Children with cerebral palsy (CP) represent one of the largest populations treated by physical therapists in the United States. 1 These children commonly demonstrate an impairment in postural control. 2 Purposeful interaction with the surrounding environment is improved through the control of posture, which is hypothesized to involve many systems including, at a minimum, the musculoskeletal, neuromuscular, and sensory systems. 3 Evaluation of these systems is important for the development of appropriate interventions. Additionally, reliable and valid measures of limitations in functional activities that reflect aspects of underlying postural control are required to document functional outcomes of intervention. 4
The Timed Up and Down Stairs (TUDS) was developed as a functional mobility outcome measure that would potentially reflect improvements in the musculoskeletal and neuromuscular systems that contribute to the control of posture. The TUDS involves the subject ascending one flight of stairs, turning around, and descending to the starting point. One can hypothesize that performance on the TUDS requires a certain amount of strength of the lower extremities and trunk, range of motion (ROM) in the lower extremities, coordination for fast reciprocal movements, and anticipatory and reactive postural control. The relationship of these types of components to functional mobility has been supported by a recent study in elderly men. 5 Competence on the Physical Performance Test (including seven activities of daily living requiring various levels of mobility from being able to eat, to picking up a penny, to turning 360 degrees and walking quickly) was found to be related to component or impairment measures, including gait speed, stride length, fall risk (Modified Gait Abnormality Rating Scale), muscle force production (grip strength), flexibility (active ankle ROM), and fitness (Modified Sitting Step Test). A linear regression revealed a strong relationship between the performance test with gait speed and muscle force production (accounting for 58% and 68% of the variability, respectively). 5 Many of the components of the functional activities studied are hypothesized to be similar to those required for the TUDS (gait speed, strength, active ROM, turning around).
Additionally, Lepage et al 6 have provided support for the relationship between functional abilities and a stair-walking task in children. They demonstrated that a timed up and down stairs task similar to the TUDS was associated with the ability to perform various life habits related to school and social situations as measured by the Life Habits Assessment (version 1.0). 6 The Life Habits Assessment tool is based on the World Health Organization’s model of disablement. Life Habits are defined as “those habits that ensure the survival and development of a person in society throughout his or her life.” 6 Lepage et al 6 found that there was an association between disruptions in life habits and types of locomotor measures. Among the locomotor measures used, the timed up and down stairs task accounted for the largest amount of variance in the disruption of four life habits categories: mobility, community, recreation, and residence for 66%, 37%, 33%, and 22%, respectively. 6 Thus, a TUDS type of measure is associated with the ability to be active in the community and home and during recreational activities. It was hypothesized that the TUDS-type task “demands more balance, coordination, strength, and muscle control than walking demands.” 6 Therefore, the TUDS appears to be an important measure in the evaluation of functional mobility and balance.
Although there is supporting evidence for the validity of the TUDS, the reliability and validity of the TUDS need to be formally evaluated for the test to be used with children. 4 Therefore, the purposes of this study are threefold. The first purpose was to determine the intrarater, interrater, and test–retest reliability in a group of children with CP and children with typical development (TD). The second purpose was to determine the concurrent validity of the TUDS with other accepted measures of functional mobility and balance. Common tests of functional mobility and balance include the Timed Up and Go (TUG), 7,8 the Functional Reach Test (FRT), 9,10 and the Timed One Legged Stance (TOLS). 11 These measures have been demonstrated to assist in the identification of disorders of postural control and resultant limitation in functional abilities. 7–12 The third purpose was to determine whether differences in functional mobility and balance are reflected in the TUDS score (construct validity) by examining relationships between 1) children with CP and children with TD, 2) different levels of functional mobility and balance in children without and with CP as classified using the Gross Motor Function Classification Scale (GMFCS), 13 and 3) different ages.
As part of a larger study on the neuromuscular control of a standing reaching task, a sample of convenience of 48 subjects was recruited from the Delaware Valley area of Pennsylvania, New Jersey, and Delaware. Subjects were recruited via flyers posted in local hospitals, doctors’ offices, and schools. In addition, the charts of two physicians were reviewed for potential candidates. All children were initially contacted by telephone to determine whether they were interested in participating in the study and to see whether they met the inclusion criteria. Inclusion criteria for participation included ages eight to 14 years old, no orthopedic surgeries within the past six months, no history of any genetic or neurological disorder besides CP, and rated as level I or II on the GMFCS (Table 1). 13 This age and functional ability range was chosen due to requirements for the larger study. One of the subjects was unable to complete the testing procedure. Of the remaining 47 subjects, 20 were children with CP (nine female) and 27 were children with TD (14 female). The ethnicity of the subjects was 11 Asian, two Hispanic, nine African American, and 25 white. The age, weight, and height, as displayed in Figure 1, were not substantially different between the children with CP and the children with TD (p = 0.69, 0.63, 0.38, respectively). Other characteristics of the children are listed in Table 2.
This study was approved by MCP-Hahnemann University’s and St. Christopher’s Hospital for Children in Philadelphia’s Internal Review Boards for use of human subjects. After the subjects and parents/guardians agreed to participate and signed the assent and consent forms, preparations for data collection and the testing procedures were completed.
Gross Motor Function Classification Scale.
The GMFCS was used to classify the children with CP into groups based on functional ability. This scale has been shown to have excellent interrater reliability (κ = 0.75) for children between two and 12 years of age. 13 Level I is the highest level of functional abilities and level V the lowest (Table 1). Due to demands of the larger study, recruitment solicited only children in levels I and II. However, there were three subjects who were classified at the time of data collection as level III who were able to complete the testing.
Timed Up and Down Stairs.
For the TUDS, the subject was asked to stand 30 cm from the bottom of a 14-step flight of stairs (19.5-cm step height). Subjects were instructed to “Quickly, but safely go up the stairs, turn around on the top step (landing) and come all the way down until both feet land on the bottom step (landing).” The subjects were allowed to choose any method of traversing the stairs. This included using a step-to or foot over foot pattern, running up the stairs, skipping steps, or any other variation. However, all subjects faced in the direction of the movement (faced up and down stairs, not to the side) as they traversed the steps. There were two handrails available on the stairs, but due to the width of the stairwell, only one handrail could be used at a time. The testing was done wearing shoes but not orthoses. The subjects were given the cues “ready” and “go.” The TUDS score was the time in seconds from the “go” cue until the second foot returned to the bottom landing. Shorter times indicated better functional ability.
Timed Up and Go.
Previous investigations into the TUG with adults and children have demonstrated good reliability and moderate to good levels of association with other clinical measures. 7,12,14 In adults, the TUG has been shown to be correlated to gait speed, postural sway, and the Berg Balance Scale. 7 In children, the TUG has been shown to have excellent interrater reliability in children with disabilities [ICC(3,1) = 0.99] 14 and is related to the index of sway on the Pediatric Clinical Test of Sensory Interaction in Balance 14 and the mobility functional skills and care-giver assistance sections of the Pediatric Evaluation of Disability Inventory (Pearson r = −0.64 and 0.69, respectively). 12 The TUG score also reflects changes in functional abilities as children age. 15 Thus, the TUG is thought to measure components related to gait speed, postural sway, functional mobility, and balance. For the TUG measurement, the child sat in an adjustable-height chair. The height of the seat was adjusted so that the subject’s knees and hips were flexed to 90 degrees when sitting with feet resting on the floor. The testing was done barefoot. The child was given a cue “ready, go.” On the “go” cue, the subject stood up, walked three meters, turned around, walked back, and sat down. The time in seconds was recorded from the “go” cue to when the child sat down in the chair. If the subject ran during the trial, that trial was repeated. Shorter times indicated better functional ability.
Functional Reach Test.
In children with disabilities 16 and without disabilities, 17 the FRT has been found to demonstrate good to excellent reliability. The FRT has been shown to be associated with ground reaction forces during a functional reach task in children. 18 Lowes et al 12 also demonstrated a correlation between the FRT and the mobility functional skills section of the Pediatric Evaluation of Disabilities Index (Pearson r = 0.63) in children with and without CP. The FRT score has been shown to reflect changes in functional mobility and balance as children age. 15 Therefore, it appears that the FRT is related to functional mobility and balance 12,15 and is an indirect measure of strength. 18 For the FRT, the subject stood with the left arm next to a ruled grid (marked in centimeters) located at shoulder height. A 20-cm wide grid was used instead of the yardstick used by Duncan et al 9 because many of the children had difficulty maintaining their arm level in line with the yardstick. The use of the custom-made grid improved the investigators’ ability to obtain accurate FRT measurements. The FRT measurement was initiated by having the subjects elevate their arm to horizontal and make a fist. The subjects were instructed to “reach as far as possible without touching the wall or taking a step.” The difference between the starting position and the final position of the third metacarpal was measured from the grid. If the subjects took a step or touched the wall, that trial was repeated. Raising up on the toes was allowed as long as the child did not lose his or her balance and did not take a step. However, no child raised his or her heels more than approximately one centimeter off the floor. Testing of the FRT was done in bare feet without shoes or orthotics. The FRT was measured in centimeters, with longer distances indicating better functional abilities.
Timed One Legged Stance
One-legged stance time is used in some developmental assessment tools. 19–21 In addition, there is some evidence that reduced one-legged stance time is associated with limitations in the stability of the ankle and a resultant limitation in postural control in children. 11 Thus, TOLS appears to be an important measure of functional mobility and balance and is probably related to the control of the center of gravity within the base of support. The TOLS has been shown to have excellent interrater and test–retest reliability. 22
For the TOLS, the subject was barefoot and stood with hands on hips 61 cm from a visual target (smiling face) located at eye level on a wall. When the subject was ready, he or she would stand on the foot of choice. The knee of the other leg was flexed to 90 degrees. During the test, the subject would continue to focus on the visual target. If the subject’s nonstance foot started to drop causing less than 80 degrees of knee flexion, a verbal cue to lift his or her foot was provided. The knees were allowed to touch during the testing, but the nonstance foot was not allowed to rest against the stance leg. The timing was started when the nonstance foot lifted off the floor, and the timing was stopped if the subject lifted his or her hands off his or her hips, looked down at his or her feet, or touched the floor with the opposite foot. This method of timing of the TOLS has been shown to be reliable (r s = 0.91 to 0.99). 22 The TOLS was timed for a maximum of 45 seconds. It was hypothesized based on previous testing 22 that any time greater than 45 seconds would not discriminate between subjects’ functional abilities. The TOLS was measured in seconds with longer times reflecting better balance ability.
The clinical measures of functional mobility and balance were measured in a random order determined through use of a random numbers table. Two pediatric physical therapists (C.A.Z., V.G.M.) did all the testing. They had 16 and eight years of experience in physical therapy and 10 and eight years in pediatrics, respectively. These investigators had also been trained in the use of the tests. The principal investigator (C.A.Z.) had two years of training with the tests used and had evaluated more than 50 children with the tests. The other investigator (V.G.M.) was trained in the measurements used for this study prior to data collection. Raters also determined the GMFCS level and recorded the child’s age. For all the measures of functional mobility and balance, the better of two trials was used in the analyses. Rest breaks were given to the children as needed during the testing.
Reliability of all measures was examined by using subgroups of the subjects who participated in the study. Twenty-five of these children were filmed performing all the tests of functional mobility and balance. These 25 children were scored in real time and then rescored later from the video by the same rater (C.A.Z.) to determine intrarater reliability. Nine of the children were scored in real time simultaneously by the two investigators to determine interrater reliability. The two investigators stood where they could see the child perform the test but could not see the other investigator’s recordings. Test–retest reliability was examined by having 24 subjects return for a second testing by the principal investigator two hours later, after completing the other portion of the study. During the interrater and test–retest reliability measures, the primary investigator did not have the original records available and the time between the testing was long enough that the investigator was not able to remember the initial scores. The interrater reliability of the GMFCS was determined on a subgroup of nine subjects. Each investigator separately rated the child after he or she had been at the study site for approximately 20 minutes.
The interrater, intrarater, and test–retest reliabilities of the TUDS and other tests were analyzed using intraclass correlation coefficients, ICC(2,1). The percentage of agreement was used to determine the interrater reliability of the GMFCS scores. The concurrent validity of the TUDS was evaluated using Spearman correlations between the TUDS and the TUG, FRT, and TOLS for the children with TD and CP separately. The Spearman correlation analysis was used due to skewed data. To determine whether the TUDS measured differences in functional mobility and balance across age (construct validity), Spearman correlations between age and TUDS scores were performed separately for the TD and CP subject groups. Also, Kruskal-Wallis analysis of variance (K-W ANOVA), followed by pair-wise testing using the Mann-Whitney U test, was run on the TD and CP group data separately, between the three age groups (eight to 10, 11 to 12, 13 to 14 years old). Last, to establish construct validity across all the children based on functional mobility and balance, a K-W ANOVA with Mann-Whitney U tests were used again. For this analysis, the children were divided into three groups, children with TD, children with CP rated level I on the GMFCS, and children with CP rated at levels II and III on the GMFCS. In both instances, nonparametric tests were used due to unequal sample sizes and lack of homogeneity of variance across the age and functional groups. For all analyses, an α level of 0.10 was used. This was chosen due to the preliminary nature of this study, the relatively small sample size, and the minimal risk of harm from a type I error. To guard against inflation of type I error, a Bonferroni correction was used for each subject group when multiple comparisons or correlations were completed. This resulted in an α level of 0.02. All data analyses were done using SPSS version 10 statistical software (SPSS Inc., Chicago, IL).
The mean and 95% confidence interval for the TUDS for the three functional groups divided by GMFCS are presented in Figure 2. The mean and standard error of the mean are presented for all the tests of functional mobility and balance in Table 3. Table 4 includes descriptive data on the TUDS across three age levels: eight to 10, 11 to 12, and 13 to 14 years.
Intrarater reliability and interrater reliability were excellent. For all tests, except FRT, the ICC(2,1) equaled 0.99. For the FRT, the intrarater reliability and interrater reliability were ICC(2,1) = 0.97 and 0.98, respectively. The test–retest reliability of the TUDS was also excellent [ICC(2,1) = 0.94]. The percentage of agreement for inter-rater reliability of the GMFCS was 100%.
Correlations between all the tests of functional mobility and balance are presented in Table 5. For all subjects, the relationship between the TUDS and the other tests of functional mobility and balance (TUG, FRT, and TOLS) demonstrated moderate to good relationships 23 (r s = 0.78, −0.57, −0.77; p < 0.001, respectively). In the TD group, none of the relationships were significant. However, the TUDS with TUG and FRT did approach significance (r s = 0.33, p = 0.046 and r s = −0.32, p = 0.053, respectively). The relationships within the CP group were stronger. The correlation between the TUDS and the TUG was moderate (r s = 0.68, p = 0.001). The correlation between the TUDS and the TOLS approached significance (r s = −0.41, p = 0.038).
The correlation between the TUDS and age was moderate and significant for both the TD and CP groups (r s = −0.61, and −0.41, p < 0.001 and 0.018, respectively). The relationship was not as strong in the CP group; however, the relationship with the higher functioning children with CP (GFMCS level I) was stronger (r s = −0.75, p = 0.010). The K-W ANOVA results showed a statistically significant difference between age groups for the TD group but not for the CP group [χ2 (2) = 9.80, p = 0.007 and χ2 (2) = 3.15, p = 0.208, respectively]. Pair-wise analyses of the TD group with the Mann-Whitney U test demonstrated a significant difference between the eight- to 10- and the 11- to 12-year-old groups (p = 0.020), the eight- to 10- and the 13- to 14-year-old group (p = 0.001) but not between the 11- to 12- and 13- to 14-year-old groups (p = 0.423). A K-W ANOVA demonstrated differences between the TUDS scores and three groups of children using the TD group and two CP groups based on GMFCS levels [χ2 (2) = 28.5, p < 0.001]. Based on Mann-Whitney pair-wise testing, there were statistically significant differences found between all groups (p < 0.02).
The study data demonstrate that excellent intrarater, interrater, and test–retest reliability can be achieved for the TUDS, comparing one experienced physical therapy rater to a physical therapy rater who was trained briefly in the test procedures. The findings of the current study also are in agreement with those of previous studies on the intrarater, interrater, and test–retest reliability of the TUG, 8 TOLS, 22 and FRT. 17 Given these results, it is appropriate to proceed with further evaluation of the validity of the TUDS.
The TUDS scores for the TD group averaged to 0.58 sec/step for ascending/descending. This average is almost half of the 1.11 sec/step for children with CP, GMFCS level I and one third of the 1.75 sec/step for children with CP, GMFCS level II/III. It would appear that use of this measure could be an easy method of monitoring change across time or with therapy. However, the responsiveness of the TUDS is yet to be determined.
The TUG scores in the TD group are similar to those found in children from Pakistan at similar ages 15 and also are similar to children as young as six years (mean TUG times, 5.45). 14 Children with CP at GMFCS level I scored an average of one second higher and children with CP at GMFCS level II scored an average of three seconds higher than the children with TD. This small spread in the scores perhaps reflects the easier nature of the TUG task and suggests that use of the TUG to document differences between children or across time may be better for children with more severe mobility problems.
The TD group FRT scores in this study were greater (approximately 3 cm in the eight- to 10-year-old group and approximately 6 to 7 cm in the 11- to 12- and 13- to 14-year-old groups) than previously recorded for children with TD, but many of the children with CP demonstrated values that placed them in a “delayed reaching skills” category as designated by Donahoe et al. 17 Children with CP and GMFCS level I reached an average of six centimeters less than the TD group, and those children in the GMFCS levels II/III reached 12 cm less than the TD group. This spread of the data across the functional groups suggests that the FRT also may be a useful test for monitoring change across time or with intervention.
The TOLS scores demonstrated that most children in the TD group could stand the maximum of 45 seconds. Times in the GMFCS level I group dropped dramatically to an average of 24.9 seconds less than the TD group. There was an additional drop in the times for the GMFCS level II/III group. This group decreased an average of 37.9 seconds compared with the TD group. In contrast to the minimal spread of the TUG scores, the TOLS test may be too difficult for many of the children with CP. It may be useful in the monitoring of children with CP who are functioning at a very high level.
The initial evaluation of the concurrent validity comparing TUDS scores to the TUG, FRT, and TOLS yielded mixed results. When all subjects were analyzed together, all the tests were moderately and significantly related, suggesting that static control of balance (TOLS), anticipatory control of balance (FRT) and strength, and balance for dynamic movement from sit-to-stand, walking, and turning on a level surface (TUG) were all related components of the TUDS task. As previously stated, the TUG has been shown to be related to measures of functional mobility and balance along with postural sway measures. These components likely include sensory processing to control of the center of gravity as indicated by the TUG’s relationship to postural sway and scores on the Pediatric Clinical Test of Sensory Interaction of Balance. 14 The 61% variation in the TUDS scores accounted for by the TUG scores supports that the TUDS likely measures the constructs of postural sway control. The control of the center of gravity within the base of support is also supported by the higher relationship between the TUDS and the TOLS (59% of the variance).
When the TD and CP subject groups were analyzed separately, however, none of the clinical tests of functional mobility and balance were significantly related to the TUDS in the TD group. The TUDS was moderately correlated with the TUG in the CP group. The difference in the correlations between all subjects and when separated into the TD and CP group separately appears to be due, in part, to a more normal distribution in these data across the test range when all subjects were combined. This resulted mathematically in higher correlations. The children with TD tended to cluster at the more functional end of the plots (lower ends for TUG and higher ends for FRT and TOLS). This was particularly true for the TOLS, where almost all the children with TD topped out at the 45-second maximum. For the children with CP, the data were spread out more but did tend to group at the less functional ends of the plots.
As Lepage et al 24 determined, differences in a timed stair task better represented different abilities to perform various types of activities of daily living than did gait speed on level surfaces. Thus, it was expected that the TUDS might represent functional mobility and balance better than the TUG. Even though these two tests show a moderate to high relationship in the subjects used in this study, the stair-climbing component may better measure differences in strength, balance, and neuromuscular coordination than does the TUG in this type of child. This is demonstrated to some extent in the larger difference in scores across the functional mobility groups noted above for the TUDS compared with the TUG. The variance in the TUDS accounted for by the TUG was also only 11% and 46% for the children with TD and CP, respectively, demonstrating some similarities but also differences between the two measures.
The FRT showed only a moderate relationship with the TUDS in all subjects combined, which may reflect a component for better anticipatory postural control to negotiate stairs more quickly. There may well be a cutoff point in functional reach distance representing adequate anticipatory postural control for fast stair climbing and descending. However, the lower correlation when the CP and TD groups were analyzed separately is curious and suggests that the two tests are measuring different aspects of more static and dynamic balance.
The TOLS correlated highly with the TUDS when the whole group was analyzed but showed low and nonsignificant relationships when the TD and CP groups were separately analyzed. The data for the TOLS as noted above by functional groups were variable. The TUDS requires fast dynamic movement, which does not require long bouts of one foot standing balance; rather, the person must move quickly through the one foot balance phase for fast stair climbing and descending. As with the FRT, there may be a cutoff point in one foot balance ability that is necessary for fast stair-climbing negotiation, which was present in the TD group and variable in the CP group.
Functional mobility and balance are known to change as children age. 25 In children with TD, changes in age were related to changes in the TUDS (37% of variance), and there was a statistically significant difference between the youngest age group and the two older age groups. Age accounted for a smaller portion of the variance in the TUDS scores in all the children with CP (22% of variance). However, if just the age and TUDS scores for the higher functioning children with CP (GMFCS level I) were examined, a greater portion of the variance was accounted for by the child’s age (56% of variance). This high relationship was not true for the children with CP in the lowest functional mobility group, and comparison between age groups in the children with CP also did not show any statistically significant differences. However, with a larger sample, a significant difference between the 8- to 10- and the 13- to 14-year-old children may have been revealed, as the median scores were 7.7 seconds different between these groups. Further testing with larger groups would be necessary to confirm this.
Considering these data, the correlations within each separate group (CP and TD) may have been higher if the age range of subjects was increased. Palisano et al 24 have developed some preliminary growth curves in children with CP across different levels of functional abilities. These growth curves were linear initially at the younger ages and then became curvilinear at the older ages when development plateaus. Typical development also follows a similar pattern. 24 This type of developmental pattern may also be present in the relationship between TUDS scores and age. Nonlinear approaches to correlations may need to be considered as the age of the subjects increases.
Age is only one of many factors related to changes in functional mobility and balance. This is supported by the relatively small amount of variance in the TUDS score accounted for by age, especially in the subjects with CP. Children with CP have other impairments affecting the development of movement related to their original central nervous system insult and subsequent changes in the musculoskeletal system due to altered movement patterns. It was our hypothesis that the TUDS would reflect changes in children’s functional mobility and balance and when functional abilities were measured more directly using the GM-FCS to assign subjects to groups. Our analysis did reveal differences in the TUDS scores across all three functional groups of children.
The Guide to Physical Therapist Practice 4 emphasizes the need to evaluate the reliability and validity of tests and measurements and weigh the time to administer, cost of administering, and subject’s tolerance of a test or measure when determining the suitability of a measure for a particular subject population. The TUDS has been demonstrated to be an quick, low-cost, reliable test for children with and without CP aged eight to 14 years. The TUDS also appears to have preliminary concurrent validity and construct validity. Concurrent validity was highest with the TUG test; however, the TUDS shows a greater range of scores across functional mobility levels and therefore may be a better measure of change across time or with intervention. The TUDS appears to account for variation in developmental and functional abilities as tested by differences across age and functional mobility categories based on GMFCS ratings. We suggest that the TUDS is a simple measure of functional mobility that can be easily done in a variety of settings and should be considered for testing and potentially documenting improvement of children with suspected limitations in functional mobility and balance. However, there is a need for further study with subjects who have a greater range in age and functional mobility and balance. Responsivity of the measure should also be evaluated to determine the usefulness of the test to measure clinically meaningful change across time or with intervention.
We gratefully acknowledge the generous donations of prizes for the children provided by Young’s Medical Equipment, Landsdown, PA; Aetna/US HealthCare, Blue Bell, PA; Theradyne, Jordon, MN; and Sundance Rehabilitation Corporation, Atlanta, GA. We also are grateful to Dr Peter Pizzutillo of St. Christopher’s Hospital of Philadelphia for his assistance in subject recruitment.