Effects of a Postoperative Resistive Exercise Program on the Knee Extension and Flexion Torque in Children With Cerebral Palsy: A Randomized Clinical Trial

Participants 

The study included 43 children with diplegic CP to the lower limbs, between the ages of 6 and 16 years (mean age ± standard deviation [SD], 9.7±2.8y). They were classified from level I to III according to the Gross Motor Function Classification System (GMFCS).24 We recruited all participants as preoperative ambulatory patients from the CP clinic of an orthopedic university hospital, where the study was conducted. They had not previously had surgery and were candidates for a multilevel procedure. Patients who were mentally or physically incapable of following the exercise program were excluded. Preoperatively, 9 patients required external assistance for walking. Written informed consent for participation was obtained from the parents. The study was approved by the local ethics committee.

Study Design 

We randomized the patients into 2 groups (table 1, fig 1): the exercise group and the control group. Before the beginning of the study, a member of the research team prepared 20 notes for each group assigned by the group name. These notes were sealed in 40 identical, opaque envelopes, which were mixed and stored in a safe place. While the patient was in the operating room, an independent person selected 1 envelope, which was not put back. The envelopes of patients that were originally allocated but excluded within the first 6 months of the study (due to injuries, drop-outs, missing examination dates) were prepared again and added to the envelopes that remained at the time when the decision for exclusion was made. Full datasets of all measurements were not available for the excluded patients and therefore no intention-to-treat analysis was performed.

Table 1. Subject Characteristics for the Exercise Group (EG) and Control Group (CG) Before Surgery

	Characteristics	EG (n=19)			CG (n=20)			EG–CG 95% CI⁎
	Age (y)	10.6±3.3			8.9±1.9			−0.1 to 3.4
	Body height (cm)	137±17			129±11			−0.4 to 18.0
	Body mass (kg)	33.3±12.6			28.5±8.9			−2.2 to 11.8
	Duration of the hospital stay (wk)	7.8±2.2			7.9±2.1			−1.4 to 1.4
								Chi Square
	Sex (boys/girls)	13/6			14/6			χ2 test=0.1, P=.81
	Assistance during walking (with/without)	2/17			7/13			χ2 test=2.1, P=.15
	GMFCS (level and no. of patients)	I	II	III	I	II	III	χ2 test=3.6, P=.16
		6	11	2	6	7	7

NOTE. Values are mean ± SD.

⁎ Statistical results of the unpaired t test (95% confidence intervals [CIs]) and the chi-square test for the difference between the groups.

View Large Image Download to PowerPoint

Fig 1. Flow diagram of the randomization procedure. Adapted from Patikas.37 Reprinted with permission. ©American Congress of Rehabilitation Medicine and American Academy of Physical Medicine and Rehabilitation.

Both groups received conventional PT after surgery as soon as mobilization was possible. During the hospital stay, the PT program was conducted by 4 physiotherapists employed at the clinic where the study was carried out. After hospital discharge, PT was carried out by the physiotherapist who had regularly treated this child before surgery. These physiotherapists were available to answer questions about the exercise program, with respect to their professional experience. If the question was too specific to the exercise program, we recommended contacting the research team.

The exercise group underwent a training program in addition to conventional physiotherapy. Strength evaluation25 and the MAS14, 15 were assessed before surgery (E0) and approximately 6 months (E½) and 1 year (E1) after surgery (E½, .41±.05y; E1, .93±.06y). For the exercise group, the examination dates E½ and E1 corresponded to 12.8±1.4 and 40.3±1.9 weeks of training at home, respectively, and E1 was conducted as soon as the training program ended. GMFM scores were assessed only at E0 and E1.26 The study period, from the first to last measurement, was almost 3 years (March 2001–January 2004).

Surgical Treatment 

The surgical treatment for all patients (exercise group, control group) included soft-tissue and bony procedures (table 2). These procedures were performed for each patient in a single-event multilevel operation by 2 surgical teams.27 The decision for surgical treatment was made after analyzing video data, 3-dimensional gait, and dynamic electromyographic recordings and with respect to the clinical, radiologic, and intraoperative examination results.3

Table 2. Surgical Procedures Carried Out in the Exercise Group (n=38 limbs) and Control Group (n=40 limbs)

	Procedure	EG	CG
	Soft-tissue procedures (total)	127	144
	Intramuscular psoas tendon lengthening	5	3
	Proximal intramuscular rectus femoris recession	5	13
	Distal transfer of the rectus femoris tendon	34	38
	Hamstring lengthening (total)	31	33
	Medial hamstring lengthening only	25	31
	Medial and lateral hamstring lengthening	6	2
	Equinus correction (total)	27	38
	Baumann procedure	21	32
	Strayer procedure	4	4
	Open Achilles’ tendon lengthening	2	2
	Additional soft-tissue procedures for foot-deformity correction	25	18
	Bony procedures (total)	65	56
	Pelvic osteotomy (Dega/Pemberton)	0	4
	Femoral derotation osteotomy (total)	33	24
	Supracondylar	26	19
	Intertrochanteric	7	5
	Tibial derotation osteotomy	2	7
	Additional bony procedures for foot-deformity correction	30	21
	Overall total	192	200

Knee flexion deformities were corrected by intramuscular and aponeurotic lengthening. To correct the stiff gait pattern (cospasticity of the knee flexors and extensors, especially the rectus femoris),28 the rectus femoris muscle tendon was transferred distally to the tendon of the gracilis muscle.29, 30 We corrected equinus deformities using the Baumann procedure or, in more severe cases, using the Strayer procedures; rarely, open lengthening of the Achilles’ tendon was performed.31 Bony procedures were used to correct any hip (sub)luxations, internally rotated gait, rotational deformities of the tibia, and bony foot deformities. Pes valgus, pes varus, or pes cavus deformities were corrected in order to create plantigrade and stable feet.

Training Protocol 

The training sessions for the exercise group started 3 to 4 weeks after surgery, when it was no longer painful to perform the exercises and there was no danger of recurring injury. During the last 3 to 4 weeks of the hospital stay, 2 physiotherapists (members of the research group) taught and supervised the training protocol. They also gave instructions to the children’s parents about executing the exercises after hospital discharge and a detailed written description of the exercises was given to the parents before the child left the hospital. The parents completed a logbook, providing information about the frequency of the training sessions at home (duration, 8.70±0.95mo). At least twice a month the research team contacted the parents at home by telephone to clarify potential issues about the assessment of the training program and to learn whether any adverse effects had developed.

We instructed the children in the exercise group to carry out the training program at least 3 times a week, with an optimal target of 4 times a week. Each training session was 30 to 45 minutes long (depending on the child), and consisted of 7 exercises for both sides (fig 2). Two sets of 5 repetitions for each exercise were performed, with approximately a 1-minute rest between each set and drill. The movement velocity was approximately 4 to 5 seconds per repetition, including slow return to the initial position in order to evoke eccentric muscle activation. The resistance was progressively increased by gradually eliminating the external support during the exercise. As soon as movements against gravity without assistance were feasible, the parents increased further the resistance for the exercises 2, 4, and 5 using elastic rubber bands.a Additional rubber-band layers were applied if the child could repeat a whole set without compensatory movements from other muscle groups. For exercises 1 and 7, the thighs were fastened together distally with rubber bands to prohibit excessive hip abduction.

View Large Image Download to PowerPoint

Fig 2. Schematic description of the training drills.

Testing Procedures 

The examinations included isometric and isokinetic tests for the knee extensors and flexors, bilaterally, using an isokinetic dynamometer.b The muscle groups we selected for testing were surgically treated in most patients (see table 2) and are functionally important during gait. The patients sat on the dynamometer chair, strapped over the trunk and distally to the thigh. The dynamometer pad was fixed above the malleoli.

Prior to each test, we conducted a trial session to familiarize the patient with the testing protocol. For the isometric measurement the knee was fixed at 90° of flexion.9 Three knee extensions and flexions were performed, with a 30- to 60-second rest between each trial. The isokinetic test was assessed concentrically, at 60° and 180°/s, with movement ranging from 100° of knee flexion to full knee extension. We tested the knee extensors and flexors separately because some children had difficulties alternating extension and flexion movements. Three sets of 5 repetitions were assessed for each movement direction and each angular velocity. The patients rested 30 to 60 seconds between each set. If the patient failed to perform the test efficiently, we repeated the test after 2 to 3 minutes of rest and further familiarized the child with the procedure when necessary. During testing, the children were allowed to grasp the side grips of the chair,32 had real-time visual feedback of the joint moment, and were verbally encouraged by the examiner.

The effect of gravity was considered and corrected for when calculating the joint moment. The peak joint moment was then determined and the best value from all repetitions in each testing mode and movement direction was calculated. The average of the 2 sides was statistically analyzed. For the isokinetic trials, only the isokinetic part of the movement was taken into consideration. Many of the children were unable to reach the angular velocity of 180°/s and, therefore, this angular velocity was not further analyzed. Furthermore, for technical reasons, data from at least 1 isometric measurement from 9 patients could not be retrieved and therefore these missing data were excluded case-wise from the statistical analysis for the isometrics.

We evaluated the resistance of the knee flexors and extensors to passive movements bilaterally using the MAS.14, 15 Furthermore, we measured the GMFM scores16 and the passive ROM of the knee in a standardized fashion, with the subject supine and the hip extended or flexed at 90° for the knee extension or flexion, respectively.33 All measurements in a particular subject were made by the same person. The 3 assessors were not aware of the results of previous measurements. Group allocation was concealed from the assessor measuring torque.

Statistics and Data Analysis 

Joint moments were normalized to body mass17 and the ratio of extension to flexion moment determined.9 The GMFM sections D (standing) and E (walking, running, jumping) were analyzed separately, in addition to the total score. All GMFM scores were expressed in percentage of the maximum possible achieved scores for each section, and the total GMFM score consisted of the average of all sections.

To determine the size of the 2 groups, we conducted a power analysis with an effect size of 1, 2-tailed α of .05, and power of .80, which resulted in 17 patients per group. Considering a possible 15% drop-out rate during the experiment, we decided to recruit 20 patients per group.

The outcome variables were analyzed using a 2×3 analysis of variance (2-way), with group as a between factor (group: exercise group vs control group) and time of the 3 examinations as a within factor (time: E0 vs E½ vs E1). To adjust the preoperative differences between the groups, we applied a 2×3 analysis of covariance (2-way ANCOVA) as a secondary analysis for the joint moment data, taking the GMFM scores (sections D, E, and total) evaluated at E0 as covariates. In this case the estimated means were calculated and adjusted for the covariates. Post hoc tests were performed using the Bonferroni adjustment for multiple comparisons. Furthermore, we assessed the differences between the 2 groups at E0, for age, body mass, height, and hospital stay using unpaired t tests and for sex, ability to walk with and without external support, and the GMFCS using chi-square nonparametric statistics. Alpha was set at .05 and the 95% confidence interval (CI) for the difference between the groups and between the examinations was calculated. The analysis was performed with the statistical package SPSS.c

Study Limitations 

There are several limitations of this study to be discussed. Although the sample size selected was large enough to achieve a power of 0.8 for a single factor, the power for the group by time interaction was much lower (≈0.4). Hence, even if an interaction effect did exist, it would be unlikely to be detected. To increase the power for the interaction (>0.8), a sample of 100 patients should be randomized; however, this could not be achieved during this 3-year, single-center research project.

Surgical intervention considerably reduces muscle efficiency in the postoperative stage. On the other hand, the training intensity that could be applied was probably not sufficient to overcome this deficit. To achieve adaptations after strength training, at least 65% of maximum load is required.34 In 4 of 7 exercises from the current study, the load was not progressively increased until the patient could execute the exercise without assistance. Even though our aim was to improve the muscle strength in the sense of explicit strength training, we had to make a compromise concerning the intensity of training, taking into account the special situation of neuromuscular disorder and the difficulties during the postoperative phase. For this reason we differentiate our training from the typical strength training, by calling it “resistive exercise.” Furthermore, the resistance of the rubber band increases proportionally with distance and this could be less optimal than isokinetic12, 13 or isotonic9, 10, 11 loading at different joint angles. Moreover, according to the principle of training specificity, the gain in torque for the exercise group may be underestimated because the muscular strength was not measured under the training conditions. Finally, it is ethically inappropriate to discourage or forbid the control group children to be more physically active because there is no argument to support that this could have any adverse effects for children with CP.

Concerning the homogeneity of surgical treatment, although distal transfer of the rectus femoris tendon and lengthening of the knee flexors was performed in almost all patients, the qualitative and quantitative discrepancy between the different surgical procedures among the patients (see table 2) and the outcome of the surgery could only be predicted to a certain degree. It is worth mentioning that, despite the higher GMFM scores, more bony procedures were performed in the exercise group than in the control group (56 vs 65), procedures which generally require a longer low weight-bearing and rehabilitation period.

The above limitations may bias the contrast between the 2 groups and therefore no systematic between-groups differentiation was observed. Such limitations were not present to this extent (if at all) in previous nonrandomized, short-term studies, or even studies without a control group (for a review, see Dodd et al6). Only recent published studies35, 36 attempted to create randomized groups and compare the effects of strength training taking a control group of patients with CP as a reference. According to these studies, patients with an age range similar to those in the present study underwent a home-based, 12-week strength training program and showed a marginal improvement in strength, although the functional and psychologic benefits leading to improvement in quality of life should be remarked.