A stiffknee gait pattern is frequently associated with several impairments including quadriceps spasticity in children diagnosed with cerebral palsy (CP). The relationship of clinical measures of quadriceps spasticity and the stiffknee gait pattern in children diagnosed with CP has not been well established. Therefore, the purpose of this study was to determine the ability of clinical measures of quadriceps spasticity (modified Ashworth scale [MAS], Ely tests, and pendulum test) to categorize a stiffknee gait pattern in children with CP. Children were categorized as having a stiffknee gait pattern based on kinematic and EMG gait data. Results of a logistic regression model revealed that the only significant measure was A1 of the pendulum test. Discriminant analysis functions were used to predict group membership (stiffknee, not stiffknee gait pattern) for each measure. The A1 of the pendulum test demonstrated the highest classification accuracy and the highest sensitivity compared to the other measures. Therefore, a negative pendulum test (indicated by an A1 value of 45 degrees or more) is more useful for ruling out a stiffknee gait pattern compared to the other clinical measures.
The clinical presentation of cerebral palsy (CP) includes a broad spectrum of motor impairments of the neuromusculoskeletal systems such as joint contractures, decreased motor control, and muscle spasticity [
A review of clinical spasticity measures reported that most clinical scales (Modified Ashworth Scale and the Ely test) are subjective and their reliability and validity have not been thoroughly evaluated [
The pendulum test is an objective measure of quadriceps spasticity that has been reported in the literature for over 50 years [
If a patient demonstrates a certain gait pattern, then certain clinical measures are performed to determine the cause of the gait deviation. However, the expected relationships between clinical measures and gait pattern are not always present, causing some to question whether a passive measure of an impairment can relate to the dynamic functional activity of walking [
Procedures were approved by a local institutional review board and informed consent was obtained. Children previously diagnosed with spastic CP (diplegia or hemiplegia) and referred to our facility over a twoyear period for clinical gait analysis were recruited for this prospective observational study (
Characteristics of subjects (
Cerebral palsy 

% 

Diagnosis  
Spastic diplegia  55  81 
Right hemiplegia  5  7 
Left hemiplegia  8  12 
GMFCS level  
I  29  43 
II  24  35 
III  14  21 
IV  1  1 
Assistive device  
None  53  78 
Walker  9  13 
Loft strand crutches  6  9 
Data were collected in the same order for all children by one examiner. Kinematic and EMG data while walking were first collected using surface reflective markers and surface electrodes following the standard gait analysis protocol (Cleveland clinic marker set) [
The surface EMG and reflective markers were then removed and then a standard clinical examination was performed including three clinical tests of quadriceps spasticity. First, the MAS was performed using an ordinal scale (0, 1, 1.5, 2, 3, and 4) [
Next, the subject was placed in prone position. The ElyS test (an assessment of quadriceps spasticity) was recorded as the presence/absence of resistance experienced by the examiner when performing prone knee flexion rapidly [
A subject’s gait pattern was classified as stiffknee if at least 4 out of 6 previously reported criteria of kinematic and electromyographic data were present [
Statistical analyses were conducted using SPSS, V22.0. (Chicago, IL). In order to compare categorical and continuous clinical measures from those with and without stiffknee gait, chisquare and two sample
Thirtyone of the 68 (46%) children were classified positive for a stiffknee gait pattern. Table
Dependent and independent measures (
Measures  Stiffknee 
Notstiffknee 

Criteria of knee gait pattern  
Knee flexion angular velocity at toeoff (degrees/sec), mean (SD)  120 (80) 
263 (95) 
Time to maximum knee flexion in swing (percent of swing phase), mean (SD)  55 (18) 
43 (11) 
Maximum swing phase knee flexion (degrees), mean (SD)  56 (13)  60 (6) 
Total knee motion (degrees), mean (SD)  38 (14) 
50 (13) 
Inappropriate quadriceps EMG activity, 
25 (81)  24 (65) 
Toe drag, 
11 (36)  1 (3) 
Pendulum test  
A1 [max knee angle − start knee angle = amount of knee flexion 
34 (15) 
57 (23) 
Modified Ashworth Score 

0  16 (52) 
32 (87) 
1  10 (32)  5 (13) 
1.5  3 (10)  
2  2 (6)  
ElyS (quadriceps spasticity) 

Yes  12 (39) 
29 (78) 
No  19 (61) 
8 (22) 
ElyF (quadriceps flexibility) 

90 degrees or less  23 (74)  15 (40) 
Greater than 90 degrees  8 (26)  22 (60) 
All dependent and predictor measures were normally distributed (skewness < ±2). None of the 68 subjects demonstrated significant multiple measure outliers. Logistic regression model was significant (chisquare 23 with 4 degrees of freedom
Logistic regression model.
Measures 

SE  Wald  df  Sig. 

95% CI for 


MAS 
0.816  0.624  1.711  1  0.191  2.261  0.666  7.674 
ElyS test 

0.734  0.134  1  0.715  0.765  0.181  3.223 
ElyF test 

0.020  0.011  1  0.915  0.998  0.960  1.037 
A1 (knee flexion during first swing of pendulum test) 

0.022  4.952  1  0.026  0.952  0.911  0.994 
Constant 

1.718  1.422  1  0.233  7.758 
Sensitivity, specificity, and positive and negative likelihood ratios for clinical measures of spasticity (
Spasticity measure  Overall accuracy  Sensitivity 
Specificity 
Positive likelihood ratio 
Negative likelihood ratio 

A1 of pendulum test  77%  0.87 
0.68 
2.69 
0.19 


MAS  69%  0.48 
0.87 
3.59 
0.60 


ElyF  62%  0.52 
0.70 
1.74 
0.69 


ElyS  71%  0.61 
0.78 
2.84 
0.50 
The impetus for this study came from interest in better understanding the relationship between measures of quadriceps spasticity and the stiffknee gait pattern of children diagnosed with CP. Because quadriceps spasticity has been proposed as one cause of the stiffknee gait pattern in children diagnosed with CP the purpose of this study was to assess the ability of measures (MAS, Ely tests, and pendulum test) to correctly categorize a stiffknee gait pattern. Logistic regression analysis revealed that only the A1 of the pendulum test was a significant measure for correctly categorizing the children’s gait pattern as stiffknee or not. For each onedegree increase in the A1 the odds are decreased by 5% that the child will have a stiffknee gait.
The regression analysis identified those clinical measures most related to the target problem (stiffknee gait pattern). From the regression analysis, the greatest amount of variance of the knee gait patterns was explained by the A1 measure of the pendulum test. However, from a clinical perspective, this information is not very applicable. Clinically, it is more important to know the discriminant ability of a test, or how well a test can identify the target problem. Therefore, discriminant analysis was performed to assess the ability of the clinical measure to correctly identify stiffknee gait pattern. The A1 measure demonstrated a higher overall accuracy to correctly classify the original groups as having a stiff or notstiffknee gait pattern 77% compared to the other measures (MAS 69%, ElyF 62%, and ElyS 71%) (Table
However, the overall accuracy of a test does not provide information regarding false positive and the false negative rate of a test [
A limitation of sensitivity and specificity is that one has to know if the target problem is truly present to calculate the sensitivity or specificity of a test. Sensitivity and specificity only indicate the probability of a correct test result (true positive, false positive, false negative, and true negative) [
By multiplying the pretest probability of a target problem by the likelihood ratio gives the posttests odds of the target problem [
Two possible explanations exist for the A1 measure of the pendulum test being the only significant measure in the logistic regression. First, four of the six criteria of a stiffknee gait pattern and the A1 measure of the pendulum test are all measures of knee displacement. Second, the methodological difference between subjective clinical measures (ElyF, ElyS, and MAS) and the objective pendulum test are significant. During the pendulum test, a constant force (gravity) is applied and the knee motions are objectively measured by the motion analysis system. In contrast, the MAS and ElyS test are subjective assessment performed at inconsistent knee flexion velocities. Therefore, the combination of a consistent application of force with a quantitative measurement of knee displacement motion, instead of subjective assessments, resulted in the pendulum test being the only significant measure to identify stiffknee gait patterns.
Previous studies using the pendulum test and EMG data have only presented a visual example of EMG data [
A limitation of this study is that ten of the children had previously undergone spasticity reducing interventions (rhizotomy or baclofen pump). Due to the small sample size no detailed analyses were performed on this subset of children. However, consistent with previous study results [
Because the logistic regression model explained 29–39% of the variance of the gait patterns, there are other factors not accounted for by the model contributing to the stiffknee gait pattern. Previous studies using computer modeling of children diagnosed with CP reported that the knee flexion velocity at toeoff had a larger influence on knee range of motion in swing than hip flexion velocity, knee angle, and hip angle at early swing phase [
Presently the pendulum test requires the use of biomechanical instrumentation techniques (electronic goniometer or threedimensional motion analysis system) to quantitatively measure the knee motion. Therefore, some clinicians may believe that the pendulum test is not a clinically useful tool. We acknowledge that this is a limitation of the pendulum test. However, a video camera or smart phone could record the knee motion during the pendulum test and then it could be played back and paused to estimate the value of A1. Therefore, future studies to assess the accuracy of visual analysis of the pendulum test are needed.
The purpose of this study was to assess the relationship of the different measures of quadriceps spasticity (MAS, Ely, and pendulum test) in the presence of a stiffknee gait pattern, measured using motion analysis laboratory. Logistic regression analysis revealed that only the A1 of the pendulum test was a significant measure for correctly categorizing the children’s gait pattern as stiffknee or not. The A1 measure demonstrated a moderate negative likelihood ratio of 0.19. Hypothetically, if a clinician was 50% confident a child has a stiffknee gait pattern. Multiplying the pretest probability of a target problem by the likelihood ratio gives the posttests odds of the target problem. Therefore, a negative pendulum test (as evident by an A1 greater than or equal to 45 degrees) decreases the pretest probability from 50% to the posttest probability of 10% that the child has a stiffknee gait pattern. By implementing the pendulum test clinicians can objectively and reliably identify if quadriceps spasticity is absent. If spasticity of the quadriceps is not present, as evident by 45 degrees or more of knee flexion during the first swing of the pendulum test, then other body structures or impairments (decrease force generation of the hip flexors and ankle plantar flexors) should be assessed for the cause of the stiffknee gait pattern in a subject diagnosed with CP.
Tim L. Uhl and Sam Augsburger are coauthors.
Funding for this project was provided by Kosair Charities of Louisville, KY. However, the authors have no potential for material gain as a result of this study.
The authors would like to thank the children and their families for their participation in this study. The authors would also like to thank Bobbie Edester for her assistance in collecting the data for this study.