Frameworks for physical therapy clinical decision making include examining a person's current activities and any functional limitations.1 Information gathered through the interview/history and systems review guides the clinician in selection of tests and measures for a more individualized examination and influence the evaluative process. When examining a child, details collected during history taking include such information as sex and age, any known medical diagnoses such as cerebral palsy or special education classification such as autism,2 as well as growth and development attributes such as height, weight, and hand dominance.1 In children, some of these factors may influence the development or execution of motor skills, specifically walking. Research supports the influence of age and height on changes in gait. As a child ages and grows, walking speed increases and cadence decreases.3 – 5 Increases in height combined with sex identification is used to graph decreases in cadence.6 The effect of weight is noted in children identified as obese; when compared with their peers of normal weight, they cover 14% less distance in a 6-minute walk test.7 Differences are established in temporospatial parameters of gait between children with and without developmental disability.8 – 11 Children with disability, such as cerebral palsy,8 autism,9 developmental coordination disorder,10 and Down syndrome,11 show decreases in velocity and increases in cadence compared to peers who are nondisabled.
A systems review is a brief examination that can guide the clinician to rule in or out the need for more detailed measures to assist in clinical diagnosing.12 A systems review is commonly thought to focus on body structures and function to identify impairments such as performing a gross screen of active range of motion. However, a systems review can focus on activity to identify functional limitations. A systems review of activity involves measurement of task-specific performances. One possible focus of a systems review of activity is a check of functional ambulation in a variable environment. For our study, we define functional ambulation as the ability to walk at one's chosen speed and overcome environmental challenges in a timely and safe manner.13
The Standardized Walking Obstacle Course (SWOC)14 is a test that allows interaction of a child with environmental demands and obstructions while examining differences in functional ambulation performances. This test has high interrater reliability for time (intraclass correlation coefficient [ICC] [2,2] = 0.99) and number of steps (ICC [2,2] = 0.94–0.99) when testing children with developmental disabilities aged 6 to 21 years who walk without assistive devices such as canes, crutches, or walkers and those aged 4 to 11 years without disabilities.14 In addition, high interrater reliability for time (ICC [1,1] = 0.77–0.91) and number of steps (ICC [1,1] = 0.87–0.94) was computed for children aged 3 years to 15 years 11 months without disabilities [unpublished data from authors]. The SWOC and Timed Up and Go were moderately correlated for time (r = 0.72–0.90; P < .05) and number of steps (r = 0.63–0.92; P < .05).14 Therefore, the SWOC has the psychometric properties necessary to measure functional ambulation in children.
Clinicians are encouraged to use standardized assessments and a systematic data collection protocol throughout examination to improve clinical decision making.15 – 17 Additional information about child characteristics may be used to determine the extent to which these factors influence task performance. Furthermore, measures of task performance, as part of a systems review, may generate new information related to the need for further testing. Therefore, the primary purpose of this study was to identify characteristics in children that influence performance on the SWOC. The secondary purpose was to determine sensitivity and specificity of the SWOC for use as a clinical screening tool for functional ambulation.
Children with and without disability were recruited from public and private schools, after school programs, private practices, and hospital-based clinics. Recruitment occurred through letters of introduction and phone calls with principals, directors of after-school programs, program administrators, and treating physical therapists. The selection criteria included children between the ages of 3 and 21 years. This age range includes children with developmental disabilities who may remain in school programs until age 21 years. Additional inclusion criteria were the ability to follow 3-step directions in proper order, to walk without any assistive devices and while carrying an object with 2 hands, and to have no known cardiopulmonary comorbidity that would impede completion of SWOC tasks. The parent or referring therapist determined if a child met all inclusion criteria. Each parent or guardian signed an informed consent and each child older than 6 years provided verbal assent before data collection. The Human Subjects Research Review Committee at Daemen College and the Institutional Review Boards at State University of New York at Syracuse and Hampton University approved this research.
The SWOC is a designated walking path 39.5 ft in length, 36 in in width, and includes 30°, 70°, and 90° turns. Obstacles placed at standardized locations along the path are a crutch, brightly colored and shag rugs, and standard-height kitchen garbage can. They are placed to step over, on, or around. Previous publications contain an illustration and more detailed descriptions of this tool.14 , 18
Total time for testing was 15 minutes. Various examiners collected data, including the authors (KK, SH, MRF)—who each have 20 or more years of experience in pediatric physical therapy—and physical therapist students. Student examiners were trained on the standardized test procedures previously found to be reliable14 and participated in practice scoring sessions with all testers for 100% accuracy. Testing occurred in various locations, such as gymnasiums, hallways, and out of doors, at no set time of day (ranging from early morning to late afternoon) and with no control for noise or people present.
Information recorded before SWOC data collection included age, sex, racial/ethnic classification, medical diagnoses and/or special education classification, and hand and leg dominance. This information was obtained from parents/guardians, teachers, therapists, or clinical charts. Children with medical or special education classification were identified as those with disability. On the day of data collection, height and weight were measured by examiners. Height was measured using a tape measure placed on a wall with the child standing with his or her back against wall. A ruler was held against the top of the head to note height in inches. All children wore their own footwear. Weight was obtained using a digital bathroom scale. All children wore their standard school/play attire. Orthotic wear was noted. Body mass index (BMI) was calculated from the height and weight measures. The 10 characteristics noted, measured, or calculated represent general demographics, measures of growth and development, and possible constraints to functional ambulation usually documented as part of a child's history.
Each participant received verbal directions and a demonstration on how to negotiate the obstacle course under the 3 conditions of walk with arms down at sides as in normal walking (walk = W); walk while carrying a tray with a place setting of plastic cup, plate, and utensils to block the view of the feet (walk with tray = WT); or walk while wearing shaded glasses to simulate a dimly lit environment (walk with glasses = WG). Verbal directions included walking at the child's usual speed—that is, walking as at home or as at school. Each child performed practice trials of W and WT before data collection. This was to ensure the understanding of the difference in tasks as suggested by previous pilot testing.14 Pilot testing determined trial 1 and trial 2 on the conditions of W and WT were statistically significantly different (P < .05) without a practice trial. However, trials 1 and 2 were not statistically significantly different (P < .05) for WG; thus, no practice was given for walk with glasses. After the participants understood the directions and were given the opportunity to ask questions, they began under 1 of the 3 randomly ordered conditions, W, WT, or WG.
During SWOC administration and scoring, a trial consisted of standing up, walking the course from the beginning to the end in 1 direction only, and sitting down. A second trial began after the participant rested as long as needed. Walking through the course in the opposite direction was the second trial, after which a new condition of the SWOC began. Each child completed 6 trials (2 per condition). During each trial, 2 examiners were present. One examiner walked behind the child to ensure safety as needed and recorded the time (via a digital stopwatch) and counted the number of steps required to complete the trial. The other examiner walked along side the child just off the path and counted the numbers of stumbles (any loss of balance or body contact with an obstacle along the course) and the number of steps off the path (all or part of the foot touched the floor along side the path) during that trial. Each examiner immediately recorded data collected after each trial, thus providing accountability of measures throughout the sessions.
Descriptive statistics were used to determine the distribution of the data. The independent variables (n = 10) of sex, age, height, weight, BMI, race/ethnicity, hand dominance, leg dominance, disability status, and orthotic wear were not normally distributed in this sample population. One author (EG) without direct involvement with data collection completed data analysis using raw data only. To describe the central tendency and dispersion of this data set medians and semiquartiles were calculated. The semiquartile describes the variability of the distribution. Semiquartile is half the difference between the value at the 75th percentile and the value at the 25th percentile of the data set. The variability of the distribution is best represented by median and semiquartile for skewed distributions. The median is not affected by outliers and semiquartile is not affected by higher values.19 Medians and semiquartiles were computed for age, height, weight, and BMI. Counts and percent distributions were determined for sex, race/ethnicity, hand dominance, leg dominance, disability status, and orthotic wear. Dependent variables (n = 12) were outcome measures (time, number of steps, number of steps off, number of stumbles) by SWOC condition (W, WT, and WG). Medians and semiquartiles were calculated for each measure. Statistically significant relationships between sample characteristics and outcome measures by SWOC conditions were determined using Chi-square analysis. Significant predictors of outcome measures by SWOC conditions were determined using multiple logistic regression analyses. The 10 predictor variables (characteristics of the child) were analyzed simultaneously (block 1) entered together into each model as they are naturally occurring in each individual. We selected model estimation of variance as criteria to define best models.
To check accuracy of the models, we examined sensitivity and specificity using the predictive probability of disability for each measure compared with the observed disability status of our sample. The documented status of the disability was the reference standard. The cutoff value for each measure was its median score. Therefore, a positive result or limitation in ambulation was a value greater than or equal to the median and a negative result or no limitation in ambulation was a value below the median. All data analyses were conducted using SPSS 17.0.
Description of the Sample
Table 1 includes frequency distributions (counts, percentages) of independent variables for the sample. Of the 440 children who participated in this study, the majority were white non-Hispanic (76.6%) male children (50.3%) aged 87 months or less (50.3%), weighing 59 lb or less (50.1%), measuring 49 in or less (50.5%), with a BMI of 17.17 kg/m2 or less (50.8%) who demonstrated right-hand (82.0%) and right-leg dominance (78.5%). Less than 20.0% of children had a documented medical diagnosis or diagnostic classification for special education1; thus, identifying them as children with disability. The children with disability included 8.6% with cerebral palsy level I or II per Gross Motor Function Classification Scale,20 3.0% with Down syndrome, and 7.4% with other conditions. These included children with autism, developmental delay, mental retardation, and learning disability. Less than 5.0% of the children with disability were wearing 1 or 2 lower extremity orthotics at the time of data collection.
Distributions of Outcome Variables by SWOC Conditions
Twelve outcome measures by SWOC conditions are presented in Table 2 for the whole sample. These outcome measures are time, number of steps, number of steps off, and stumbles under W, WT, and WG. Variability was observed under all conditions for time and number of steps. The shortest time and least number of steps were observed during W. No variability was observed in number of steps off and number of stumbles under W and WG conditions. The variables of time, number of steps, stumbles, and steps off are further described via medians and semiquartiles for upper and lower values of each outcome measure by SWOC condition (Table 3).
Summary of Bivariate Analysis
The bivariate analysis examined the association between sample characteristics and outcome measures by SWOC conditions. Chi-square results were computed for all relationships between sample characteristics and outcome measures under W, WT, and WG conditions. Of 120 bivariate relationships, 64.2% (n = 77) were statistically significant at P ≤ .05.
Results for Time
Statistically significant relationships (86.6%; n = 26) were identified for time under all conditions. Chi-square results are reported in Table 4. There were no statistically significant relationships between BMI and time under the W and WG conditions, hand dominance and time under the W condition, and sex and time under the WG condition.
Results for Number of Steps
Statistically significant relationships (90.0%; n = 27) were identified for the number of steps under all conditions. Chi-square results are reported in Table 5. There were no statistically significant relationships between sex and number of steps under the W and WG conditions, and between BMI and the number of steps under the W condition.
Results for Number of Steps Off Path
Statistically significant relationships (50%; n = 15) were identified for the number of steps off path under all conditions. Chi-square results are reported in Table 6. There were statistically significant relationships between the number of steps off path and hand dominance, leg dominance, disability status, and orthotic wear, respectively, under W. There were statistically significant relationships between the number of steps off path and height, weight, age, and disability status under WT. There were statistically significant relationships between the number of steps off path and height, age, race/ethnicity, hand dominance, leg dominance, disability status, and orthotic wear under WG.
Results for Number of Stumbles
Statistically significant relationships (30%; n = 9) were identified for the number of stumbles under all conditions. Chi-square results are reported in Table 7. There were statistically significant relationships between the number of stumbles and disability status and the number of stumbles and orthotic wear under all SWOC conditions. Statistically significant relationships were found between the number of stumbles and hand dominance and the number of stumbles and BMI under the W condition. There was a statistically significant relationship between the number of stumbles and sex under the WT condition.
Summary of Multivariate Analysis
A multivariate analysis examined the simultaneous influence of 10 predictor variables on each SWOC condition. These variables included sex, age, height, weight, BMI, race/ethnicity, hand dominance, leg dominance, disability, and orthotic use. Each variable was significant in at least 1 condition. The analysis of these variables on each outcome measure by SWOC condition generated 24 multiple logistic regression models. Six models were computed per outcome measure, that is, time, number of steps, number of steps off path, and number of stumbles.
Results for Time
Two best models were identified for time under the WT condition. In the logistic regression model for predictors of less than 14.5 seconds under the WT, the significant predictors were age and weight. Children older than 87 months (7 years 3 months) were 5.1 times more likely to finish the SWOC faster than children younger than 87 months. Children weighing more than 59 lb were 2.6 times more likely to finish the SWOC faster than children 59 lb or less (Table 8). In the logistic regression model for predictors of 14.5 seconds or more under the WT condition, the significant predictors were disability and race. Children having a disability were 76.8 times more likely to take more time than children without disability. Children who belonged to a race/ethnic group other than white non-Hispanic were 2.3 times more likely to take more time than children in the white non-Hispanic group (Table 9). Both models roughly explained 59.1% of the variability in time on the SWOC under the condition of WT.
Results for Number of Steps
Two best models were identified for the number of steps under WT. In the logistic regression model for predictors of less than 25 steps under the WT condition, the significant predictors were age and weight. Children older than 87 months (7 years 3 months) were 15.5 times more likely to take fewer steps than children younger than 87 months. Children weighing more than 59 lb were 3.7 times more likely to take fewer steps than children weighing less than 59 lb (Table 10). In the logistic regression model for predictors of 25 steps or more under the WT condition, the only significant predictor was disability. Children having a disability were 69 times more likely to take more steps while carrying a tray (Table 11) than children without disability. Both models roughly explained 64.1% of the variability in number of steps on the SWOC under the condition of WT.
Results for Number of Steps Off the Path
Two best models were identified for the number of steps off the path under the WG condition. In the logistic regression model for predictors of no steps off path under the WG condition, the significant predictors were age and hand dominance. Children older than 87 months were 4.5 times more likely to not step off the path than children 87 months or younger. Children who were right-hand dominant were 2.3 times more likely to not step off the SWOC while wearing glasses (Table 12). In the logistic regression model for predictors of 1 or more steps off the path, the significant predictor was disability. Children having a disability were 30.1 times more likely to step off the path at least 1 time while wearing glasses than were children without disability (Table 13). Both models roughly explained 41.5% of the variability in the number of steps off path on the SWOC, under the WG condition.
Results for Number of Stumbles
Two best models were identified for number of stumbles under W. In the logistic regression model for predictors under the W condition, the significant predictor was weight. Children who weighed more than 59 lb were 5.8 times more likely not to stumble during the SWOC than children weighing 59 lb or less (Table 14). In the logistic regression model for predictors under the W condition, the significant predictors were BMI, disability, and wearing an orthotic. Children with a BMI more than 17.17 kg/m2 were 3.4 times more likely to stumble at least once during the SWOC than children with a BMI 17.17 or below. Children with a disability were 8.6 times more likely to stumble at least once during the SWOC than children without disability. Children wearing an orthotic were 4.2 times more likely to stumble at least once on the SWOC (Table 15) than children not wearing an orthotic. Both models roughly explained 30.1% of the variability in number of stumbles during the SWOC under the W condition.
Sensitivity and Specificity of the SWOC
For the measures of time, under all conditions, sensitivity ranged from 93% to 95% and specificity from 60% to 61%. Accuracy ranged from 66% to 67%. For the measures of number of steps under all conditions, sensitivity ranged from 72% to 81% and specificity from 66% to 72%. Accuracy ranged from 62% to 73%. For the measure of number of stumbles, under all conditions, sensitivity ranged from 24% to 41% and specificity from 94% to 97%. Accuracy ranged from 82% to 84%. For the measure of number of steps off path, the results for the conditions of W and WG were different from the WT condition. Sensitivity was 68% for the W condition and 75% for the WG condition, with specificity 89% and 84%, respectively. For the condition of WT, sensitivity was 83% and specificity was 54%. Accuracy ranged from 60% to 86%. Tables 16, 17, and 18 present sensitivity and specificity (95% confidence intervals) when all measures are combined.
One purpose of this study was to identify characteristics of children that influence performance on the SWOC. We identified 4 statistically significant logistic regression models for time and the number of steps in the WT condition, 2 statistically significant logistic regression models for the number of steps off path in the WG condition, and 2 statistically significant logistic regression models for the number of stumbles in the W condition. Disability was either the sole or strongest predictor in the models that predicted more than the median time, steps, steps off path, and stumbles. Children with disability were slower, took more steps, took more steps off the path, and stumbled more frequently than the median value for each condition. This research supports previous findings that gait parameters are different in children without disability from those with various disabilities such as cerebral palsy,8 autism,9 developmental coordination disorder,10 and Down syndrome.11 In combination with disability, race/ethnicity, or BMI and orthotic use were also significant predictors. All the children wearing an orthotic in this sample had an identified disability. Children with disability who also wear orthotics tend to have problems with stability,8 , 21 thus making them more likely to stumble. In our sample, 53% of the children classified with a disability and nonwhite had BMI scores greater than 30 indicating obesity. Obesity in children is associated with racial/ethnic group identity.22 In addition, there is a relationship between increased BMI and children with disability.23 Finally, increased BMI has been associated with decreased postural stability24 and decreased walking speed.7 Thus, this research supports previous findings.
Age (>87 months) was the strongest predictor in the models that predicted less than the median time, number of steps, and number of steps off path. Weight (>59 lb) was the only predictor of stumbles. As previously discussed, increased age is a predictor of faster time, fewer steps, and more stability in children without disability.25 – 27 Walking patterns should be more stable after the age of 7 years4 , 27 and children should be more adaptable to threats to their balance.22 Increased weight is correlated with increased age in children28 and, as noted earlier, obesity is associated with postural instability. In our sample, 84% of the children had BMI less than 30, which could have made them more stable.
Finally, in relation to stability, age and right-hand dominance were significant factors in 1 condition. This could have been an artifact of the predominance of right-handedness in this sample, but the significance of handedness to functional ambulation needs further examination.
To summarize, the best models for estimation of variance provide clinical information about the SWOC as a screening tool. On the basis of model estimation of variance, we suggest that the WT condition might be the best choice for screening for disability.
The second purpose of this study was to determine sensitivity and specificity of the SWOC as a clinical screening tool. The measures under the W condition were the most accurate (77%) to positively identify the presence of disability on the basis of sensitivity and specificity analyses of the combined SWOC measures. However, the WG condition was 75% accurate and WT was 70%. This suggests that any of the conditions could be used in a systems review of activity. On the basis of the positive likelihood ratio under the W condition and the pretest probability, posttest probability improves by 25%. This means that a positive test result will most likely occur in a child with disability. If the pretest probability is 50% using the positive likelihood ratio under the W condition, the posttest probability increases to approximately 75% with an initial positive result. On the basis of the negative likelihood ratio and the pretest probability, posttest probability decreases by 29%. This means that a negative test result will most likely occur in a child without disability. If pretest probability is 50% using the negative likelihood ratio under the W condition, the posttest probability decreases to 21%. Considering these statistics, the SWOC is better at ruling in a disability with positive results than ruling out with negative results.29
The study included several limitations. One limitation was use of a sample of convenience. However, we had 23% nonwhite children, which exceeds that reported in the 2000 US Census,30 and only a slight minority of female children (49.6%); 16% of the sample had BMI more than 30 and 82% right handed matching current statistics.22 , 31 The number of children with a disability that affected their mobility in our sample was 19%. This exceeds the approximate 1.2% of children with disabilities in mobility between the ages of 5 and 17 years in the US population.32 Although we did not randomly select our sample, we did use a random order of testing.
Our study results support use of the SWOC as a screening tool for children with and without disability. Knowing the strongest predictors of performance under multiple SWOC conditions can assist clinical decision making about the need for further testing beyond an initial systems review of functional ambulation.
1. Drnach M. The Clinical Practice of Pediatric Physical Therapy from the NICU to Independent Living. Philadelphia, PA: Lippincott Williams & Wilkins; 2008:190–250.
2. Effgen SK. Meeting the Physical Therapy Needs of Children. Philadelphia, PA: F. A. Davis; 2005:379.
3. Hausdorff JM, Zemany L, Peng C-K, Goldberger AL. Maturation of gait dynamics stride-to-stride variability and its temporal organization in children. J Appl Physiol. 1999;86:1040–1047.
4. Sutherland D. The development of mature gait. Gait Posture. 1997;6:163–170.
5. Beck RJ, Anriacchi TO, Kuo KN, Fermier RW, Galante JO. Changes in the gait patterns of growing children. J Bone Joint Surg Am. 1981;63:1452–1457.
6. Todd FN, Lamoreux LW, Skinner SR, et al. Variations in the gait of normal children: a graph applicable to the documentation of abnormalities. J Bone Joint Surg Am. 1989;71:196–204.
7. Morinder G, Mattsson E, Sollander C, Marcus C, Larsson UE. Six-minute walk test in obese children and adolescents: reproducibility and validity. Physiother Res Int. 2009;14:91–104.
8. Prosser LA, Lauer RT, VanSant AF, Barbe MF, Lee SCK. Variability and symmetry of gait in early walkers with and without bilateral cerebral palsy. Gait Posture. 2010;31:522–526.
9. Rinehart NJ, Tonge BJ, Iansek R, et al. Gait function in newly diagnosed children with autism: cerebellar and basal ganglia related motor disorder. Dev Med Child
10. Woodruff SJ, Bothwell-Myers C, Tingley M, Albert WJ. Gait pattern classification of children with developmental coordination disorder. Adapt Phys Act Q. 2002;19:378–391.
11. Buzzi UH, Ulrich BD. Dynamic stability of gait cycles as a function of speed and system constraints. Motor Control. 2004;8:241–254.
12. Guide to Physical Therapist Practice. Second Edition. American Physical Therapy Association. Phys Ther. 2001:81:9–746.
13. Graham HK, Harvey A, Rodda J, Nattrass GR, Pirpiris M. The functional mobility scale. J Pediatr Orthop. 2004;24:514–520.
14. Held SL, Kott KM, Young B. Standardized Walking Obstacle Course (SWOC): reliability and validity of a functional measurement tool in children who are developing typically and atypically. Pediatr Phys Ther. 2006;18:23–30.
15. Schenkman M, Deutsch JE, Gill-Body KM. An integrated framework for decision making in neurologic physical therapist practice. Phys Ther. 2006;86:1681–1702.
16. O'Neil ME, Fragala-Pinkham MA, Westcott SL, et al. Physical therapy clinical management recommendation for children with cerebral palsy—spastic diplegia: achieving functional mobility outcomes. Pediatr Phys Ther. 2006;18:49–72.
17. Palisano R. A collaborative model of service delivery for children with movement disorders: a framework for evidence-based decision making. Phys Ther. 2006;86:1295–1305.
18. Kott KM, Held SH. Effects of orthoses on upright functional skills of children and adolescents with cerebral palsy. Pediatr Phys Ther. 2002;14:199–207.
21. Burtner PA, Woollacott MH, Craft GL, Roncesvalles MN. The capacity to adapt to changing balance threats: a comparison of children with cerebral palsy and typically developing children. Develop Neurorehab. 2007;10:249–260.
23. Chen AY, Kim SE, Houtrow AJ, Newacheck PW. Prevalence of obsesity among children with chronic conditions. Obesity. 2009;18:210–213.
24. Hue O, Simoneau M, Marcotte J, Berrigan F, Dore J, Marceau P. Body weight
is a strong predictor of postural stability. Gait Posture. 2007;26:32–38.
25. Beck RJ, Andriacchi TP, Kuo KN, et al. Changes in the gait patterns of growing children. J Bone Joint Surg Am. 1981;63:1452–1457.
26. Norlin R, Odenrick P, Sandlund B. Development of gait in the normal child
. J Ped Orthop. 1981;1:261–266.
27. Sutherland DH, Olshen R, Cooper L, Woo SL. The development of mature gait. J Bone Joint Surg Am. 1980;62:336–353.
28. Tanner JM, Whitehouse RH, Takaish M. Standards from birth to maturity for height, weight, height velocity, and weight velocity: British children, 1965. I. Arch Dis Child
29. Portney LG, Watkins MP. Foundations of Clinical Research Application to Practice. 3rd ed. Upper Saddle, NJ: Pearson; 2009:620–630.
Keywords:© 2011 Lippincott Williams & Wilkins, Inc.
activities of daily living; adolescent; age factors; body weight; child; preschool; developmental disabilities/physiopathology; disability evaluation; motor skills/physiology; physical therapy (specialty)/instrumentation; physical therapy (specialty)/methods; postural balance/physiology; sensitivity; specificity; predictive validity; predictive value of tests; walking/physiology