Cerebral palsy (CP) is defined as a group of permanent disorders of movement and posture causing activity limitations attributed to a static lesion in the developing brain often accompanied by secondary impairments. Predominant clinical manifestations found in CP include weakness, loss of selective motor control, spasticity, and antagonist cocontraction.1 Significant impairments caused by this disorder may compromise motor function, and as a result, individuals with CP experience functional limitations that affect activities of daily life ranging from mild incoordination to total body involvement.2
Muscle weakness is one of the most common impairments related to function in children with CP. Substantial weakness, even among those who are mildly involved, has been noted in this population when compared with peers who are developing typically.3 Secondary to their antigravity role, the quadriceps are affected by unloading and disuse4,5 and are particularly important for control during the stance phase of gait and other functional activities.6,7 Furthermore, components of the Gross Motor Function Measure, including standing, walking, running, and jumping, have been shown to be significantly correlated to the strength of the quadriceps alone.7 Because muscle size is known to be directly proportional to the force generating capacity of skeletal muscle,8 precise measures of quadriceps muscle thickness are useful when investigating weakness in children with CP. In healthy adults, muscle thickness has been reported to be a better predictor of isometric elbow flexor force than cross-sectional area9 and has been reported to be highly correlated to muscle volume10 and cross-sectional area.9,11 Accurate measures of voluntary strength are often difficult to obtain in youth with CP because of diminished selective motor control and cognitive, behavioral, or other communicative impairments. Measurements of muscle size, such as muscle thickness, may, therefore, be an alternative way to evaluate strength in children with CP. In fact, we have previously shown that muscle thickness of the rectus femoris (RF) and vastus lateralis is highly predictive of maximum torque and has the potential to serve as a surrogate measure of voluntary strength.12
With the use of musculoskeletal ultrasound imaging techniques, muscle thickness measurements of the quadriceps can be performed noninvasively and directly in CP.12,13 In this study, 2 different measurement methods were used to determine the muscle thickness of the RF. One measure of muscle thickness, MT50, was obtained at 50% of thigh length and the second measure, maximum muscle thickness (MaxMT), was obtained through multiple measures along the length of the muscle. Traditionally, 50% of thigh length is estimated to be the maximum cross-sectional area of the quadriceps. Therefore, muscle thickness measured at 50% of thigh length is frequently reported in the literature in both subjects who are healthy and subjects who are impaired.9,11–14 However, it is unknown whether this relationship is similar in CP. Therefore, the purpose of this study was to compare the 2 measurement methods, MT50 and MaxMT, to determine agreement between the measures of the RF thickness in children with CP and to determine whether the two can be used interchangeably.
Thirteen children with CP (14.4 ± 3.6 years) who were ambulatory and part of a randomized clinical trial participated in the study. Participants were within Gross Motor Function Classification System levels I (n = 7), II (n = 1), and III (n = 5). The study was approved by our Institutional Review Board, and parental consent and child assent were obtained for each participant.
Real-time ultrasound imaging (GE Logiq i) of the RF was obtained bilaterally in 2-dimensional B-mode. Both lower extremity measurements were taken with participants in the supine position with knees comfortably resting in the natural resting position. During scanning, participants were instructed to relax their muscles. If a muscle contraction did occur, it was detected in real time and the image was discarded. Gel was applied generously to the skin to eliminate compression or distortion of the muscle and to aid in acoustic coupling. First, 50% of the distance between the superior boarder of the patella and the anterior superior iliac spine (MT50) was measured and marked with a surgical pen on the skin as a horizontal line. Next, the mid-belly of the RF was located with the probe oriented axially at the MT50 line. Once the mid-belly was located, the probe was then turned perpendicular to the plane with the probe oriented longitudinally, and the position of the probe was marked on the skin as illustrated in Figure 1.13 Three MT50 RF images were taken at the measurement site and averaged. The probe was then slowly moved proximally from the MT50 measurement line toward the insertion of the muscle at the anterior inferior iliac spine and recorded as a video clip. Each video was approximately 6 seconds in duration, and muscle thickness measurements were obtained every 2 milliseconds until the MaxMT of the RF was determined. Once the muscle began to clearly taper with 5 successive measurements decreasing in muscle thickness, measurement stopped. For both methods, muscle thickness was measured as the distance between the superficial and deep aponeurosis as illustrated in Figure 1. A single, experienced examiner performed all of the ultrasound assessments. However, a separate examiner calculated the measures from the images and video. Intraclass correlation coefficients of 0.99 for intrasession reliability have been previously reported for RF muscle thickness in children and adolescents with CP using the 2-dimensional ultrasound imaging methods described previously.13
Many method-comparison studies are often incorrectly analyzed using correlation coefficients and t tests and can be misleading. Correlation reflects the strength of the relationship between 2 variables but not the agreement between them. The appropriate statistical analysis for assessing the agreement between 2 methods of clinical measurement has been reported by Bland and Altman.15 This type of analysis should be conducted to establish whether 2 methods can be used interchangeably or a new method can replace an old one. Therefore, the data from this study were analyzed using the Bland-Altman method with 95% limits of agreement (LOA). The first step was to construct a scatterplot with a 45° line of equality to visually inspect the data. The differences between the ultrasound MT50 and MaxMT measurements taken for each subject were then plotted against the mean differences of the measures (Bland-Altman plot). For acceptable agreement, the 95% LOA (±1.96 SD of the mean difference) should include 95% of the differences between the 2 methods of measurement.15 Acceptable tolerances for the absolute differences between the 2 methods were evaluated using a tolerance table.16 The tolerance table sets acceptable tolerances for the differences between the 2 measurements and the percentage of measures that would fall within the tolerance levels. For the purposes of this article, the left and right legs were considered independent of one another.
Despite the fact that MT50 and MaxMT were highly correlated (r = 0.98; P < .001) as illustrated in Figure 2a, the 45° line of equality in Figure 2b illustrates that most measures are above the line, indicating that the majority of MaxMT measurements were higher. The difference between the ultrasound MT50 and MaxMT measurements taken for each subject were plotted against the mean of the measures as illustrated in the Bland-Altman plot in Figure 3. For this analysis, all data points fell within the 95% LOA except for a single outlier, indicating excellent agreement. However, a constant bias toward higher MaxMT values was observed with a mean difference between the 2 measures of 0.09 cm with a 95% LOA of −0.11 to 0.28 cm. On average, MaxMT was 1.78 ± 0.48 cm and MT50 was 1.70 ± 0.46.
A tolerance table was constructed to further assess agreement between the 2 measures as illustrated in Table 1. The tolerance table sets acceptable tolerances for the differences between the 2 measurements and the percentage of measures that would fall within the tolerance levels.16 For example, Table 1 demonstrates that at least 88.5% of the measurements agree on the 0.2-cm tolerance level and 96.2% agree on the 0.3-cm tolerance level. It is up to the researcher or clinician to determine what would be an acceptable difference between the measures when deciding which measurement method to use.
Valid measures of muscle size in a population such as CP, where atrophy and weakness are prevalent, have strong clinical significance. We previously reported that MT50 of the RF is only 68% of an age-matched group of typically developing youth.13 The range of values reported here is similar to that previously reported for the MT50 measure in children and adolescents with CP (range: 0.89–2.04 cm) and is representative of the differences in functional abilities across subjects.12,13 Because the size of a muscle is directly proportional to the strength capacity of the muscle, muscle thickness can be a useful measure of strength. Although muscle thickness cannot account for neural activation deficits, we have previously reported that RF muscle thickness measured with the MT50 method explains 32% of the variance in isometric peak torque of the quadriceps while controlling for age whereas vastus lateralis muscle thickness explained 85% of the variance while controlling for Gross Motor Function Classification System level and age.12 In addition, the MT50 method can be an important outcome measure to assess potential changes in muscle size before and after an intervention.17 Noninvasive measures of muscle thickness and other muscle architectural parameters, using ultrasound imaging, have allowed researchers to examine differences between children with CP and their peers who are typically developing across a wide range of ages through cohort comparison studies.13,18,19 A better understanding of the mechanisms underlying force generation impairments will ultimately lead to improved interventions addressing these impairments.
According to the results of the Bland-Altman analysis, all data points were within the 95% LOA, except for a single outlier, demonstrating excellent agreement. However, there was a constant bias toward higher MaxMT values, meaning that the difference between the 2 methods was consistent, regardless of the thickness of the muscle, and we are 95% confident that the difference lies between 0.11 cm lower and 0.28 cm higher (LOA). This occurrence is also easily visualized in Figure 3, where the majority of values for the differences between the 2 measures (MaxMT − MT50) are located above the zero line and also can be observed in Figure 2b. Therefore, muscle thickness of the RF taken with the commonly used MT50 method at 50% thigh length is, on average, 0.09 cm less than the maximum thickness as determined with the MaxMT method. This implies that the true maximum thickness of the RF lies proximal to 50% of the distance between the anterior superior iliac spine and the superior pole of the patella. However, this could be influenced by the presence of patella alta. Depending on the research or clinical question, this may or may not be of concern. Practical applications must be considered rather than emphasizing whether or not agreement obeys a strict statistical analysis.
We have provided a table with various tolerance levels so that one can see the percentage of values that fall within a tolerance level, depending on the particular question (Table 1). For example, if it is important to obtain the true maximum thickness of the RF, one must determine an acceptable tolerance level when using the MT50 method. If your acceptable tolerance level for differences between MT50 and MaxMT is 0.1 cm, is it acceptable that only 58% of the measurements agree on this tolerance level? For this example, 0.1 cm is equivalent to 6% of the average thickness value reported here. If it is more important to have a standardized site of measurement for muscle thickness that is less time consuming, then the MT50 method may be the most appropriate choice, also given the excellent agreement with MaxMT. In addition, Bland and Altman have proposed that with a constant bias, the mean difference value can be simply added or subtracted to the appropriate measure.20 For example, when using the MT50 method, you could add 0.09 cm to the measured value to obtain the MaxMT value.
Although we have proposed that muscle thickness can be used as a surrogate measure of strength, a limitation of this measure is that it cannot account for differences in neural activation.
In addition, although great care was taken to align the probe in the mid-belly of the muscle in the mediolateral direction, the field of view was limited to this plane in 2 dimensions and did not encompass the entire muscle belly.
The measurement of muscle thickness, using ultrasound imaging, is a less time-consuming, less costly, and more practical way to measure the force-generating capacity of the muscle as compared with other methods such as magnetic resonance imaging and computed tomographic scans. Ultrasound imaging is also more likely to be available in an outpatient setting, allowing for greater accessibility for our patients and families, and could be easily implemented in a rehabilitation setting with proper training. These methods become increasingly important when assessing the underlying muscle adaptations to interventions in youth with CP. In addition, these methods may provide a surrogate measure for strength in those with CP who have severe cognitive, visual, hearing, motor control, or other issues that preclude the ability to perform objective tests of muscle strength.
In conclusion, it is up to the researcher or clinician to decide whether the differences between the 2 measurement methods are clinically significant as determined by the tolerance table and depending on the research question. Given the more time-consuming nature of measuring multiple frames from video as done for the MaxMT measure, the MT50 measurement may be a more feasible alternative when estimating MaxMT of the RF in children and adolescents with CP.
The authors thank Katy Holthaus for assistance with data collection and processing.
1. Bax M, Goldstein M, Rosenbaum P, et al. Proposed definition and classification of cerebral palsy, April 2005. Dev Med Child Neurol. 2005;47:571–576.
2. Damiano DL, Martellotta TL, Quinlivan JM, Abel MF. Deficits in eccentric versus concentric torque in children with spastic cerebral palsy. Med Sci Sports Exerc. 2001;33(1):117–122.
3. Wiley ME, Damiano DL. Lower-extremity strength profiles in spastic cerebral palsy. Dev Med Child Neurol. 1998;40(2):100–107.
4. de Boer MD, Maganaris CN, Seynnes OR, Rennie MJ, Narici MV. Time course of muscular, neural and tendinous adaptations to 23 day unilateral lower-limb suspension in young men. J Physiol. 2007;583(Pt 3):1079–1091.
5. de Boer MD, Seynnes OR, di Prampero PE, et al. Effect of 5 weeks horizontal bed rest on human muscle thickness and architecture of weight bearing and non–weight bearing muscles. Eur J Appl Physiol. 2008;104(2):401–407.
6. Mizner RL, Snyder-Mackler L. Altered loading during walking and sit-to-stand is affected by quadriceps weakness after total knee arthroplasty. J Orthop Res. 2005;23(5):1083–1090.
7. Goh HT, Thompson M, Huang WB, Schafer S. Relationships among measures of knee musculoskeletal impairments, gross motor function, and walking efficiency in children with cerebral palsy. Pediatr Phys Ther. 2006;18(4):253–261.
8. Wickiewicz TL, Roy RR, Powell PL, Edgerton VR. Muscle architecture of the human lower limb. Clin Orthop Relat Res. 1983;(179):275–283.
9. Akagi R, Kanehisa H, Kawakami Y, Fukunaga T. Establishing a new index of muscle cross-sectional area and its relationship with isometric muscle strength. J Strength Cond Res. 2008;22(1):82–87.
10. Miyatani M, Kanehisa H, Kuno S, Nishijima T, Fukunaga T. Validity of ultrasonograph muscle thickness measurements for estimating muscle volume of knee extensors in humans. Eur J Appl Physiol. 2002;86(3):203–208.
11. Sipila S, Suominen H. Ultrasound imaging of the quadriceps muscle in elderly athletes and untrained men. Muscle Nerve. 1991;14(6):527–533.
12. Moreau NG, Simpson KN, Teefey SA, Damiano DL. Muscle architecture predicts maximum strength and is related to activity levels in cerebral palsy. Phys Ther. 2010;90(11):1619–1630.
13. Moreau NG, Teefey SA, Damiano DL. In vivo muscle architecture and size of the rectus femoris and vastus lateralis in children and adolescents with cerebral palsy. Dev Med Child Neurol. 2009;51(10):800–806.
14. Kanehisa H, Ikegawa S, Fukunaga T. Comparison of muscle cross-sectional area and strength between untrained women and men. Eur J Appl Physiol Occup Physiol. 1994;68(2):148–154.
15. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1(8476):307–310.
16. Chan YH. Biostatistics 104: correlational analysis. Singapore Med J. 2003;44(12):614–619.
17. Shortland AP, Fry NR, Eve LC, Gough M. Changes to medial gastrocnemius architecture after surgical intervention in spastic diplegia. Dev Med Child Neurol. 2004;46(10):667–673.
18. Shortland AP, Harris CA, Gough M, Robinson RO. Architecture of the medial gastrocnemius in children with spastic diplegia. Dev Med Child Neurol. 2002;44(3):158–163.
19. Mohagheghi AA, Khan T, Meadows TH, Giannikas K, Baltzopoulos V, Maganaris CN. In vivo gastrocnemius muscle fascicle length in children with and without diplegic cerebral palsy. Dev Med Child Neurol. 2008;50(1):44–50.
20. Bland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Med Res. 1999;8(2):135–160.
Keywords:© 2012 Lippincott Williams & Wilkins, Inc.
adolescence; cerebral palsy; child; muscle strength/physiology; quadriceps muscle/ultrasonography; reliability and validity of measures