; Motion Lab Systems, Baton Rouge, Louisiana). Preoperative electromyographic data for the tibialis anterior and gastrocnemius were available for forty-nine extremities. Postoperative electromyographic data for the tibialis anterior and gastrocnemius were available for fifty extremities. Preoperative and postoperative electromyographic data for the tibialis anterior and gastrocnemius were available for forty-six extremities. Seven patients did not have recorded values for tibialis anterior, gastrocnemius, or both because of either technical or clinical problems. There is no consensus on the objective quantitative assessment of electromyographic data in children with cerebral palsy. To be consistent with clinical interpretation paradigms, the electromyographic activity of the tibialis anterior and gastrocnemius during the first half of swing phase was visually assessed (as “on” versus “off”) by a single, experienced clinician (J.R.D.), on the basis of the magnitude (>20% of the recorded gait maximum) and timing (relative to kinematic determination of toe-off) of the electrical signal.
Statistical Analysis
Preoperative and postoperative differences were compared with use of paired t tests for continuous clinical measures. For the clinical measures with a limited range of outcomes, preoperative and postoperative differences were evaluated with use of the nonparametric sign test and with use of a chi-square test (outcomes were identical for both evaluations). Proportions were compared with use of binomial tests. All comparisons were evaluated at a 0.05 level of significance. Using a stepwise approach, we estimated a logistic regression model of the preoperative ankle dorsiflexor function in swing as a function of other preoperative measures.
Source of Funding
The authors did not receive any outside funding or grants in support of their research for the preparation of this work.
Cohort Demographics
There were fifty-three children (twenty-nine [55%] were boys and twenty-four [45%] were girls) in the study cohort. Thirty-three children (62%) had hemiplegic cerebral palsy, and the remaining twenty children (38%) had diplegic cerebral palsy. The mean age at the time of the initial study in the Motion Analysis Laboratory was eight years and eleven months (range, five years and five months to sixteen years and five months). Thirty-two left lower extremities (60%) and twenty-one right lower extremities (40%) were selected for the analysis. Eight children (15%) had single-level surgery. The remaining forty-five children (85%) had single-event multilevel surgery. The mean time between the initial and postoperative follow-up study was two years and three months (range, one year and eight months to five years and five months). The follow-up study in the Motion Analysis Laboratory was performed between one and three years after the index surgery for fifty subjects (94%). At the time of the initial study, thirty-seven children (70%) were classified as having GMFCS level II, and the remaining sixteen children (30%) were classified as having level I. At the time of the follow-up study, twenty-six children (49%) were classified as having GMFCS level II. The remaining twenty-seven children (51%) were classified as having GMFCS level I. The change in GMFCS classification level between the two Motion Analysis Laboratory studies was significant (p = 0.021), with fifteen children (28%) improving from level II to I, four children (8%) deteriorating from level I to II, and thirty-four children (64%) showing no change in level.
Physical Examination Variables
Significant improvements were noted in passive ankle dorsiflexion with the knee flexed (p < 0.001) and extended (p < 0.001) following the ankle plantar flexor surgery (Table I). Thirty-two extremities (60%) showed an improvement of ≥5° in passive ankle dorsiflexion with the knee flexed. Thirty-six extremities (68%) showed an improvement of ≥5° in passive ankle dorsiflexion with the knee extended.
Preoperative selective control of ankle dorsiflexor function was normal in twenty-one extremities (40%), partial or limited in sixteen extremities (30%), and absent in sixteen extremities (30%). Postoperative selective control of ankle dorsiflexor function was normal in twenty-eight extremities (53%), partial or limited in seventeen extremities (32%), and absent in eight extremities (15%). Significant improvement was noted in ankle dorsiflexor selective control (p = 0.002) following the ankle plantar flexor surgery, with twenty extremities (38%) improving one level or greater.
Preoperative ankle dorsiflexor strength was normal or partially diminished in twenty-one extremities (40%), no greater than antigravity in twelve extremities (23%), and absent in twenty extremities (38%). Postoperative ankle dorsiflexor strength was normal or partially diminished in thirty extremities (57%), no greater than antigravity in eleven extremities (21%), and absent in twelve extremities (23%). Significant improvement was noted in ankle dorsiflexor strength (p = 0.001) following the ankle plantar flexor surgery, with sixteen extremities (30%) improving two grades or more.
Preoperative ankle plantar flexor strength (as indicated by the number of single-limb heel rises with handheld support for balance performed on demand) was normal or slightly diminished in nine extremities (17%), moderately diminished in fifteen extremities (28%), extremely diminished in four extremities (8%), and absent in twenty-five extremities (47%). Postoperative ankle plantar flexor strength was normal or slightly diminished in eight extremities (15%), moderately diminished in sixteen extremities (30%), extremely diminished in five extremities (9%), and absent in twenty-four extremities (45%). No significant difference was noted in ankle plantar flexor strength (p = 0.777) following the ankle plantar flexor surgery, with fourteen extremities (26%) improving and fourteen extremities (26%) worsening one grade or more. The remaining twenty-five extremities (47%) showed no change in strength following surgery.
The preoperative confusion test was positive in thirty-seven extremities (70%). The postoperative confusion test was positive in forty-three extremities (81%), with six extremities changing from negative to positive following ankle plantar flexor surgery. The change between preoperative and postoperative confusion test measures was significant (p < 0.001).
Clonus of the ankle plantar flexors was present in twenty-nine extremities (55%) prior to plantar flexor surgery. Clonus was present in eleven extremities (21%) following surgery, with nineteen extremities changing from present to absent and one extremity changing from absent to present. The change between preoperative and postoperative clonus measures was significant (p = 0.007).
Swing Phase Kinematic Variables
Significant improvements were noted in peak and mean ankle dorsiflexion in swing phase following the ankle plantar flexor surgery (p < 0.001 for each measure) (Table II). Thirty-four extremities (64%) showed improvement of peak ankle dorsiflexion in swing phase of >5°. Thirty-five extremities (66%) showed improvement of mean ankle dorsiflexion in swing phase of >5°.
Active ankle dorsiflexor function in swing phase was present in forty-two extremities (79%) prior to ankle plantar flexor surgery. Swing phase dorsiflexor function was present in fifty-one extremities (96%) following surgery, with ten extremities changing from absent to present and one extremity changing from present to absent. The change between preoperative and postoperative active ankle dorsiflexor function in swing phase was not significant (p = 0.299). A logistic regression model, built with use of all physical examination and kinematic variables to predict the absence or presence of active ankle dorsiflexor function in swing phase prior to surgery, achieved perfect sensitivity (100%) but poor specificity (9.09%). The classification values did not differ greatly from that of a random predictor, and in the end a significant predictor of active ankle dorsiflexor function in swing phase was not identified.
Dynamic Electromyographic Variables
Prior to surgery, normal phasic activity of the tibialis anterior muscle (on) and the gastrocnemius (off) during the first half of swing phase was noted in forty-one (89%) of forty-six extremities. Inappropriate coactivation of the tibialis anterior (on) and gastrocnemius (on) was noted in the remaining five extremities (11%). Following surgery to the ankle plantar flexor muscles, the electromyographic activity of the gastrocnemius in the first half of swing phase changed from abnormal (on) to normal (off) in five extremities, and from normal (off) to abnormal (on) in three extremities. Tibialis anterior activity in three extremities during the first half of swing phase changed from normal (on) to abnormal (off) following surgery. Due to offsetting changes in the activation patterns, the postoperative prevalence of normal phasic activation and abnormal phasic coactivation of the tibialis anterior and gastrocnemius during the first half of swing phase was unchanged (89% and 11%, respectively) following surgery to the ankle plantar flexor muscles.
Cerebral palsy is a complex upper motor neuron disorder, which produces significant secondary musculoskeletal impairments. The mechanisms by which the central nervous system lesions or structural abnormalities disrupt normal muscle function remain obscure. Classic simplistic models of excessive chronic stimulation of agonist muscles and disuse of antagonist muscles are not supported by the range of biologic, histologic, and mechanical data generated from the study of skeletal muscle in normal individuals and children with cerebral palsy19-22.
Study of the complex interrelationships among spasticity, weakness, and selective motor control is complicated by difficulties in defining each of these entities and evaluating them in children with cerebral palsy23,24. A study assessing the distribution of spasticity, joint motion deficits, and selective muscle control problems in children with cerebral palsy found wide variability of correlations between clinical measures of these three features that spanned all five levels of the GMFCS23. Those authors found the strongest correlation between selective motor control problems and gross motor function deficits (as measured by the Gross Motor Function Measure). In another recent study, significant direct effects were found between spasticity and gross motor function, strength and gross motor function, and gross motor function and functional outcome (as measured by the Pediatric Evaluation of Disability Inventory) for children with cerebral palsy. Spasticity and strength had significant indirect effects on functional outcome through effect on gross motor function25.
Dynamic electromyography has been used to study the relation between spasticity, strength, and coactivation of agonists and antagonist muscles in children with cerebral palsy during maximum voluntary isometric contraction (MVIC)26-28. Those studies have shown that coactivation of agonist and antagonist muscles is present in normal children, but is more common in children with cerebral palsy, and that coactivation is greater in adjacent agonist muscles than antagonist muscles28. Ankle plantar flexor stretch reflexes were positively correlated with coactivation ratios and were negatively correlated with ankle plantar flexor and dorsiflexor strength measures26. Neuromuscular activation and motor unit firing characteristics (firing rate, recruitment, and short-term synchronization) during maximum voluntary isometric contraction were significantly reduced in children with cerebral palsy relative to age-matched normal children, supporting the concept that weakness in cerebral palsy has a strong central component27,29,30.
Dynamometers have also been used to study the relationship between spasticity, strength, and coactivation of agonists and antagonist muscles in children with cerebral palsy27,31. Those investigators found no relation between agonist spasticity and strength or between agonist spasticity and antagonist strength. All muscles (agonists and antagonists) were weaker in children with cerebral palsy relative to age-matched normal children. Those authors postulated that weakness may be a more significant component of motor impairment than spasticity in children with cerebral palsy27.
Despite its centrality to the current paradigm of surgical management and rehabilitation of static and dynamic muscle deformities in children with cerebral palsy, the change in antagonist function following surgical lengthening of the agonist muscle has been considered in few studies5,6. Teasing out the influence of a single surgical procedure performed in the context of simultaneous multilevel surgery is always a challenge. However, the kinematic data from the current study support the long-held clinical impression that ankle dorsiflexion during swing phase is improved following ankle plantar flexor surgery in the properly selected child with cerebral palsy32-34.
There are several possible mechanisms by which this improvement may be achieved. Correction of a constraining fixed equinus contracture of the ankle plantar flexors may unmask preexisting ankle dorsiflexor function. This mechanism is supported by the data from the current study showing kinematic evidence of active ankle dorsiflexor function and electromyographic evidence of normal phasic activation of the tibialis anterior and gastrocnemius muscles, in the first half of the swing phase, in most patients (79% and 89%, respectively) prior to the ankle plantar flexor surgery. The confusion test was positive in the majority of subjects before and after surgery (71% and 80%, respectively), suggesting that surgery may improve function by the unmasking of this persistent primitive reflex.
Restoration of a more normal neurological relation between agonist and antagonist muscles is another possible mechanism explaining improved ankle dorsiflexion in the swing phase following plantar flexor surgery. The data showing significantly diminished presence of clonus of the gastrocnemius after lengthening surgery suggest a peripheral neurological correction of the muscle's exaggerated response to a quick stretch following surgical lengthening. The data showing significantly improved strength and selective motor control of the ankle dorsiflexor muscles following ankle plantar flexor surgery, and the restoration of active ankle dorsiflexor function in ten of the eleven patients in whom it was completely absent prior to ankle plantar flexor surgery, support the possibility of improvement in reciprocal inhibition between agonist and antagonist muscles.
Logistic regression modeling failed to identify any combination of predictors of active ankle dorsiflexor function in swing phase from the variables considered in the current study. This suggests that there are multiple causes for poor ankle dorsiflexor function in swing phase in children with cerebral palsy. The electromyographic data from the current study showed a lower level of coactivation between the tibialis anterior and gastrocnemius muscles during the first half of the swing phase that would have been anticipated on the basis of the prior literature examining this phenomenon26-28. In those studies, the presence of coactivation among agonist, antagonist, and other agonist muscles was established during maximum voluntary isometric contraction, which may be a poor proxy for functional activities, such as gait, that do not incorporate isometric muscle activity. The subjects in the current study were all walking at relatively well-functioning GMFCS level I or II, and it is our impression, from performing quantitative gait studies, that muscle coactivation is more common for children walking at a lower GMFCS level III.
The muscle strength data from the current study support the concept that both spastic agonist and antagonist muscles are weak in children with cerebral palsy. Unfortunately, the relative strength (or weakness) of the tibialis anterior and gastrocnemius muscles cannot be reliably determined in the current study because of the use of different strength assessment scales for these muscles in the SMALnet protocol.
The relationships among improved passive ankle motion, strength, selective motor control, diminished spasticity, and improved dorsiflexor function in swing phase following ankle plantar flexor surgery remain unclear. However, data from the literature and the current study provide a foundation or rationale for clinicians managing gait disorders in children with cerebral palsy. Weakness is common and is present in both agonist and antagonist muscles. Surgical lengthening of the muscle tendon unit, which presumably weakens it further, should only be considered for the correction of fixed muscle contractures that are recalcitrant to correction by nonoperative treatments such as manual stretching, serial stretch casting, and strength training35-37. The data from the current study support the concept that, when surgical lengthening of a spastic agonist is required, improved function of the antagonist should be expected and optimized by appropriately focused strength training and other modalities during the recovery and rehabilitation phases. The central mechanisms that disrupt gait in children with cerebral palsy are not directly addressed by peripheral interventions such as surgical lengthening of the muscle tendon unit. A better understanding of the mechanisms by which such peripheral interventions are efficacious should improve clinical decision-making for both the operative and nonoperative management of gait disorders in children with cerebral palsy.