Journal Logo

Applied Sciences: Biodynamics

Role of cocontraction in the O2 cost of walking in children with cerebral palsy


Author Information
Medicine & Science in Sports & Exercise: December 1996 - Volume 28 - Issue 12 - p 1498-1504
  • Free


Gait abnormalities in children with cerebral palsy have been shown to increase submaximal walking energy expenditure almost three-fold compared with healthy children (7). One consequence of this increased energy cost is that children with CP complain of fatigue at low walking intensities (5,11).

The explanation for this phenomenon remains incomplete. One factor that has been implicated, but not documented, is the amount of simultaneous contraction of agonist and antagonist muscle (18). Such“cocontraction” can be both beneficial and detrimental. It is a way of achieving joint stability, particularly at the ankle and knee joints and specifically during locomotion. However, in instances when the antagonist muscle is activated excessively, movement is achieved at a high metabolic cost, reflecting the effort of both muscles(12,13,28).

Attainment of stable or steady state oxygen consumption values is an important aspect of any assessment of walking and running economy in able-bodied subjects (23). The attainment of steady state values in able-bodied children has been addressed(1,9,10). However, this issue remains unresolved in people with a neuromuscular disability. In all three energy expenditure studies conducted by Rose et al.(25-27) involving children with CP, there is no mention of whether steady state has been obtained. Consequently, their rationale for choosing 2-min submaximal stages has not been explained. Therefore, to quantify the energy costs of walking, the optimal protocol that satisfies both the attainment of steady state ˙VO2 and minimizes the amount of time the subject with CP spends on the treadmill needs to be considered.

One possible reason for the paucity of data on O2 cost of walking in children with CP is the unfamiliar challenges that this type of testing presents: for example, the need to walk without assistive devices on a moving belt; to exhale and inhale through the mouth while the nose is blocked; and to seal their lips around the mouthpiece (children with CP often drool because of their inability to seal the lips). To overcome the challenges of using a mouthpiece, McLenaghan et al. (22) developed an alternative subject/instrument interface to collect metabolic data in children with CP. However, the mode of exercise chosen was cycle ergometry as opposed to treadmill walking. In light of these constraints, a major challenge of the present study was to overcome the logistical problems of actually getting subjects to cope with a setup involving simultaneous multiple measures while walking.

The aims of the study were threefold: 1) to compare the O2 cost of locomotion between children with CP and able-bodied controls, 2) to assess the contribution that cocontraction may have upon ˙VO2 in children with CP, and 3) to assess the optimal time to achieve stable submaximal˙VO2 values. The two groups were compared at one common walking speed (3 km·h-1) and at one speed that reflected a common relative effort (90% fastest walking speed (FWS)).



Eighteen subjects volunteered for the study. Nine (7 males, 2 females) were children with spastic CP (age 12.7 ± 2.8 yr, mean ± SD). Classification of CP was made according to published taxonomy(4). Seven were assessed as having diplegia, one of whom used crutches and orthotics as his usual walking aids. One child was classified as having hemiplegia and one had quadriplegia and used crutches as his normal walking aid. Nine children served as able-bodied controls (7 males, 2 females). Because of technical problems with the equipment, one control subject's data were unusable. Therefore, data on only eight able-bodied subjects (7 males, 1 females) were considered (age 13.6 ± 2.1 yr). Pubertal status was determined through the self-assessment of the Tanner stage(29) for breast development in the females and pubic hair development in the males. The validity of this technique has recently been established (21). As seen in Table 1, the two groups had similar age, height, body mass and maturational status.

Verbal consent was obtained from the children followed by a written informed consent by a parent or guardian. The study was approved by the Ethics Review Board of the McMaster University Medical Centre. Subjects completed a physical activity questionnaire modified from Bar-Or (2). This provided information on longterm (6 months) physical activity. The physical activity profiles of the two groups were very similar. Six subjects with CP were classed as “moderately active,” two as“active,” and one “very active.” Controls were classified as one “inactive,” one “moderately active,” four “active,” and two “very active.” Also, all subjects completed a pretest questionnaire detailing information regarding activity and food intake on the day of the test. At this initial screening the RJL-BIA 101A bioelectrical impedance system was used to assess body composition.

Visit 1. All subjects visited the laboratory twice. In the first visit, the children with CP completed a 15-min practice period on the treadmill (Quinton, Q65, Seattle, WA), consisting of three stages. This involved the subject walking 1) while holding onto the treadmill support bars and being supported around the waist, 2) while just being supported around the waist, and 3) totally unaided. These three periods were separated by 2-3 min of rest. By the end of the habituation, all subjects, including the two subjects with CP who normally walked using forearm crutches, were able to not hold on to the treadmill support bars for the duration of both testing sessions. However, as an additional precaution, a researcher straddled the treadmill belt behind each CP subject during each test. The first visit was identical for the control subjects except that the three-stage practice protocol was not used. However, the control subjects had a preliminary 4-min walk to become used to the treadmill. The treadmill belt had been adjusted to accommodate speeds as low as 1.68 km·h-1. The treadmill was calibrated by measuring the belt length, placing a marker on the treadmill belt, and measuring the time taken for the belt marker to make 20 revolutions. All calibrations were performed with the subjects on the treadmill, and calibrations were performed for every subject at all submaximal speeds.

Subjects then performed an 8- to 10-min maximal treadmill walking test, with collection of expired gases. As part of this protocol, the FWS was determined through progressive 2-min speed increments until the FWS was achieved. This procedure took approximately 6 min per subject. The FWS was defined as the point of transition from walk to run, as reflected by the appearance of a flight phase. After the establishment of FWS, the walking speed was reduced to 0.2-0.5 km·h-1 below the FWS and the gradient was raised every 2 min until a peak ˙VO2 was achieved(15). At the end of this first visit, electrodes were attached to the subjects to acquaint them with the electromyographic (EMG) collection equipment that would be worn during visit 2.

Visit 2

Submaximal exercise protocol. The subjects arrived at the laboratory having refrained from exercise for at least 5 h and from eating for the 2 h prior to testing. In the laboratory they rested while anthropometric measurements and questionnaires were completed. Prior to the onset of exercise, pre-exercise metabolic measures were made with the subjects seated for 4 min. A single discontinuous walking submaximal steady state protocol was then used for both groups. The protocol had two 4-min intermittent stages at 0% gradient with 8-min rest periods between. The two speeds selected were 3 km·h-1 and 90% of each child's predetermined FWS.

Electromyographic Measures

Measurement of maximal voluntary contraction (MVC). In both CP and able-bodied subjects MVC of the right quadriceps, hamstring, tibialis anterior, and soleus were performed to establish a reference for EMG data collected during the submaximal exercise protocol. MVC data were collected by having the children perform three near isometric (3°·s-1 and a 10° range of motion) knee flexion, knee extension, ankle plantar flexion, and dorsi flexion contractions on the Kincom II (Chattecx Corporation, Chattanooga, TN) isokinetic dynamometer(30).

Collection of EMG data. EMG data were collected (Watscope, Waterloo, Ontario) for a 5-s interval during the final 30 s of each exercise bout. Disposable surface electrodes (Ag/AgCl) were placed in pairs over the vastus lateralis, the middle of the hamstrings group, tibialis anterior, and soleus on the right leg. For details see Unnithan et al.(30). The mass of the electrodes, wires, and a light-weight junction box worn by the subjects was 200 g.

EMG analysis. Raw EMG data were full wave rectified, low pass filtered (3Hz cut-off frequency), and normalized to the largest value of activation observed in each muscle. This value was from the MVC, however, when greater activation was seen in the treadmill bouts than in the MVC maneuver, the former was used for normalization. It was also used for all CP subjects who were unable to attain an MVC from ankle dorsi/plantar flexion. Calculation of a cocontraction index (CI) was done by overlapping the linear envelopes of vastus lateralis and hamstrings and likewise tibialis anterior and soleus, calculating the area of overlap and dividing by the number of data points(Fig. 1). The entire 5-s interval was used for the cocontraction calculation, so that at least two complete strides or cycles would be present for both able-bodied and CP subjects. All subjects satisfied this criterion. In addition, this averaging approach across time was particularly suited to the CP subjects, who demonstrated nonsystematic gait patterns from stride to stride. The sampling frequency used was 1000 samples·s-1.

Metabolic Measures

Metabolic data were collected using a computerized metabolic cart (Beckman Horizon MMC, Anaheim, CA), which was calibrated prior to and following each test. To minimize dead space ventilation, a Hans Rudolph valve of either 20.1-, 48.9-, or 102.9-ml dead space was used, depending on body size. To minimize the discomfort of a head support for the valve, subjects wore bicycle helmets (mass 260 g) to which was connected a 3-cm internal diameter tube from the expired side of the one-way valve. This tube was led over the top of the helmet and into the expired air port of the metabolic cart.

The metabolic cart generated data points for physiologic variables every 15 s. These values were averaged for each min of the submaximal walking stages. The final minute of each submaximal walking stage was taken to represent the steady state value. Heart rate (HR) was monitored using a PE 3000 monitor(Polar Electro Fitness Technology, Kempe, Finland). The following variables were measured: absolute oxygen consumption (˙VO2abs), absolute oxygen consumption-net (˙VO2abs-net), mass-relative oxygen consumption (˙VO2rel), mass-relative oxygen consumption-net(˙VO2rel-net),% ˙VO2max, ventilation (˙VE), respiratory exchange ratio (RER), HR, and HRnet. Net measurements were defined as exercise minus pre-exercise sitting values.

Statistical Analysis

Owing to the small sample size, data were pooled for males and females. A general linear model was used to compare differences between and within speeds, groups, and muscle sites for all cocontraction variables. Simple linear regression was used to explain the variability of ˙VO2 with respect to lower leg and thigh cocontraction indices. One-way analysis of variance (ANOVA) was used to determine the differences between CP and controls at the two walking speeds and a two-way factorial analysis of variance was used to determine how soon a steady state is achieved during the 4-min stage.P < 0.05 was considered significant. Minitab statistical software was used to process the data.


No significant differences were noted between the groups for age, height, body mass,% fat, or maturational status (Table 1). Nor were there significant differences for the pre-exercise sitting˙VO2, whether expressed in relative (CP: 4.6 ± 0.85 vs Control: 4.4 ± 1.08 ml·kg-1·min-1) or absolute (CP: 0.214 ± 0.85 vs Control: 0.211 ± 0.05 l·min-1) terms. The control subjects had significantly(P < 0.05) higher peak ˙VO2 values compared to the children with CP, expressed in both relative and absolute terms (45.2 ± 8.4 vs 32.7 ± 4.8 ml·kg·min-1 and 2.29 ± 0.80 vs 1.58 ± 0.68 l·min-1).

As seen in Table 2, at 3 km·h-1, significant differences between groups were noted for ˙VO2abs-net,˙VO2rel, ˙VO2rel-net,%˙VO2max,˙VE, HR, and HRnet; in all instances the children with CP had higher values. At a relative intensity of 90% FWS, significant differences were noted only for%˙VO2max. As seen inFigures 2 and 3, the subjects with CP, but not the controls, had a positive relationship between ˙VO2 and CI. Lower leg CI in the subjects with CP explained 42.8% of the variance in˙VO2 (r = 0.663), whereas thigh CI explained 51.4% of the variance(r = 0.717).

Significantly higher (P < 0.05) CI values were generated at 3 km·h-1 for lower leg and thigh cocontraction sites in the CP group (Lower leg, CP: 22.69 ± 4.6 vs Control: 5.4 ± 1.80 and Thigh, CP: 11.49 ± 8.95 vs Control: 3.39 ± 2.07, mean ± SD). Similarly, at 90% FWS the children with CP had higher CI values at both muscle sites (Lower leg, CP: 27.04 ± 11.31 vs Control: 10.95 ± 4.26 and Thigh, CP: 16.26 ± 9.02 vs Control: 6.76 ± 3.32. General Linear Model analysis demonstrated significant main effects for group(P = 0.00). That is, irrespective of walking speed or cocontraction site, the CP subjects had systematically higher values compared with the controls. Significant main effects for speed and site were also noted(P = 0.01 and P = 0.000, respectively); i.e., systematically higher values were noted at 90% FWS irrespective of site or group, and irrespective of group/speed the leg cocontraction values were significantly higher than those seen for thigh. No interaction effects were significant.

As seen in Table 3 at a speed of 3 km·h-1, by the end of the 2nd minute 88% of the 4th minute˙VO2 was attained for the children with CP. This had increased to 94% by the end of the third minute. At 90% FWS the figures were 90 and 98% respectively. To determine how soon a steady state is established, a two-way factorial analysis of variance was conducted to establish differences in the submaximal ˙VO2 values generated at minutes 2, 3, and 4 at the two walking speeds. At 3 km·h-1, a significant group (children with CP had higher ˙VO2 values irrespective of time) effect was noted. No significant time or group-by-time interactions were achieved. At 90% FWS there was no significant group, time, or group-by-time interaction.


The major findings of this study are that: 1) higher O2 costs existed for the CP versus the control group at 3 km·h-1, but the differences diminished when the two groups were compared at 90% FWS; 2) cocontraction was a major factor in the elevated submaximal energy cost of walking in children with CP but did not affect the oxygen cost of the controls; and 3) no significant difference was found between submaximal˙VO2 values at 2, 3, or 4 min for either group at either running speed.

In accord with Campbell and Ball (7) the energy cost of walking at a given absolute walking speed was higher for the CP group compared with the controls in all ˙VO2 variables except ˙VO2 where the P value approached significance (P = 0.06). An additional indication of the proportionally higher energy cost of walking in the children with CP is reflected by the higher% ˙VO2max compared with control. At 3 km·h-1, the children with CP were working at 53.5% ˙VO2max compared with 22.5% in the controls. Hence, the implications are that the children with CP would be working at a much higher proportion of their maximal aerobic power and would fatigue more easily in prolonged exercise. The significantly higher HR and ˙VE of the CP subjects at 3 km·h-1 are coupled with the need to sustain a higher ˙VO2. While previous studies have commented upon ventilatory inefficiency resulting from chest wall distortion and possibly respiratory muscle spasticity in children with CP (17), the ventilatory equivalent in the present study was not significantly higher in the group with CP. Local muscle factors most likely represent the single most important reason for the elevated O2 cost of walking in children with CP. This claim is substantiated by data from Dahlbäck and Norlin(11), who demonstrated at moderate walking speeds in children with CP that exhaustion occurred when RER, ventilatory equivalent of oxygen, and blood lactate were at levels normally seen at 50-60%˙VO2max. Further support for the importance of local factors is derived from work by Lundberg (17), who concluded that differences in mechanical efficiency between those with CP and controls during cycling stemmed from the different energy expenditures associated with involuntary local muscle activity.

Based upon CI from both lower leg and thigh, cocontraction of these muscle groups appears to contribute to the higher energy cost of walking seen in the CP subjects at 3 km·h-1. The source of the pathological pattern of muscle activity seen in the children with CP is open to speculation. It could be a consequence of an impaired maturation of the locomotor pattern and neuronal adaptation to altered peripheral muscle function in early infancy before walking is learned (6). Alternatively, as a result of the insult to the brain stem and cerebellum at birth, normal inhibition patterns could be affected. The absence of normal inhibition patterns(3) gives rise to unsuccessful repression of undesired contractions generating excessive cocontraction levels, abnormal gait, and a marked increase in energy consumption (3,20). While the higher oxygen cost of walking can be explained by cocontraction in the children with CP, the same cannot be said for the control subjects. As reported by Gatev (13) and Sutherland et al.(28) there appears to be a certain amount of cocontraction that is beneficial for joint stability and that is associated with a certain O2 cost. Beyond this level, as seen inFigures 1 and 2, any additional cocontraction may induce an excessively high oxygen cost of walking.

It has been established that children have faster response kinetics for oxygen consumption than adults (19). However, little is known about the transients of gas exchange in children with chronic disease(8) or with a motor disability. The results from the present study indicate that a 2-min submaximal protocol would provide a reasonable estimate (88 and 90%) of steady state ˙VO2. However, perhaps the real determinant for the length of the submaximal stage is the level of discomfort that the child with CP should face on the treadmill while still attaining a reasonable estimate of steady state ˙VO2.

In experiments by Rose et al. (25-27) O2 cost of locomotion was determined among children with CP and able-bodied controls. It is hard, however, to compare findings of those studies with this investigation. The main difference is that Rose et al. allowed their subjects to hold on to the handrails as necessary while walking. This has direct implications for the natural gait and for ˙VO2. Supporting the arm on the treadmill prevents reciprocal arm-swing, disrupts gait patterns, and almost certainly reduces the energy cost of walking(24). Green and Foster (14) demonstrated that the level of reduction in oxygen cost of treadmill exercise depends upon the degree by which the subjects grasp the handrails. It has been clearly demonstrated in the comparison of cardiorespiratory measures among different age groups of children that growth differences can give rise to differences in running and walking energy expenditure(16). Hence, to demonstrate cardiorespiratory differences resulting from a disability (CP) rather than the confounding effects of growth, the subjects need to be similar in age, height, and mass. In the Rose et al. study (25) the groups differed in mass and height, and there was no indication of any between-groups statistical comparison. In addition, the fitness levels of the children with CP were not documented in any of the studies by Rose et al. (25-27).

In conclusion, the sources of the greater O2 cost of walking in children with CP have yet to be fully explained. Simultaneous electromyographic and metabolic analysis in this study has established cocontraction of antagonist muscles as a factor that partially explains this phenomenon. More research incorporating kinematic data may yield further insight into factors that cause the high O2 cost. One practical limitation of the methodological design used in the present study was the inability of a child with CP and a mental handicap to cope with the demands of our testing regimen. Consequently, the development of test procedures for the child with CP and a mental handicap needs to be considered.

Figure 1-Sample illustration of cocontraction index (CI) calculation. Calculation of the cocontraction index involved overlapping the linear envelopes of muscle 1 and muscle 2 calculating the area of overlap and dividing by the number of data points. In this generic example, this gave a CI of 6.5. Dotted line illustrates average cocontraction index over time. Reprinted with permission. :
Ref. 30 . V. B. Unnithan, J. J. Dowling, G. Frost, B. Volpe Ayub, and O. Bar-Or. Cocontraction and phasic activity during gait in children with cerebral palsy. EMG Clin. Neurophysiol. (in press).
Figure 2-˙VO2 versus lower leg cocontraction index at 3km·h-1: CP versus Controls.
Figure 3-˙VO2 versus thigh cocontraction index at 3km·h-1: CP versus Controls


1. Armon, Y., D. M. Cooper, R Flores, S. Zanconato, and T. J. Barstow. Oxygen uptake dynamics during high-intensity exercise in children and adults. J. Appl. Physiol. 70:841-848, 1991.
2. Bar-Or, O. Pediatric Sports Medicine For The Practitioner: From Physiologic Principles To Clinical Applications, New York: Springer-Verlag, 1983, pp. 10-11.
3. Basmajian, J. Motor learning and control: a working hypothesis. Arch. Phys. Med. Rehabil. 58:38-41, 1977.
4. Bax, M. C. O. Terminology and classification of cerebral palsy. Dev. Med. Child. Neurol. 6:295-307, 1964.
5. Berg, K. Effect of physical training of school children with cerebral palsy. Acta. Paediatr. Scand. Suppl. 204:27-33, 1970.
6. Berger, W., J. Quintern, and V. Dietz. Pathophysiology of gait in children with cerebral palsy. Electroencephalogr. Clin. Neurophysiol. 53:538-548, 1982.
7. Campbell, J. and J. Ball. Energetics of walking in cerebral palsy. Orthop. Clin. North Am. 9:374-377, 1978.
8. Cooper, D. M. Development of the oxygen transport system in normal children. In: Advances in Pediatric Sport Sciences, Vol. 3, O. Bar-Or (Ed.). 1989, pp 67-96.
9. Cooper, D. M., M. R. Kaplan, L. Baumgarten, and D Weiler-Ravell. Coupling of ventilation and CO2 production during exercise in children. Pediatr. Res. 21:568-572, 1987.
10. Cooper, D. M., C. Berry, N. Lamarra, and K Wasserman. Kinetics of oxygen uptake at the onset of exercise as a function of growth in children. J. Appl. Physiol. 59:211-217, 1985.
11. Dahlback, G. O. and R. Rorlin. The effect of corrective surgery on energy expenditure during ambulation in children with cerebral palsy. Eur. J. Appl. Physiol. 54:67-70, 1985.
12. Frost, G., J. Dowling, O. Bar-Or, and K. Dyson. Comparison of cocontraction in three groups of children during treadmill running. Can. J. Appl. Physiol. 19:P16:1994.
13. Gatev, V. Role of inhibition in the development of motor coordination in early childhood. Develop. Med. Child Neurol. 14:336-341, 1972.
14. Green, M. A. and C. Foster. Effect of magnitude of handrail support on prediction of oxygen uptake during treadmill testing(Abstract). Med. Sci. Sports Exerc. 23:S166, 1991.
15. Hoofwijk, M., V. Unnithan, and O. Bar-Or. Maximal treadmill performance in children with cerebral palsy. Pediatr. Exerc. Sci. 7:305-313, 1995.
16. Krahenbuhl, G. S., D. W. Morgan, and R. P. Pangrazi. Longitudinal changes in distance -running performance of young males.Int. J. Sports Med. 10:92-96, 1989.
17. Lundberg, A. Longitudinal study of physical working capacity of young people with spastic cerebral palsy. Dev. Med. Child. Neurol. 26:328-334, 1984.
18. Lundberg, A. Oxygen consumption in relation to work load in students with cerebral palsy. J. Appl. Physiol. 40:873-875, 1976.
19. Macek, M. and J. Vavra. The adjustment of oxygen uptake at the onset of exercise: A comparison between prepubertal boys and young adults. Int. J. Sports Med. 1:70-72, 1980.
20. Mann, R. Biomechanics in cerebral palsy. Foot Ankle 4:114-119, 1983.
21. Matsudo, S. M. M. and V. K. R. Matsudo. Physician assessment of sexual maturation in Brazilian boys and girls: concordance and reproducibility. Am. J. Hum. Biol. 6:451-455, 1994.
22. McLenaghan, B., S. Hill, and C. Okazaki. Interface design for indirect calorimetry in children with cerebral palsy. Arch. Phys. Med. Rehabil. 69:548-551, 1988.
23. Morgan, D. W., P. E. Martin, and G. S. Krahenbuhl. Factors affecting running economy. Sports Med. 7:310-330, 1989.
24. Ragg, K. E., T. F. Murray, L. M. Karbonit, and D. A. Jump. Errors in predicting functional capacity from a treadmill exercise stress test. Am. Heart J. 100:581-583, 1980.
25. Rose, J., W. L. Haskell, and J. G. Gamble. A comparison of oxygen pulse and respiratory exchange ratio in cerebral palsied and nondisabled children. Arch. Phys. Med. Rehabil. 74:702-705, 1993.
26. Rose, J., J. G. Gamble, A. Burgos, J. Medeiros, and W. L. Haskell. Energy expenditure index of walking for normal children and for children with cerebral palsy. Dev. Med. Child Neurol. 32:333-340, 1990.
27. Rose, J., J. G. Gamble, J. Medeiros, A. Burgos, and W. L. Haskell. Energy cost of walking in normal children and in those with cerebral palsy: comparison of heart rate and oxygen uptake. J. Pediatr. Orthop. 204:276-279, 1989.
28. Sutherland, D. H., R. Olshen, L. Cooper, and S. Woo. The development of mature gait. J. Bone Joint Surg. 62A:336-353, 1980.
29. Tanner, J. M. Growth and Adolescence, 2nd Ed. Oxford: Black-well, 1962.
30. Unnithan, V. B., J. J. Dowling, G. Frost, B. Volpe Ayub, and O. Bar-Or. Cocontraction and phasic activity during gait in children with cerebral palsy. EMG Clin. Neurophysiol. (in press).


©1996The American College of Sports Medicine