Investigation of the timed ‘Up & Go’ test in children

(1)

timed ‘Up & Go’ test in children

Elizabeth N WilliamsDip Physio, School of Physiotherapy, University of Melbourne, Parkville, Victoria;

Sara G CarrollMSc BAppSc(Physio), School of Physiotherapy, Curtin University of Technology, Perth, Western Australia;

Dinah S ReddihoughMD BSc FRACP FAFRM, Department of Child Development and Rehabilitation, Royal Children’s Hospital and Murdoch Childrens Research Institute;

Bev A PhillipsPhD PGDipHealthSci DipPhysio, School of Physiotherapy, University of Melbourne;

Mary P Galea*PhD BAppSc(Physio) BA, School of Physiotherapy, University of Melbourne, Austin Health, Parkville, Victoria, Australia.

*Correspondence to last author atSchool of Physiotherapy, University of Melbourne, Parkville, Victoria 3010, Australia.

E-mail: [email protected]

The timed ‘Up & Go’ test (TUG) is a test of basic or functional mobility in adults which has rarely been used in children.

Functional mobility was defined for this study as an individual’s ability to manoeuvre his or her body capably and independently to accomplish everyday tasks. Reliability and validity of TUG scores were examined in 176 children without physical disabilities (94 males, 82 females; mean age 5y 9mo [SD 1y 8mo]; range 3 to 9y) and in 41 young people with physical disabilities due to cerebral palsy or spina bifida (20 males, 21 females; mean age 8y 11mo [SD 4y 3mo], range 3 to 19y). Mean TUG score for children without physical disability was 5.9s (SD 1.3). Reliability of the TUG test was high, with intraclass correlation coefficients (ICC) of 0.89 within session, and 0.83 for test–retest reliability. Mean score of the group aged 3 to 5 years was significantly higher (6.7s SD 1.2) than that of the older group (5.1s, SD 0.8; p=0.001). Scores in the younger group reduced significantly over a 5-month follow-up period (p=0.001), indicating that the TUG was responsive to change.

Within-session reliability of the TUG in young people with disabilities was very high (ICC=0.99). There were significant differences in TUG scores between children classified at levels I, II, and III of the Gross Motor Function Classification System (p=0.001). TUG scores showed a moderate negative correlation with scores on the Standing and Walking dimensions of the Gross Motor Function Measure (n=22, rho=–0.52, p=0.012).

There was no significant difference in TUG scores between typically developing male and female children. The TUG can be used reliably in children as young as 3 years using the protocol described in this paper. It is a meaningful, quick, and practical objective measure of functional mobility. With further investigation, the TUG is potentially useful as a screening test, an outcome measure in intervention studies for young people with disabilities, a measure of disability, and as a measure of change in functional mobility over time.

in clinical practice as an outcome measure to assess basic or functional ambulatory mobility, or dynamic balance in adults (Shumway-Cook et al. 2000, Creel et al. 2001, Gray et al.

2001, Morris et al. 2001, Steffen et al. 2002, Andersson et al.

2003). However, its reliability and validity have not been investigated comprehensively in children.

In an ambulatory person, basic or functional mobility can be defined as ‘the ability to manoeuvre his or her body capably and independently in order to accomplish everyday tasks’

(adapted from Creel et al. 2001, p 789; and Haley 1991, p 282).

In a child without physical disabilities, motor skills develop rapidly from birth, with independent walking achieved at around 12 months of age. Mature gait patterns become estab- lished somewhere between 3 years and 3¹⁄²to 4 years of age (Sutherland et al. 1980, 1988; Sutherland 1997). Children with physical disabilities who achieve walking do so at variable ages depending on their diagnosis and level of impairment.

Published reports of the use of the TUG in children are few.

The TUG has been used as a measure of balance in Pakistani children aged 5 to 13 years (Habib and Westcott 1998, Habib et al. 1999). Other studies conducted in Australian children (Williams and Carroll 1999, Williams et al. 2001), Indonesian children (Takarini et al. 2003), and Filipino children (M Ruiz, personal communication 2003) have been published only as abstracts. Although little information is available about the use of the test in young people with disabilities, TUG has been used as an outcome measure in adults (Andersson et al. 2003) and children with cerebral palsy (CP; Lowes and Westcott 1995).

The standard TUG test for frail elderly people requires them to to rise from a seat with arms, stand momentarily, walk 3m, turn, return to the same seat, and sit down (Podsiadlo and Richardson 1991). The score is the time in seconds from the instruction ‘go’ to the person regaining their seat. The TUG has been shown to have high sensitivity and specificity as a test for older adults at risk of falls (Shumway-Cook et al. 2000).

Test–retest reliability and interrater reliability have been shown to be good in frail elderly persons (Podsiadlo and Richardson 1991) and those with Parkinson’s disease (Morris et al. 2001).

We sought to establish whether the TUG was a suitable test for children without disabilities (study 1), and for young people with physical disability due to CP or spina bifida (study 2).

The aim in study 1 was to determine whether children could reliably perform the test, and whether the TUG was responsive to change over time. In study 2 the aim was to explore within- session reliability and examine validity in relation to the Gross Motor Function Classification System (GMFCS; Palisano et al.

1997) and the Gross Motor Function Measure (GMFM; Russell et al. 1989), considered a ‘criterion standard’ outcome measure of function for children with CP.

Method

MODIFICATIONS TO STANDARD TUG TEST

Pilot studies of children without disabilities (aged 3 to 6y), and an 8-year-old child with CP (spastic diplegia) indicated that modifications to the standard TUG test, as described by Podsiadlo and Richardson (1991), were required for the test to be used in children. These were as follows.

(1) A concrete task was used in that children were asked to touch a target on on a wall, compared to the more abstract instructions of the standard TUG. Abstract instructions

(2)

have been shown to limit performance in children with CP (Van der Weel et al. 1991).

(2) Instructions were repeated during the test. A seat with a backrest but without arms was selected from the children’s environment. The seat height was acceptable if the child’s knee angle was 90˚(SD 10) flexion with feet flat on the floor.

(3) Children were allowed to behave spontaneously, so no qualitative instructions (e.g. ‘walk as fast as you can’) were given to ensure a naturalistic performance for ecological validity (Bronfenbrenner 1977).

(4) Timing was started as the child left the seat, rather than on the instruction ‘go’, and stopped as the child’s bottom touched the seat, in order to measure movement time only.

PROCEDURE

Ethical approval was obtained from the Human Research Ethics Committees of The University of Melbourne and the Royal Children’s Hospital (RCH), Melbourne, Australia. Partic- ipants were recruited from children attending nearby schools, kindergartens, child-care centres, special schools, and the physiotherapy department of the RCH. Testing was conducted in familiar settings with other children and teachers present.

Written informed consent from was obtained from the parent or guardian of each child.

In both studies, the seat was selected and the height of the seat recorded. With the child seated and feet flat on the floor, goniometric measurement of the right knee angle was recorded. The following instructions were then delivered to the child slowly and clearly: ‘This test is to see how you can stand up, walk, and touch the star, then come back to sit down. The stopwatch is to time you. I am going to ask you to do the test three times (hold up three fingers) or twice (if child had a disability). After I say “go”, walk and touch the star, and then come back and sit down. Remember to wait until I say “go”. This is not a race; you must walk only, and I will time you. When you are ready… go! Don’t forget to touch the star, come back and sit down’. Although a demonstration was provided initially, this was not required for subsequent tests.

The investigator sat on a chair, in clear view of the child.

Children were tested in small groups. Every completed TUG test was scored and behavioural variations were noted. Some children displayed exuberant antics, such as hopping, which interrupted the performance of the test. The same investigator conducted all testing procedures for the study of typically developing children. Three raters, all physiotherapists who had completed a training session with the TUG modified for children, undertook testing of the children with disabilities, and each rater tested a different group of these children. In a study of TUG undertaken by our group in children in Indonesia, interrater reliability for three raters at two time points was high (intraclass correlation coefficient [ICC] [2,1]=0.81; Takarini et al. 2003).

In Study 1, three trials were conducted within a single session (time 1) then another three trials were repeated after a break of 10 to 20 minutes on the same day (time 2). Three trials were repeated at 1 week (time 3) and after 5 months (time 4) in the same children.

In Study 2, two trials were conducted at time 1 and then another two trials after 10 to 20 minutes (time 2). Participants with CP were classified by their treating physiotherapist as being at one of five levels of the GMFCS (Palisano et al. 1997) and those with spina bifida according to lesion level in con-

junction with GMFCS levels. A subgroup with CP were tested on the GMFM (Russell et al. 1989) by three independent paediatric physiotherapists; each tested different participants and all had GMFM criteria qualifications.

STATISTICAL ANALYSIS

Statistical analysis was performed using SPSS for Windows (version 10.0.5). Reliability was analyzed using the ICC, including the 95% confidence intervals (CIs). This index is a reliability coefficient calculated using variance estimates obtained through an analysis of variance (ANOVA) and, therefore, reflects both the degree of correspondence and the degree of agreement among ratings. The closer the ICC coefficient is to 1, the better the agreement of repeated trials. Single measure ICC (1,1) was used for within-session reliability at time 1 and time 2 for children with and without disabilities. Using the mean of three scores for each child on the first and subsequent occasions, the average measure reliability ICC (1,3) was calculated for between-session reliability (response stability) on the same day (time 1 and time 2) and for test–retest reliability after 1 week (time 1 and time 3) for children without disability. The mean of two scores was used to calculate average measure ICC (1,2) for between-session reliability (response stability) on the same day for the group of young people with disability.

In addition to measuring within- and between-test reliability, consistency of repeated responses over time was measured and expressed in the units of the measurement using the standard error of measurement (SEM). The SEM for a set of scores from a large sample was estimated as follows:

SEM=s_x√(1–r_xx), where s_xis the standard deviation of the set of observed scores, and r_xxis the reliability coefficient, i.e. the ICC, for those data (Portney and Watkins 2000).

Parametric tests were used as data for the typically developing group were normally distributed. Differences between males and females, and between younger and older children were analyzed at time 1 using independent t-tests. Age was not considered as a factor in the study of young people with disabilities for this paper. Responsiveness, i.e. whether scores changed after 5 months for the non-impaired group, was examined by comparing the mean scores at time 1 and time 4 using paired t-tests, and analyzing the total group as well as the two age groups separately. For both the independent and paired t-tests the level of significance was set to 0.025 using a Bonferroni correction.

To compare the mean scores of TUG to the classification levels of the GMFCS, the non-parametric Kruskal–Wallis one- way ANOVA was used because the assumption of homogene- ity of variances was violated. A Mann–Whitney Utest was then used to determine which levels of GMFCS were significantly different (p=0.017). Descriptive information was recorded using accepted taxonomic classification according to diagnosis, part affected, and severity.

Finally, the strength of the relations between TUG and the Standing and Walking dimension scores of the GMFM was eval- uated using Spearman’s non-parametric correlation coefficient.

Results

STUDY1: TUG IN CHILDREN WITHOUT PHYSICAL DISABILITIES

One hundred and seventy-six children aged between 3 and 9 years performed three trials at time 1. Three preschool children failed to return for same-day retest (time 2, n=173).

(3)

One hundred and fifty-one children returned for testing 1 week later, then 5 months later 128 children were retested.

Attrition was due to children being absent on the day, and one preschool centre was unable to accommodate retesting after 5 months. No children were excluded for gross motor anom- alies. At time 1, 86 children were in the preschool group (3 to 5 years) and 90 were in the primary group (5 to 9 years). Most children for whom behavioural variations were recorded were in the preschool group (27%; see Table I). The height of the seat used, consistent between testing, was confirmed as

appropriate by measurement of the right knee angle (mean 93.3˚, SD 8.8).

RELIABILITY

Good reliability was shown for within-session trials for times 1, 2, and 3, with ICCs of 0.80, 0.89, and 0.85 respectively (see Table II).

Response stability (same-day retest) using mean scores for time 1 and time 2 was good (ICC=0.89; 95% CI=0.86 to 0.92), as was test–retest reliability (time 1 and time 3; ICC=0.83; 95%

Table II: Within session, same day retest, and test–retest (after 1 week) reliability for participants without disabilities

Reliability test Variable tested n Mean (SD) ICC 95% CI

Within-session (Time 1) t1.1 t1.2 t1.3 176 5.9 (1.3) 0.80 0.75–0.84

t1.1 6.0 (1.5)

t1.2 5.9 (1.3)

t1.3 5.9 (1.3)

t2.1 5.9 (1.5)

t2.2 5.9 (1.5)

t2.3 5.9 (1.5)

t3.1 5.7 (1.2)

t3.2 5.8 (1.2)

t3.3 5.7 (1.2)

Same-day retest Time 1, 2 173 5.9 (1.3) 0.89 0.86–0.92

Time 1 5.9 (1.3)

Time 2 5.9 (1.5)

Test–retest (1 week) Time 1,3 173 5.8 (1.2) 0.83 0.77–0.88

Time 1 5.9 (1.3)

Time 3 5.7 (1.1)

Same-day retest, preschool Time 1,2 83 6.7 (1.2) 0.82 0.72–0.88

Time 1 6.7 (1.2)

Time 2 7.0 (1.3)

Same day, preschool (B) Time 1,2 51 6.7 (1.2) 0.88 0.8–0.93

6.7 (1.3) 6.7 (1.0)

Test–retest (1 week) preschool Time 1,3 83 6.6 (1.1) 0.61 0.39–0.75

6.7 (1.2) 6.5 (1.0)

Test–retest (1 week) preschool (B) Time 1,3 51 5.1 (0.8) 0.71 0.5–0.83

6.7 (1.3) 6.5 (1.0)

Same-day retest, primary Time 1,2 90 5.1 (0.8) 0.76 0.61–0.85

Time 1 5.2 (0.8)

Time 2 4.9 (0.8)

Test–retest (1 week) primary Time 1,3 90 5.1 (0.8) 0.83 0.73–0.89

Time 1 5.2 (0.8)

Time 3 5.0 (0.8)

ICC, intraclass correlation coefficients; CI, confidence interval; t, trial (within-session); B, preschool behavioural variations group excluded.

Table I: Demographics of participants without disabilities at time 1

Demographics All Male Female Preschool Preschool with Primary Primary with

behavioural behavioural

variations variations

n 176 94 82 86 27 90 3

Mean age (SD), y:m 5:10 (1:8) 5:9 (1:7) 5:11 (1:9) 4:4 (0:5) 4:3 (0:5) 7:3 (1:1) 5.1 (0:1)

(4)

CI=0.77 to 0.88; see Table II).

Examination of subgroups revealed that for the preschool group (n=83) there was good response stability (same-day retest; ICC=0.82) and moderate test–retest reliability (1 week; ICC=0.61). When children with behavioural variations were excluded and data re-examined in the preschool group (n=51), both same day (ICC=0.88) and test–retest reliability improved (ICC=0.71; see Table II). In the primary group (n=90), response stability (ICC=0.76) and test–retest reliability (ICC=0.83) were good.

Mean (SD) scores for all groups have been tabled (see Table II). The mean score of three trials at time 1 for the whole group (n=176) was 5.9s (SD 1.3), with a range of 3 to 13s. The range of scores was 3 to 13s for preschool children and 3.1 to 8s for primary school children. The magnitude of random error in original units of measurement (SEM) calculated for the three trials at was 0.6s time 1, and 0.4s for the three trials at time 2.

Expressed as a percentage of the mean scores, the size of the SEM was generally less than 10%.

There was no statistically significant difference in mean scores between males (n=94, 6.0s, SD 1.3) and females (n=82, 5.9s, SD 1.3) at time 1 (p=0.859). Preschool children (mean age 4y 4mo, SD 5mo; range 3 to 5y) took significantly longer 6.7s (SD 1.3) to complete the TUG than primary school children (mean age 7y 3mo, SD 1y 1mo; range 5y 1mo to 9y 3mo; 5.2s, SD 0.8; p=0.001; Fig. 1).

RESPONSIVENESS

Responsiveness to change over time, or sensitivity, was examined in children who returned for testing after 5 months (time 4; n=128). There was a significant decrease in mean TUG scores from time 1 (5.7s, SD 1.3) to time 4 (5.2s, SD 0.9;

p=0.001).

When the two age groups were examined separately, there was a substantial and significant decrease in mean TUG scores

for preschool children from time 1 (6.7s, SD 1.3, 95% CI 6.6 to 6.7s) to time 4 (5.7, SD 1, 95% CI 5.4 to 6.0s; p=0.001), and a smaller significant reduction in primary school children (time 1: 5.3, SD 0.8, 95% CI 5.0 to 5.3s; time 4: 5.0s, SD 0.7; 95%

CI 4.8 to 5.1s; p=0.02). The change in preschool children and relative stability of primary school children is demonstrated in Figure 2.

STUDY2: TUG IN YOUNG PEOPLE WITH DISABILITIES

Study 2 participants were a convenience sample of 41 young people diagnosed with CP (n=33) or spina bifida (n=8; see Table III). The age distribution was bimodal with 63% aged 3 to 12 years. Comments were made about behavioural variations during the performance of the TUG for 5% of participants but this was not considered significant. All young people were able to walk independently; orthoses or assistive devices, such as crutches or walkers, were kept consistent between trials.

There was good within-session reliability of the two trials at time 1 (ICC=0.98; 95% CI 0.97 to 0.99) and at time 2 (ICC=0.98;

95% CI 0.88 to 0.99). Reliability between the same-day retest at time 1 and time 2 was also high (ICC=0.99; 95% CI 0.91 to 0.99; Table IV).

939 1840 4

Table III: Demographics of participants with disabilities

Diagnosis All Cerebral palsy Spina bifida

n 41 33 8

Male 20 16 4

Female 21 17 4

Age

Range (y) 3–19 3–17.5 5–19

Mean (SD), y:m 8:11 (4:3) 8:7 (4:1) 10:10 (4:10)

Preschool Primary school Start 10min 1wk 5mo

Age group Time

14

12

10

8

6

4

2

7.5

7.0

6.5

6.0

5.5

5.0

4.5

Time (s) Mean score (s)

Primary group Preschool group

Figure 2:Change in mean timed ‘Up & Go’ scores over time according to age group. Error bars represent 95%

confidence intervals.

Figure 1:Box plot showing preschool (n=86) and primary school (n=90) groups comparing timed ‘Up & Go’ scores for all participants at time 1. Rectangles represent 50% of participants, whiskers show range of values, and bold line shows median value. ●●, mild outliers; *, extreme outlier.

(5)

TUG SCORES IN RELATION TO GMFCS AND LESION LEVEL

Analysis of TUG scores in relation to functional ability (GMFCS level) using a one-way Kruskal–Wallis ANOVA indicated significant differences (p=0.001) between participants at GMFCS level I (n=25; mean 8.3s, SD 1.8), level II (n=8; mean 10.8s SD 1.8) and level III (n=8; mean 28.1s, SD 13.5). A Mann–Whitney Utest showed that level I was significantly different from level II (p=0.017) and level III (p=0.001), and that level II was

significantly different from level III (p=0.001; Fig. 3).

Within the group with CP, the severity of impairment using the accepted taxonomic classification was also reflected in the mean TUG scores at time 1. Results were: spastic hemiplegia (n=4) mean 8.4s, SD 1.3; spastic diplegia (n=22) mean 10.1s, SD 2.4; and spastic quadriplegia (n=6) mean 28s, SD 26. Most of the group with spina bifida had myelomeningocele with low lesion level (n=7) and their mean TUG score was 8s (SD 1.5).

Mean scores of two trials at time 1 in two participants who did not fall into any of these groups were 38s for a child classified as athetoid/ataxia, and 5s for a child with lipoma (Fig. 4).

GMFM SCORES IN RELATION TO TUG SCORES

A subgroup of young people with CP (n=22) was concurrently tested using the GMFM. This analysis revealed that there was a moderate negative correlation between TUG scores and the GMFM (rho=–0.524, p=0.012), with lower TUG scores being associated with higher percentages of GMFM scores for the Standing and Walking dimensions.

Discussion

We have shown that all children tested could competently complete the test and that, after a demonstration, the TUG was reliable in children without disability aged 3 to 9 years, and in young people with a physical disability aged 3 to 19 years. The test should be repeated if TUG performance is interrupted by normal behavioural variation (e.g. hopping). A contraindica- tion for use of the TUG in children is an inability to understand instructions. This reinforces a Canadian study that noted that the TUG was not suitable for people with cognitive deficits (Rockwood et al. 2000). Because ICC model 1 was used, the results could be generalized to other raters using the same instruction and training, although further studies are needed to examine interrater reliability.

Mean TUG score in 176 Australian children aged 3 to 9 years for three trials at time 1 was 5.9s with a range of 3 to 13s. This range reflected the range of scores without consideration of behavioural variations. The range for the older group of children was 3.1 to 8s. These TUG scores are similar to those reported in studies of children in other countries. The mean TUG score of children from Pakistan, aged 5 to 13 years (n=180), was 5.1s (Habib et al. 1999); in Indonesian children between 4 to 9 years (n=501) the mean score was 6.1s (Takarini et al. 2003); and the mean score in Filipino children aged 6 to 12 years (n=144) the TUG score was 6.7s (M Ruiz, personal communication 2003). This suggests potential for intercultural studies. Although TUG scores have been examined in a wide range of typically developing participants (Podsiadlo and Richardson 1991, Newton 1997, Habib et al.

1999, Rockwood et al. 2000, Eekhof et al. 2001, Steffen et al.

2002, Takarini 2003), there are no reported TUG data for people aged between 13 and 60 years.

Younger children demonstrated a decrease in TUG scores over time, indicating that the TUG was responsive to changes in a child’s performance over time. Although the reduction was statistically significant, it amounted to a period of 1s, so the improvement may be due to an increase in body size and strength rather than neural maturation. This interpretation is supported by Sutherland’s (1997) observations that a mature gait pattern emerges at 3¹⁄²to 4 years, and by Habib and Westcott’s (1998) findings that height predicted TUG scores in older children. The slower mean times (10.5 to

SH SD SQ A/AtxQ MM LP

(n=4) (n=22) (n=6) (n=1) (n=7) (n=1) Type of disability

I lI III

GMFCS level

60

50

40

30

20

10

0 60

50

40

30

20

10

0

Time (s)Time (s)

Figure 4:Timed ‘Up & Go’ scores according to type of disability: SH, spastic hemiplegia; SD, spastic diplegia; SQ, spastic quadriplegia; A/AtxQ, athetoid/ataxia

quadriplegia; MM, myelomeningocele; LP, lipoma.

Rectangles represent 50% of participants, whiskers show range of values, and bold lines show median value.

Figure 3:Comparison of timed ‘Up & Go’ scores for Gross Motor Function Classification System (GMFCS) levels I (n=25), II (n=8), and III (n=8). Rectangles represent 50%

of participants, whiskers show range of values, and bold lines show median value.

(6)

14.8s) reported for people over 80 years with no pathology (Eekhof et al. 2001, Steffen et al. 2002) could perhaps also be explained by age-related changes.

TUG IN PEOPLE WITH DISABILITIES

TUG scores differentiated between young people with disabilities classified at GMFCS levels I, II, and III, However, further research with larger groups of ambulant participants is required to determine whether a clear demarcation exists between the levels. The ‘criterion standard’ functional measure for children with CP, the GMFM, takes an hour to admin- ister and requires specific training for results to be reliable.

The moderate negative correlation (rho=–0.52) of TUG scores with the Standing and Walking dimensions of GMFM indicates the potential for TUG to be administered between GMFM testing sessions to provide an indication of progress or deterioration with regard to functional mobility. A further study in a larger group of young people with disability may provide stronger evidence of the validity of TUG as a functional measure compared with GMFM. The TUG appears to be sensi- tive to changes in functional mobility, as demonstrated by Andersson et al. (2003) who showed that both TUG and GMFM scores significantly improved in a study of static strength training in adults with CP compared with a control group.

TUG AS A MEASURE OF FUNCTIONAL MOBILITY

For an ambulatory child to be independent in functional mobility, he or she must be able to rise up, walk to perform tasks, and then sit down to rest. This requires sophisticated control of balance and movement through planning, initiating, exe- cuting, and completing an integrated movement sequence.

For social acceptance, these actions must be able to conform to the pace of peers. The TUG requires this level of motor control and it is commonly cited as a measure of functional or basic mobility (Shumway-Cook et al. 2000, EeKhof et al.

2001, Geiger et al. 2001), as a test of balance (Habib et al. 1999, Gill-Body et al. 2000), and postural stability (Westcott et al.

1997). In functional movement, measures of static and dynamic balance need to be differentiated and studied (De Weerdt and Spaepen 1999, p 211) and the TUG would appear to be an appropriate test for dynamic balance.

Although young people with hemiplegia and spastic diplegia had slower mean scores than the children without disabilities, those with spastic quadriplegia had mean scores that were six times slower than typically developing children of a similar age. This has implications for the activity and partici- pation of these young people at home, at play, and at school.

These results also indicate that TUG could be a useful objective measurement (qualifier) of disability for the mobility domain of the World Health Organization’s (WHO) International Classification of Functioning, Disability and Health (ICF; WHO

2001). A review of functional mobility using the TUG would determine the level of assistance required for optimal function and include how these children could be assisted in different environments.

The TUG test may be useful clinically to monitor change over time, particularly following interventions that aim to improve gait, speed of transitions, and turns in those with physical disabilities. Further study of the TUG in young people with a disability is necessary because in this study only same day retesting was performed. The ICC values obtained, though high, do not, of themselves, provide evidence of the amount of fluctuation in performance expected with this measurement tool. The SEM of scores obtained on separate tests performed at least 1 week apart would address this issue by providing the degree of variation, in scale units, for the population in question (Keating and Matyas 1998), and would assist in determining whether a change in score can be interpreted as clinically significant.

In this study, TUG tests were conducted in the child’s usual environment to maximize ecological validity. Measurement of spontaneous activity of children in their environment has been considered a benchmark for assessment of children (Bronfen- brenner 1977, Haley et al. 1993). Environmental and personal (or contextual) factors are an important feature of the ICF (WHO 2001a). Because the TUG comprises a series of common integrated or linked movements that are frequently performed in the child’s environment, this test could be considered a rea- sonably ecologically valid tool, particularly compared with other tests, such as those conducted in a gait laboratory or hospital clinic.

We modified the procedure for the use of the TUG in children to ensure that children clearly understood the task. We also ensured that movement time rather than a combination of reaction time and movement time was measured. In this study, the TUG was performed without a practice trial or provision of qualitative instructions. Our results showed that, as expected, it was behavioural variations in younger children aged less than 5 years (although expected or normal responses for children of that age) that affected both the range and reliability of TUG scores. This supports the view that tests should be repeated if younger children exhibit behavioural variations, as these may mask true physical ability.

LIMITATIONS OF THE STUDY

We did not formally screen the children without disabilities to ensure that they were typically developing. The participants were a convenience sample and, therefore, may not be repre- sentative of a wider population, although they were represen- tative of the population of children attending preschool and primary school in Melbourne. Some of the children were very young and this may have influenced the results. A limitation of the study of young people with disabilities was that the number

Table IV: Range of scores within Gross Motor Classification System (GMFCS) levels, and same-day retest reliability for participants with disabilities

n Mean age (SD), y:m Mean score SD ICC Range

All participants 41 9:0 (4:3) 12.7 9.7 0.99 4.5–50.2

Level I 25 8:6 (3:9) 8.3 1.8 0.88 4.5–12.1

Level II 8 11:0 (5:2) 10.9 1.8

Level III 8 9:3 (5:2) 28.1 13.5

(7)

of participants was small, and neither interrater reliability nor test–retest reliability after 1 week was examined. Further studies in this population are recommended.

Conclusion

We have shown that the modified TUG is reliable in children as young as 3 years of age, provided that the child can understand instructions and the test is performed in an integrated manner without behavioural variation. The responsiveness of the TUG was demonstrated by significant change over time in the younger children. We have also shown that the TUG is both reliable and valid for ambulant young people with physical disabilities. The TUG integrates transitions and walking skills, and provides a measure of capability that is meaningful to most people. Examination of GMFCS levels and TUG scores suggests that further studies of TUG performance are indicated to define performance levels that could clearly separate level I and level II as well as to investigate the degree of fluctuation in scores over time to determine clinically significant changes.

TUG could be an evaluative outcome measure in rehabilitation, provided that interventions are directed towards the movement elements affecting its performance. Population studies to norm-reference the TUG could provide a useful benchmark to establish baseline levels of functional mobility in different environments (e.g. at school) to inform health-care providers of the need for assistive devices. Further studies in children of this meaningful, practical, and inexpensive test are warranted.

DOI: 10.1017/S0012162205001027 Accepted for publication 2nd August 2004.

Acknowledgement

We acknowledge the financial support of the Physiotherapy Research Foundation, Australia.

References

Andersson C, Grooten W, Hellsten M, Kaping K, Mattson E. (2003) Adults with cerebral palsy: walking ability after progressive strength training. Dev Med Child Neurol45:220–228.

Bronfenbrenner U. (1977) Towards an experimental ecology of human development. Am Psychol32:513–521.

Creel GL, Light KE, Thigpen MT. (2001) Concurrent and construct validity of scores on the timed movement battery (TMB). Phys Ther 81:789–798.

De Weerdt W, Spaepen A. (1999) Balance. In: Durward BR, Baer GD, Rowe PJ, editors. Functional Human Movement: Measurement and Analysis.Oxford: Butterworth-Heinemann. p 204–218.

Eekhof J, De Bock GH, Schaapveld K, Springer MP. (2001) Functional mobility assessment at home. Timed up and go test using three different chairs. Can Fam Physician47:1205–1207.

Geiger RA, Allen JB, O’Keefe J, Hicks RR. (2001) Balance and mobility following stroke. Effect of physical therapy interventions with and without biofeedback/forceplate training. Phys Ther81:995–1005.

Gill-Body KM, Beninato M, Krebs DE. (2000) Relationship among balance impairments, functional performance, and disability in people with peripheral vestibular hypofunction. Phys Ther 80:748–758.

Gray C, Wall JC, Farris J, Rich A, van Camp J. (2001) Balance and the post-polio syndrome. Phys Ther81:64.

Habib Z, Westcott S. (1998) Assessment of anthropometric factors on balance tests in children. Ped Phys Ther10: 101–109.

Habib Z, Westcott S, Valvano J. (1999) Assessment of balance abilities in Pakistani children: a cultural perspective. Ped Phys Ther11:73–82.

Haley SM. (1991) Motor assessment tools for infants and young children: a focus on disability assessment. In: Forssberg H, Hirschfeld H, editors. Movement Disorders in Children.Basel:

Karger. p 278–283.

Haley SM, Baryza MJ, Blanchard Y. (1993) Functional and naturalistic frameworks in assessing physical and motor disablement. In: Wilhelm IJ, editor. Physical Therapy Assessment in Early Infancy. New York: Churchill Livingstone. p 225–256.

Keating J, Matyas T. (1998) Unreliable inferences from reliable measurements. Austr J Physiother44:5–10.

Lowes LP, Westcott S. (1995) Relationship of force output and range of motion to mobility tests in children with cerebral palsy.

Combined sections meeting: American Physical Therapy Association. Ped Phys Ther7:200. (Abstract)

Morris S, Morris ME, Iansek R. (2001) Reliability of measurements obtained with the timed ‘up & go’ test in people with Parkinson disease. Phys Ther81:810–818.

Newton RA. (1997) Balance screening of an inner city older adult population. Arch Phys Med Rehabil78:587–591.

Palisano R, Rosenbaum P, Walter S, Russell D, Wood E, Galuppi B.

(1997) Development and reliability of a system to classify gross motor function in children with cerebral palsy. Dev Med Child Neurol 39:214–223.

Podsiadlo D, Richardson S. (1991) The timed ‘up and go’: a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc39:142–148.

Portney LG, Watkins MP. (2000) Foundations of Clinical Research:

Applications to Practice. 2nd edn.New Jersey: Prentice Hall Health. p 578–579.

Rockwood K, Awalt E, Carver C, MacKnight C. (2000) Feasibility and measurement of the functional reach and the timed up and go tests in the Canadian study of health and aging. J Gerontol A Biol Sci Med Sci55:M70–M73.

Russell DJ, Rosenbaum PL, Cadman DT, Gowland C, Hardy S, Jarvis S.

(1989) The gross motor function measure: a means to evaluate the effects of physical therapy. Dev Med Child Neurol 31: 341–352.

Shumway-Cook A, Brauer S, Woollacott M. (2000) Predicting the probability for falls in community-dwelling older adults using the timed up and go test. Phys Ther80:896–903.

Steffen TM, Hacker TA, Mollinger L. (2002) Age- and gender-related test performance in community-dwelling elderly people: six- minute walk test, Berg balance scale, timed up & go test, and gait speeds. Phys Ther82:128–137.

Sutherland D, Olshen R, Cooper L, Woo SLY. (1980) The development of mature gait. J Bone Joint Surg62A:336–353.

Sutherland DH, Olshen R, Biden EN, Wyatt MP. (1988) The Development of Mature Walking. Clinics in Developmental Medicine No. 104/105.London: Mac Keith Press.

Sutherland D. (1997) The development of mature gait. Gait Posture 6:163–170.

Takarini NT, Williams EN, Denehy L. (2003) TUG in Indonesian children. Paper presented at the World Confederation of Physical Therapists Congress, Barcelona, 2–12th June 2003.

Van der Weel FR, Van der Meer AHL, Lee DN. (1991) Effect of task on movement control in cerebral palsy: implications for assessment and therapy. Dev Med Child Neurol33:419–426.

Westcott SL, Lowes LP, Richardson PK. (1997) Evaluation of postural stability in children: current theories and assessment tools. Phys Ther77:629–645.

Williams EN, Carroll SR. (1999) The timed up-and-go test in normal children of different ages. Paper presented at the National Paediatric Physiotherapy Conference, Sydney, Australia, 7–10th September 1999.

Williams EN, Carroll SR, Galea MG. (2001) The timed up-and-go test for children: TUG for children with a disability. Austr J Physiother 42:140–142.

World Health Organization. (2001) WHO family of international classifications. http://www.who.int/classification/ (accessed 10th May 2005).