ORIGINAL ARTICLE
Further development of the High-level Mobility Assessment Tool
(HiMAT)
GAVIN WILLIAMS1,2, JULIE PALLANT3, & KEN GREENWOOD4
1Epworth Hospital, Melbourne, Australia, 2School of Physiotherapy, 3School of Rural Health, University of Melbourne,
Australia, and 4School of Health Sciences, RMIT University, Melbourne, Australia
(Received 17 December 2009; accepted 27 April 2010)
Abstract
Primary objectives: The high-level mobility assessment tool (HiMAT) was developed to measure high-level mobility
limitations following traumatic brain injury (TBI). Rasch analysis was used in the development to ensure cognitive deficits
would have a minimal impact on performance. The main aim of this study was to investigate the dimensionality of the
HiMAT using recently developed advanced testing procedures.
Research design: Results from the original sample of 103 adults with TBI used to develop the HiMAT were re-analysed using
the RUMM2020 program. Revised minimal detectable change (MDC95) scores were also calculated.
Main outcomes and results: Rasch analysis of all 13 HiMAT items suggested that the scale was multidimensional, showing
a clear separation between the stair and non-stair items. The nine non-stair items of the HiMAT showed good overall fit,
excellent internal consistency, with no disordered thresholds or misfitting items, however removal of one item was required
to ensure a unidimensional scale. The final 8-item solution showed good model fit (p¼0.93), excellent internal consistency
(PSI¼0.96), no disordered thresholds, no misfitting items and no differential item functioning for age or sex. The revised
HiMAT total score is 32 points and the MDC95 was calculated to be 2 points.
Conclusion: The results of this study demonstrate that the revised HiMAT is unidimensional and valid to use in
rehabilitation and community settings where there is no access to stairs.
Keywords: Brain injuries, rehabilitation, gait disorders, neurologic, outcome assessment
Introduction
The high-level mobility assessment tool (HiMAT)
was originally developed to quantify the high-level
mobility limitations of people with traumatic brain
injury (TBI) [1, 2]. A key priority in the development
of the HiMAT was to ensure that only the
physical component of high-level mobility was
quantified. This priority was instrumental in ensuring
cognitive deficits would have a minimal impact
on performance. Rasch analysis was used in the
developmental stages to ensure the unidimensionality
of the HiMAT items [2]. Recent advances in
Rasch analysis have indicated that further
investigation is required using the more advanced
approaches currently available to investigate dimensionality
[3]. Therefore, the main aim of this study
was to investigate the dimensionality of the HiMAT
using the recently developed advanced testing
procedures.
A second priority in the development of the
HiMAT was clinical utility. In order to facilitate
the uptake of the HiMAT into a wide range of
clinical and community settings, only items which
were minimally dependant on time, equipment and
resources were considered. Similar to many mobility
scales used in rehabilitation [4–9], the HiMAT also
Correspondence: Dr Gavin Williams, Physiotherapy Department, Epworth Hospital, 89 Bridge Rd, Richmond, 3121, Victoria, Australia. Tel: þ613 9426
8727. Fax: þ613 9426 8734. E-mail: gavin.williams@epworth.org.au
ISSN 0269–9052 print/ISSN 1362–301X online 2010 Informa Healthcare Ltd.
DOI: 10.3109/02699052.2010.490517
includes the ability to ascend and descend stairs.
The ability to negotiate stairs is important for dayto-
day mobility. Expert clinicians strongly endorsed
the stair items during the consultative phase in the
development of the HiMAT [1], yet in some
rehabilitation and community-based settings, a suitable
flight of stairs may not be available for testing.
Therefore, the secondary aim of this study was to
investigate the validity of summing the non-stair
items to generate a total (but reduced) HiMAT
score.
Methods
Participants
Data from the original sample of 103 people with
TBI recruited in the development of the HiMAT
were re-analysed in this study [2].
Procedure and data analysis
Rasch analysis was used to investigate the validity of
the HiMAT scoring system without the stair items.
The Conquest program [10] was used in the original
developmental stages of the HiMAT. However, due
to recent advances in Rasch modelling, the
RUMM2020 [11] program was chosen for use in
this study to investigate item fit, dimensionality,
internal consistency, differential item functioning,
response dependency and targeting of the non-stair
HiMAT items. In particular, additional testing using
the t-test procedure to compare scores from sub-sets
of items has been recommended when investigating
dimensionality [3].
The procedures undertaken to assess the psychometric
properties of the HiMAT were consistent
with those recommended by Pallant and Tennant
[12] and Tennant and Conaghan [13].
The overall
fit of the scale was evaluated using a chi-square
statistic with a non-significant p-value indicative of
adequate fit to the Rasch model. Fit of the individual
items and persons were assessed using two indicators:
a non-significant chi-square statistic and a fit
residual value within the range 2.5. Differential
item functioning testing was conducted for age and
sex using analysis of variance.
The presence of
response dependency amongst the items was investigated
by inspecting the residual correlation matrix
for values exceeding þ0.3. Dimensionality testing
involved conducting a series of t-tests to compare
Rasch derived scores from two sub-sets of items
identified from principal components analysis (PCA)
of the residuals.
Less than 5% of tests should be
significantly different (or the lower bound of the
binomial confidence interval should overlap by 5%)
for the scale to be considered unidimensional.
A graph showing the distribution of Rasch derived
scores and corresponding item thresholds was used
to evaluate the targeting of the scale.
The main purpose of the original HiMAT was to
establish a scale which quantified the high level
mobility requirements for the societal roles of young
adults.
Therefore, it was important to establish that
the HiMAT was minimally susceptible to a ceiling
effect [14]. The exclusion of the stair items was
unlikely to introduce a ceiling effect to the revised
HiMAT because of their item location [2]. The
targeting of the non-stair HiMAT items was investigated
to ensure the revised scale remained
Figure 1. Distribution of scores and item thresholds on the revised 8-item HiMAT scale.
1028 G. Williams et al.
discriminative for a range of abilities, particularly at
the highest level.
A shorter version of the HiMAT, without the stair
items, may also change the original minimal detectable
change (MDC95) scores. The MDC95 scores
are used to determine the 95% confidence intervals
for detecting clinically meaningful change. The
MDC95 scores for the revised HiMAT were recalculated
from the original data sets [14, 15] using the
formula:
MDC95 ¼ Mean difference 1:96 SE
where SE is the standard error of measurement
derived from a reliability study [15].
Results
Rasch analysis of all HiMAT items
Rasch analysis of all 13 HiMAT items showed
adequate fit to the model (p¼0.82) with no
misfitting items. Dimensionality testing, however,
indicated that the scale was multidimensional,
showing a clear separation of the stair and nonstair
items.
Over 12% of respondents recorded
significantly different Rasch derived scores for
these two sets of items. This value exceeds the
recommended guideline of 5% and suggests that
these two sets of items should not be combined to
form a total score.
Rasch analysis of HiMAT non-stair items
Rasch analysis of the nine non-stair items of the
HiMAT indicated good overall fit to the model
(p¼0.99).
However, dimensionality testing indicated
that the scale was multidimensional, with
over 13% of respondents showing statistically significant
different scores on two sets of items identified
from PCA of the residual correlation matrix.
Removal of item 8 ‘bound–affected leg’ resolved
this dimensionality issue, with subsequent testing
identifying 5.10% of cases with significantly different
scores.
Although this value exceeded the recommended
value of 5%, the binomial confidence
interval (CI: 1–9%) included the value of 5%,
supporting the unidimensionality of the scale.
The final 8-item solution (with item 8 removed)
showed good model fit (p¼0.93) (Table I), excellent
internal consistency (PSI¼0.96), no disordered
thresholds, no misfitting items (Table II) and no
differential item functioning for age or sex. All
residual correlations were less than 0.34, suggesting
Table II. Individual item fit statistics for the final models of the HiMAT scale.
No. Item Location SE Fit Resid df Chi Sq df p
HiMAT—without stair items
1 walk 0.28 0.21 0.78 82 0.64 2 0.72
2 walk (backward) 2.36 0.20 0.23 82 1.04 2 0.59
3 walk (toes) 0.88 0.18 0.05 82 0.51 2 0.77
4 walk (obstacle) 1.55 0.19 1.07 82 0.96 2 0.62
5 run 0.94 0.17 0.65 82 1.80 2 0.41
6 skip 1.92 0.17 0.21 82 1.06 2 0.59
7 hop 2.41 0.17 0.15 82 0.41 2 0.81
9 bound (nonaffected leg) 0.20 0.18 0.79 82 2.19 2 0.33
SE¼Standard error, Fit Resid¼Fit Residual, Chi Sq¼Chi-square, df¼degrees of freedom, p¼probability.
Table I. Model fit statistics for original and revised HiMAT scale.
Action
Analysis
no.
Overall
model fit
Item Fit Resid
M (SD)
Person Fit Resid
M (SD) PSI
%
signif t-tests
HiMAT—all items 1 2¼19.38;
df¼26; p¼0.82
0.08 (0.52) 0.23 (0.68) 0.98 12.87%
(CI: 9–17%)
HiMAT—without stair items 2 2¼7.07;
df¼18; p¼0.99
0.23 (0.63) 0.25 (0.90) 0.96 13.27%
(CI: 9–18%)
Removal of item 8 3 2¼8.62;
df¼16; p¼0.93
0.29 (0.56) 0.21 (0.81) 0.96 5.10%
(CI: 1–9%)
Fit Resid¼Fit Residual, df¼degrees of freedom, p¼probability, SD¼standard deviation, PSI¼Person Separation Index,
CI¼confidence interval.
Revised HiMAT 1029
no serious response dependency among the items.
Targeting was appropriate for this group of patients,
with item thresholds covering the full range of scores
(Figure 1).
MDC95 scores
Due to the reduction in the number of items from 13
to 8, the maximum score for the revised version of
the HiMAT is now 32.
The MDC95 were recalculated
for the revised 8-item HiMAT. The revised SE
of measurement was calculated at 0.79 (revised ICC
0.99, revised mean difference between test and
re-test scores was 0.42 and the revised standard
deviation of the test and re-test scores were 7.82 and
7.97). The revised MDC95 was calculated to be
1.13 to 1.97.
Discussion
Although the intention of this study was to evaluate
how the HiMAT functioned when clinicians did not
have access to a flight of stairs, recent developments
in Rasch modelling identified a problem with
dimensionality, demonstrating a clear separation
between the stair and non-stair items. Removal of
the stair items (and one additional item) resulted in
a short 8-item scale with excellent psychometric
properties.
The separation of the stair and non-stair stair
items into two distinct sub-groups is a somewhat
surprising result.
The majority of mobility outcome
measures used in neurological rehabilitation incorporate
stair items [4–9], yet dimensionality has rarely
been evaluated in the development of these scales.
The use of Rasch analysis to develop or refine
mobility outcome scales is infrequent but increasing
[16]. In addition to the HiMAT, Belvedere and
de Morton [16] identified two further mobility scales
used in adult neurological rehabilitation which have
been evaluated with Rasch analysis.
The ABILOCO
[17] was developed using RASCH analysis and the
Dynamic Gait Index (DGI) [8] was refined from an
8-item to a 4-item scale following evaluation with
Rasch analysis. A fourth scale, the Rivermead
Mobility Index (RMI), has also been refined with
Rasch analysis [18].
The recent advances in Rasch analysis used to
investigate dimensionality involve the use of t-tests to
compare scores derived from sub-sets of items
identified using principal components analysis of
the residuals. This method was not used to investigate
the dimensionality of the ABILOCO, DGI or
RMI. The ABILOCO is a 13-item mobility questionnaire
developed for stroke which includes reciprocal
stair ascent [17].
Although developed and
validated with Rasch analysis, evaluation was
performed on questionnaire responses and physical
performance was not directly observed or tested.
A difference in dimensionality or item hierarchy
may exist when comparing observed performance
with self-report.
The DGI is an 8-item scale developed
for elderly people with balance and vestibular
disorders. The original version was revised to four
items when refined using factor analysis and Rasch
analysis [19].
The stair item was eliminated from the
final 4-item version due to a low single-construct
loading value identified using factor analysis. Rasch
analysis was also used in the refinement of the RMI
which contains two stair items.
The t-test procedure
was not used to investigate the dimensionality of the
RMI and the only change to this questionnairebased
scale was to omit the most difficult item due to
a low response rate.
This current study is the first to investigate the
dimensionality of a mobility scale which includes
stair items. Further investigation is required to
determine whether stair items should be included
in mobility scales, particularly in view of the importance
to be able to negotiate steps and stairs in
everyday life.
The results of this study demonstrate
that the revised HiMAT is unidimensional and valid
to use in rehabilitation and community settings
where there is no access to stairs.
The exclusion of
the stair items from the revised HiMAT further
enhances the practicality, feasibility and clinical
utility of the scale.
The original goal for developing the HiMAT was
to construct a measure of high-level mobility. The
eight non-stair items retained on the revised version
of the HiMAT still accomplishes this goal as all three
non-stair items which are more difficult than the
most difficult stair item have been retained.
Compared to the original version of the HiMAT,
the revised HiMAT no longer contains the two
easiest items (ascending and descending stair–
dependent).
This may make the revised HiMAT
more susceptible to a floor affect because it is less
able to quantify people with severe mobility limitations,
yet many other mobility scales such as the
ABILOCO [17], HABAM [20], FIM [21] or the
RMI [5] adequately quantify this level of ability.
Most importantly, the ability of the revised version
of the HiMAT to measure the high-level limitations
has been retained.
When the non-stair items were evaluated, a
residual problem with dimensionality was still evident.
Removal of the ‘bounding on to the moreaffected
leg’ item resolved this issue.
‘Bounding onto
the more-affected leg’ is very similar to ‘bounding
onto the less-affected leg’, yet the problem with
dimensionality was only associated with the first
item. One possible explanation may relate to how the
items were scored during the data collection phase
1030 G. Williams et al.
for the HiMAT. ‘Bounding onto the less-affected
leg’ requires subjects to push-off their more-affected
leg which may be difficult, but is not usually
perceived to be unsafe as the subject can then land
on their preferred or less-affected leg.
In contrast,
‘bounding on to the more-affected leg’ may enable
subjects to generate a greater push-off than can be
safely controlled when landing on the less-preferred
of more-affected leg. Hesitancy in landing on the
more-affected leg may have led some of the participants
to alter their performance or to refuse to
attempt the task.
When ‘bounding onto the moreaffected
leg’, failure and refusal were both scored as
‘0’. A ‘0’ score due to inability to bound may be an
accurate representation of performance, whereas a
‘0’ score due to refusal may be an accurate representation
of ability or the impact of fear or anxiety on
performance. Inability to differentiate between the
‘0’ scores limits investigations into the reasons for
the dimensionality problem with this item.
The MDC95 values for the revised 8-item HiMAT
were calculated to reflect the new change scores for
detecting clinically significant change. Results indicate
that a 2-point increase over a 3-month period is
an indication of clinically significant improvement.
Conversely, a 1-point deterioration over a 3-month
period is an indication of clinically significant deterioration.
These values are important for assisting
clinicians to interpret scores and establishing treatment
plans and goal-setting.
Conclusions
The revised 8-item HiMAT has excellent psychometric
properties and is valid for measuring highlevel
mobility in TBI. Exclusion of the stair items
improves the clinical utility of the revised HiMAT
due to the reduced number of items and time and
equipment requirements.
Declaration of interest: The authors report no
conflicts of interest. The authors alone are responsible
for the content and writing of the paper.
References
1. Williams G, Robertson V, Greenwood K, Goldie P,
Morris ME. The high-level mobility assessment tool
(HiMAT) for traumatic brain injury. Part 1: Item generation.
Brain Injury 2005;19:925–932.
2. Williams GP, Robertson V, Greenwood KM, Goldie PA,
Morris ME. The high-level mobility assessment tool (HiMAT)
for traumatic brain injury. Part 2: Content validity and
discriminability. Brain Injury 2005;19:833–843.
3. Tennant A, Pallant JF. Unidimensionality Matters! Rasch
measurement transactions 2006;20:1048–1051.
4. Carr J, Shepherd R, Nordholm L, Lynne D. Investigation of a
new Motor Assessment Scale for stroke patients. Physical
Therapy 1985;65:175–180.
5. Collen FM, Wade DT, Robb GF, Bradshaw C. The
Rivermead Mobility Index: A further development of the
Rivermead Motor Assessment. International Disability
Studies 1991;13:50–54.
6. Duncan P. Duke mobility skills profile. In: Shumway-Cook
A., Woollacott M, editors. Motor control: Theory and
practical applications. Baltimore: Williams and Wilkins;
1995. pp 319–322.
7. Lincoln N, Leadbitter D. Assessment of motor function in
stroke patients. Physiotherapy 1979;65:48–51.
8. Shumway-Cook A. Motor control: Theory and practical
implications. Baltimore: Williams & Wilkins; 1995.
9. Howe JA, Inness EL, Venturini A, Williams JI, Verrier MC.
The community balance and mobility scale-a balance
measure for individuals with traumatic brain injury. Clinical
Rehabilitation 2006;20:885–895.
10. Wu ML, Adams RJ, Wilson MR. ACER ConQuest:
Generalised item response modelling software. Melbourne:
The Australian Council for Educational Research; 1998.
11. Andrich D, Lyne A, Sheridan B, Luo G. RUMM2020. Perth:
RUMM Laboratory Pty Ltd; 2003.
12. Pallant JF, Tennant A. An introduction to the Rasch
measurement model: An example using the Hospital
Anxiety and Depression Scale (HADS). British Journal of
Clinical Psychology 2007;46:1–18.
13. Tennant A, Conaghan PG. The Rasch measurement model
in rheumatology: What is it and why use it? When should it be
applied, and what should one look for in a Rasch paper?
Arthritis & Rheumatism 2007;57:1358–1362.
14. Williams G, Robertson V, Greenwood K, Goldie P,
Morris ME. The concurrent validity and responsiveness of
the high-level mobility assessment tool for measuring the
mobility limitations of people with traumatic brain injury.
Archives of Physical Medicine and Rehabilitation 2006;87:
437–442.
15. Williams GP, Greenwood KM, Robertson VJ, Goldie PA,
Morris ME. High-Level Mobility Assessment Tool
(HiMAT): Interrater reliability, retest reliability, and internal
consistency. Physical Therapy 2006;86:395–400.
16. Belvedere S, De Morton NA. Rasch analysis in health care: A
systematic review of the characteristics of Rasch application
in the development or refinement of mobility scales. Journal
of Clinical Epidemiology. In press.
17. Caty GD, Arnould C, Stoquart GG, Thonnard J-l,
Lejeune TM. ABILOCO: A Rasch-built 13-item questionnaire
to assess locomotion ability in stroke patients. Archives
of Physical Medicine and Rehabilitation 2008;89:284–290.
18. Antonucci G, Aprile T, Paolucci S. Rasch analysis of the
Rivermead Mobility Index: A study using mobility measures
of first-stroke inpatients. Archives of Physical Medicine and
Rehabilitation 2002;83:1442–1449.
19. Marchetti GF, Whitney SL. Construction and validation of
the 4-item dynamic gait index. Physical Therapy 2006;86:
1651–1660.
20. MacKnight C, Rockwood K. Rasch analysis of the hierarchical
assessment of balance and mobility (HABAM). Journal of
Clinical Epidemiology 2000;53:1242–1247.
21. Granger CV, Hamilton BB, Sherwin FS. Guide for the use of
the uniform data set for medical rehabilitation. New York:
Uniform Data System for Medical rehabilitation Project
Office, Buffalo General Hospital; 1983.
Revised HiMAT 1031
Copyright of Brain Injury is the property of Taylor & Francis Ltd and its content may not be copied or emailed
to multiple sites or posted to a listserv without the copyright holder’s express written permission. However,
users may print, download, or email articles for individual use.
===========ABOUT ASSIGNMENT TASK ===========
AssignmentTask.com provides high-quality assignments help at wide range from Australia – UK & USA academic experts and also guarantees to make sure delivery assignments on time @ reasonable rate. OUR ASSIGNMENT EXPERT WRITERS MAKES EFFORTS TO DELIVER BEST QUALITY ASSIGNMENT HELP, ESSAY HELP, CASE STUDY HELP AND DISSERTATION EDITING HELP on each Order. 100% satisfaction is our primary motto. Our customers get 24 x 7 hours live chat supports.