TRUNK CONTROL TEST AS A FUNCTIONAL PREDICTOR IN STROKE

PATIENTS
E. Duarte, E. Marco, J.M. Muniesa, R. Belmonte, P. Diaz, M. Tejero and F. Escalada
From the Physical Medicine and Rehabilitation Department, Hospital de lEsperanca, IMAS, Universitat Auto`noma of
Barcelona, Barcelona, Spain
The purpose of this study was to evaluate prospectively the
Trunk Control Test (TCT) correlation at admission to
rehabilitation with length of stay, functional independence
measure (FIM), gait velocity, walking distance and balance
measured at discharge in 28 hemiparetic patients. FIM and
TCT were registered on admission. Outcome measures at
discharge were: FIM, gait velocity, walking distance and
balance assessed with the Berg Balance Scale and computer-
ized posturography. TCT was signicantly correlated with
length of stay (r = 0.722), discharge FIM (r = 0.738),
discharge motor FIM (r = 0.723), gait velocity (r = 0.654),
walking distance (p = 0.003), centre of gravity symmetry
(r = 0.601) and Berg Balance Scale (r = 0.755). Initial TCT
predicts the 52% of the variation in length of stay and 54%
in the discharge FIM. The predictive value of a compound
variable (TCT and admission FIM) reaches 60% of the
variation in length of stay and 66% in the FIM at discharge.
Key words: cerebrovascular accident, equilibrium,
functional gain, functional independence measure, stroke
assessment, rehabilitation.
J Rehabil Med 2002; 34: 267272
Correspondence address: Esther Duarte, Physical
Medicine and Rehabilitation Department, Hospital de
lEsperanca, IMAS, Sant Josep de la Muntanya 12,
ES-08024 Barcelona, Spain. E-mail:
[email protected]
Submitted December 7, 2001; accepted May 17, 2002
INTRODUCTION
The search for predictors of functional stroke outcome has
always been a matter of research in Physical Medicine and
Rehabilitation. The appropriate screening of patients in the acute
stage of stroke allows the clinician to decide early the placement
in which the rehabilitation programme will be most efcient (1).
Even though this distribution depends on available local
resources, the clinician must establish the functional prognosis
as soon as possible, by identifying neurological, functional and
psychosocial variables that have a demonstrated relationship
with the nal functional outcome in stroke patients. Functional
prediction models have been developed to identify stroke
patients with different functional outcomes and inpatient
rehabilitation length of stay (LOS) (2, 3). Balance status has
been identied as a signicant predictor of LOS in previous
studies (4). Sitting balance has a positive correlation with nal
disability in hemiparetic patients (57). There is a marked
positive correlation between sitting balance improvement and
disability measured with the Barthel Index during the hospita-
lization period in a rehabilitation unit (8). Sitting balance at 2
weeks after suffering a cerebrovascular accident (CVA) is also
correlated signicantly with walking ability at 6 months (9).
The Trunk Control Test (TCT) proposed by Colin & Wade,
administered at 6 weeks post-CVA, is a predictor of walking
ability at 18 weeks (10). TCT reliability and validity has been
demonstrated in stroke patients, as has its positive correlation
with disability at hospital discharge from inpatient rehabilitation
measured with the functional independence measure (FIM) (11).
It is not well known whether the TCT correlates with other
measures of static and dynamic balance which have been
validated in stroke patients, such as the Berg Balance Scale
(BBS) (12) and computerized Balance Master posturography
(13). It also remains unknown if trunk balance at admission
could correlate with outcome variables such as gait velocity and
walking distance perimeter at hospital discharge.
The aim of this study was to assess the TCT predictive value
at admission to inpatient rehabilitation of static and dynamic
balance, gait velocity, walking distance perimeter, LOS and
functional motor status achieved at hospital discharge in
hemiparetic patients who have suffered a CVA.
METHODS
Patients were eligible for inclusion in this study if they met the following
criteria: (1) hemiplegia secondary to a CVA in the 4 weeks preceding
admission to our inpatient rehabilitation unit, (2) no previous history of
motor disability and (3) absence of cognitive impairment that would
prevent the patients from following the instructions required to complet e
the tests.
From September 14 to November 30, 2000, a total of 36 subjects were
admitted to our stroke rehabilitation unit to follow a rehabilitative
programme with the diagnosis of hemiplegia post-stroke; 28 of them met
inclusion criteria for this study. Reasons for exclusion were: lack of
collaboration and/or cognitive level to complete the tests in 4 cases, 2
patients missed after being transferred to other hospital wards because of
medical complications and, nally, 2 cases of previous CVA.
Clinical characteristics of the sample were: mean age 64.5 years (SD
13.1), mean length of time from stroke onset to admission 15.33 days
(SD 6). There were 24 (85.7%) ischaemic and 4 (14.3%) haemorrhagi c
CVA cases. Hemiplegi a was right-sided in 12 (43%) and left-sided in 16
(57%) patients.
All patients were studied prospectivel y during their inpatient
rehabilitation stay. Disability was measured with the FIM (14) by a
trained rater. Trunk control was assessed with the TCT (10) in the rst
72 hours after admission to our stroke rehabilitation unit. The FIM
was divided into two subscales: motor (motFIM) and cognitive FIM
(15).
Outcome measurement s at hospital discharge from rehabilitation
2002 Taylor & Francis. ISSN 16501977 J Rehabil Med 34
J Rehabil Med 2002; 34: 267272
were: LOS, FIM and the motFIM, FIM and motFIM gain (difference
between admission and discharge of the same values), efciency (FIM
gain related to LOS) and motor efciency (motFIM gain related to LOS).
Gait velocity was assessed in a 10 m straight walkway. Patients were
asked to walk at their comfortable speed and then at their maximal safe
speed (with technical devices and/or standby supervision when
necessary) (16). Walking distance was categorized in three levels:
none, less than 50 m and more than 50 m.
Static and dynamic walking balance was assessed with computerized
posturography (Neurocom
1
Balance Master System 6.1) (Fig. 1).
Variables collected were: weight-bearing symmetry, centre of gravity
(COG), sway standing with open and closed eyes. Step length, step
width, walk speed and COG end sway are registered in the walking test
on a short 1.53 m platform from a position of still standing (Fig. 2). It
must be noted that gait testing using this method is only partially
representative of gait in a real environment , but it provides functionally
important values such as step width and the end-walk sway. Reliability
and validity of this system has been demonstrated to test static and
overall dynamic balance in stroke patients (13). Balance was also
clinically assessed with the BBS whose validity, reliability and
sensitivity to change have been also demonstrated in stroke patients
(12, 17).
The same team conducted all tests: two rehabilitation specialists
evaluated patients without knowledge of admission scores.
Statistics
Normality of the quantitative variables was checked both graphically
and with the Shapiro-Wilks test. Students t-test was used to assess the
relationship between groups in dichotomous variables and Scheffes
multiple comparison procedure in variables with more than 2 categories.
Relationship between the quantitative variables was assessed using
Pearsons correlation coefcient (r). To predict the value of the FIM at
discharge and the LOS in rehabilitation on the basis of the values of the
TCT and the FIM at admission, two stepwise multivariate regression
analyses were performed with LOS and discharge FIM as dependent
variables and TCT and admission FIM as explanator y variables. But
collinearity was observed (eigenvalue = 0.028, condition index = 10.2)
thus, a principal component analysis was done with the two explanatory
variables, and a new compound variable was created multiplying by
0.561 the sum of standardized values for TCT and admission FIM, i.e.:
compound

TCT 76
:
4
24
:
03

admission FIM84
:
0
22
:
38

0
:
561
:
Results were considered statistically signicant at p-values 0.05.
RESULTS
Twenty-eight patients were assessed in the rst 72 hours
following their hospital admission to our inpatient rehabilitation
unit: mean time after stroke onset was 15.3 days (SD 6). Mean
initial disability measured with the FIM and the motFIM was 84
(SD 22.4) and 52.7 (SD 19.2). Mean TCT score was 76.4 (SD
24). Of note are the lower scores observed in the rolling to the
sound side performance (18.6 (SD 8.1)) in comparison with the
rolling to the weak side performance (23.1 (SD 4.6)).
The mean LOS in rehabilitation was 19.2 days (SD 7.6). The
mean FIM and motFIM scores reached 109.5 (SD 11.5) and 76.8
(SD 10.9), respectively, at hospital discharge. The mean
motFIM gain during inpatient stay was 24.1 (SD 13.7) and the
mean FIM gain was 25.4 (SD 16.3). Efciency related to the
FIM and the motFIM was 1.3 (SD 0.7) and 1.29 (SD 0.6),
respectively.
Within the 48 hours preceding hospital discharge, the length
of time required to walk a distance of 10 m at a comfortable
walking velocity was 26 seconds (SD 17.4). At maximal safe
walking velocity, the mean time to cover the same distance was
19 seconds (SD 13.5).
As to walking distance, all patients were ambulant at hospital
discharge, but only 15 of them (53.6%) were able to cover
distances futher than 50 m.
In the computerized posturography evaluation, when static
standing, the COG was displaced to the sound side: the strong
limb sustained 54.3% of body weight. The 12 patients unable to
roll independently towards the sound side (TCT2 12) pre-
sented more asymmetric loading on the platform (61% of the
weight sustained by the sound limb) in contrast with the 16
patients with TCT2
>
12 (49.4% of the weight sustained by the
sound limb) (p = 0.006).
The mean COG sway was 0.4/sec (SD 0.2) with open eyes
and 0.6/sec (SD 0.4) with closed eyes. In the walking test, the
step length was 21.1 cm (SD 7.6), the step width was 17.8 cm
(SD 2.5), mean velocity was 23.5 cm/sec (SD 11.4) and nal
COG sway when patients were asked to stop walking was 2.65/
sec (SD 0.9).
Mean total score in BBS at hospital discharge was 42.7 (SD
13.3) over a theoretical maximum of 56.
When the relation between the TCT and the measured
variables was analysed, we found the following results: the
Fig. 1. Static and dynamic balance was assessed with computerized
posturography (Neurocom
1
Balance Master System 6.1) at
hospital discharge. Weight bearing symmetry, centre of gravity,
sway standing with open/closed eyes and a walking test were
registered.
J Rehabil Med 34
268 E. Duarte et al.
Fig. 2. Walk test assessed with computerized posturograph y (Neurocom
1
Balance Master System 6.1). The gure represent s the walking
test tracing: patients walk on a short 1.53 m platform: step length and gait speed are only partiall y representativ e of gait in a real
environment , but it provides functionall y important values such as step width and the end-walk sway. Three trials of the same test are done
for each patient and mean values are registered: drawings to the left represent the centre of gravity progression of the three walk trials.
J Rehabil Med 34
Trunk control test in stroke 269
TCT showed a positive correlation with the motFIM and the
FIM at admission (r = 0.648 and 0.587, respectively). No
statistical differences between these positive correlations were
found: in other words, the TCT correlated equally with the
motFIM and the FIM at admission to rehabilitation.
Admission TCT value showed correlation with the motFIM
and the FIM at hospital discharge (r = 0.723 and r = 0.738,
respectively). Correlation was inversely signicant between the
TCT and the LOS (r = 0.722) (all p-values 0.05) (see Fig. 3).
The TCT did not correlate with the motFIM/FIM gain and
efciency.
The TCT showed a statistically signicant difference
(p = 0.003) between patients whose walking distance at dis-
charge was longer than 50 m (mean TCT 88.9 (SD 14.3)) and
patients whose walking distance was shorter than 50 m (mean
TCT 61.9 (SD 25.2)).
Correlations were also statistically signicant between the
TCT and the time required to walk a 10 m straight walkway at a
comfortable and at maximal safe pace (r = 0.644 and
r = 0.654, respectively): the better initial TCT was, the higher
gait velocities at discharge were (Fig. 3).
Correlations with results measured by computerized posturo-
graphy were signicant as well, so that the higher TCT at
admission was, the less displacement (r = 0.601) and the better
Fig. 3. Correlations between Trunk control test and (a) discharge FIM (r = 0.738), (b) length of stay (LOS) (r = 0.722), (c) time required to
walk a 10 m straight walkway at a comfortable speed (r = 0.644) and (d) at maximal safe speed (r = 0.654), (e) centre of gravity (COG)
displacement (r = 0.601), and (f) posturograph y gait speed (r = 0.482). All p-values
<
0.05.
J Rehabil Med 34
270 E. Duarte et al.
posturography gait speed (r = 0.482) were (all p-values 0.05)
(see Fig. 3). No signicant correlations between the TCT and the
standing COG sway (with open and closed eyes, r = 0.207 and
r = 0.272 respectively), step length (r = 0.328) and step width
(r = 0.024) were found. The BBS presented a positive correlation
either with the motFIM and the FIM at admission (r = 0.630 and
r = 0.578) and the motFIM and the FIM at discharge (r = 0.838
and r = 0.809). Correlation was alsosignicant between the initial
TCTandthe BBS(r = 0.755) (all p-values 0.05). No statistically
signicant differences between these correlations were observed;
in other words, we cannot consider that the TCT correlates better
than the FIMat admission with the FIM at discharge. In the same
way, we cannot consider that the TCT correlates better than the
FIM at admission with the BBS at discharge.
Two multivariate regression analyses were performed with
LOS and FIM at discharge as dependent variables and
TCT FIM at admission as explanatory. Although in both
regressions the R
2
increase signicantly when the second
variable (admission FIM) was included: 0.521 vs. 0.60 for
LOS (p = 0.036) and 0.545 vs. 0.665 for discharge FIM
(p = 0.006), collinearity was observed. A new compound
variable was created using a principal component analysis.
This compound was highly correlated with LOS (r = 0771,
p
<
0.001) and discharge FIM (r = 0.815, p
<
0.001) (Fig. 4).
DISCUSSION
Trunk balance in the acute stage of stroke is a functional
outcome predictor (57). Clinical examination of balance is
done in daily practice when stroke patients are assessed in
rehabilitation units, but it is not usually done in a standardized
and objective way.
When analysing TCT results, we found in most of the patients
that rolling to the weak side (TCT1) was easier than rolling to
the sound side (TCT2). This observation might be explained by
the existing difculty in mobilizing hemiplegic limbs, as has
also been reported by other authors in previous TCT studies
(10, 11). Most of the patients (89.3%) achieve sitting in a
balanced position on the edge of the bed without assistance for a
duration of 30 seconds. This suggests that discriminative ability
of the TCT4 is lower than the remaining TCT movements (11).
In our study, the TCT correlates with the FIMat admission and
at discharge (both FIM and motFIM). Franchignoni et al. (11)
observed that the correlation of the TCT at admission with the
motFIM at discharge is even higher than correlation of the
motFIM at admission with the motFIM at discharge. In our case,
correlation between the TCT at admission and the FIM at
discharge is not signicantly higher than correlation between the
FIM at admission and the FIM at discharge. The four patients
excluded for not being able to complete TCT at admission for
cognitive impairment reported lower mean FIM admission and
discharge scores: 57.3 (SD32.1) and83.6 (SD24.5), respectively.
We have founda signicantly negative correlation between the
TCT and the LOS: hemiparetic patients with worse trunk control
at admission stay longer in a rehabilitation ward. However, we
have not found correlation between the initial TCT and FIMgain
during the hospitalization period. This might be explained by the
fact that patients with better TCT scores present higher FIM
scores at admission and, therefore, their gain potential is lower.
This fact would also explain why we have not found a correlation
with efciency measured with the motFIM and the FIM.
As reported in a recent study, Suzuki et al. point out that
balance and muscle strength in the acute stage post-CVA are
predictors of walking speed in stroke patients (18). In our study,
there is also a positive correlation between the TCT score at
admission and the walking speed (comfortable and fast) at
discharge. Walking distance difference is also signicant:
hemiparetic patients whose walking distance at discharge is
>
50 m present higher TCT scores on admission than those
whose walking distance is
<
50 m. In other words, the better
initial trunk control patients have, the longer walking distance
and the faster speed they achieve at hospital discharge.
Fig. 4. Correlation between the compound variabl e
TCT76.4
24.03

admission FIM84.0
22.38

0
:
561 and (a) length of stay (LOS) (r = 0.771,
p
<
0.001) and (b) discharge FIM (r = 0.815, p
<
0.001).
J Rehabil Med 34
Trunk control test in stroke 271
We have not found studies that correlate trunk balance with
static or dynamic balance assessed with computerized postur-
ography. Besides, our study reects a positive correlation
between the TCT score at admission and standing COG
symmetry. Furthermore, patients with less ability to roll towards
the sound side showed more COG displacement at discharge.
Juneja et al. show that the BBS correlates with disability
measured with the FIM in stroke patients (19). We have found a
positive correlation of the BBS at discharge with the TCT at
admission and the motFIM/FIM at either admission or dis-
charge. We have also compared these correlations and have not
found signicant differences between them.
LOS in rehabilitation is conditioned by clinical, functional
and sociodemographic characteristics of stroke patients, though
local resources and local health economic policy should be
considered simultaneously. Different LOS prediction models
have been assessed using available admission information.
Brosseau et al. explain 43.6% of the variation in LOS with a
model consisting of age, functional status at 1 week post-
rehabilitation admission, perceptual status and balance status
(4). The model proposed by Stineman & Williams includes age
and modied Barthel index: rehabilitation LOS prediction
reaches 30% (2). The FIM Function Related Groups system
explains 31.5% of the variance in rehabilitation LOS (3). We
have not found studies including TCT in their prediction models.
In our study admission TCT becomes a signicant predictor
reaching large prediction values: 52% of the variation in LOS
and 60% in discharge FIM (model 1). Adding the admission FIM
to the TCT (model 2), the prediction value increases only to
60%, whereas the discharge FIM prediction is still better: 67%.
However our sample size is very small, so the high predictive
values achieved could be sample dependent. Our results suggest
that TCT should be included in future prediction studies of LOS
and functional stroke outcomes.
Other limitations of the present study should be noted. First,
as usually happens in rehabilitation units, there is an initial bias
due to the fact that our patients are pre-selected on the basis of
their potential to follow an intensive rehabilitation programme.
Information about remaining stroke patients, whose TCT scores
could be different from those of our sample, is not available.
Second, the sample size did not allow us to determine a cut-off
TCT threshold in our predictive model to be used in daily
practice. Initial design of this study considered TCT as an early
predictor and not an outcome variable. It would have been
interesting to repeat TCT at discharge to provide further
evidence of correlation with the other outcome variables.
Further research is needed to correct design drawbacks.
To summarize, we would like to point out that the TCT is a
short and simple test that can be used to predict functional
outcome in stroke patients. Its correlation with the FIM and the
LOS is signicantly positive, though it does not correlate with
FIM gain or treatment efciency. Likewise we can conclude that
the TCT correlates well with some specic motor results such as
walking speed, distance walking perimeter and balance,
measured with computerized systems such as posturography or
with clinical scales such as the BBS. The prediction power of the
TCT in our sample, as a single test, accounts for 52% of the
variation in LOS in rehabilitation and 54% in the FIM at
discharge. The predictive value of a compound variable
(TCT admission FIM) reaches 60% of the variation in LOS
and 66% in the FIM at discharge.
ACKNOWLEDGEMENT
The authors wouldlike to acknowledge the statistical contribution of Mr J.
Vila fromInstitut Municipal dInvestigacio Me`dicade Barcelona(IMIM).
REFERENCES
1. Johnston MV, Wood K, Stason WB, Beatty P. Rehabilitative
placement of poststroke patients: reliability of the critical practice
guideline of the Agency for Health Care Policy and Research. Arch
Phys Med Rehabil 2000; 81: 539548.
2. Stineman MG, Williams SV. Predicting inpatient rehabilitation
length of stay. Arch Phys Med Rehabil 1990; 71: 881887.
3. Stineman MG, Escarce JJ, Tassoni CJ, Goin JE, Granger CV,
Williams SV. Diagnostic coding and medical rehabilitation length of
stay: their relationship. Arch Phys Med Rehabil 1998; 79: 241248.
4. Brosseau L, Philippe P, Potvin L, Boulanger YL. Post-stroke
inpatient rehabilitation, I: Predicting length of stay. Am J Phys Med
Rehabil 1996; 75: 422430.
5. Kwakkel G, Wagenaar RC, Kollen BJ, Lankhorst GJ. Predicting
disability in stroke: a critical review of the literature. Age Ageing
1996; 25: 479489.
6. Nichols DS, Miller L, Colby LA, Pease WS. Sitting balance: its
relation to function in individuals with hemiparesis. Arch Phys Med
Rehabil 1996; 77: 865869.
7. Loewen SC, Anderson BA. Predictors of stroke outcome using
objective measurement scales. Stroke 1990; 21: 8286.
8. Sandin KJ, Smith BS. The measure of balance in sitting in stroke
rehabilitation prognosis. Stroke 1990; 21: 8286.
9. Feigin L, Sharon B, Czaczkes B, Rosin AJ. Sitting equilibrium 2
weeks after a stroke can predict the walking ability after 6 months.
Gerontology 1996; 42: 348353.
10. Colin C, Wade D. Assessing motor impairment after stroke: a pilot
reliability study. J Neurol Neurosurg Psychiatry 1990; 53: 576579.
11. Franchignoni FP, Tesio L, Ricupero C, Martino MT. Trunk Control
Test as an early predictor of stroke rehabilitation outcome. Stroke
1997; 28: 13821385.
12. Berg K, Wood-Dauphinee S, Williams JI. The Balance Scale:
reliability assessment for elderly residents and patients with acute
stroke. Scand J Rehabil Med 1995; 27: 2736.
13. Liston RAL, Brouwer BJ. Reliability and validity of measures
obtained from stroke patients using the Balance Master. Arch Phys
Med Rehabil 1996; 77: 425430.
14. Keith RA, Granger CV, Hamilton BB, Sherwin FS. The Functional
Independenc e Measure: a new tool for rehabilitation. In: Eisenberg
MG, Grzesiak RC, eds. Advances in Clinical Rehabilitation. New
York, NY: Springer; 1987, p. 618.
15. Linacre JM, Heinemann AW, Wright BD, Granger CV, Hamilton
BB. The structure and stability of the Functional Independenc e
Measure. Arch Phys Med Rehabil 1994; 75: 127132.
16. Goldie PA, Matyas TA, Evans OM. Decit and change in gait
velocity during rehabilitation after stroke. Arch Phys Med Rehabil
1996; 77: 10741082.
17. Stevenson TJ, Garland SJ. Standing balance during internally
produced perturbations in subjects with hemiplegia: validation of
the balance scale. Arch Phys Med Rehabil 1996; 77: 656662.
18. Suzuki K, Imada G, Iwaya T, Handa T, Kurogo H. Determinants and
predictors of the maximum walking speed during computer-assisted
gait training in hemiparetic stroke patients. Arch Phys Med Rehabil
1999; 80: 179182.
19. Juneja G, Czyrny JJ, Linn RT. Admission balance and outcomes of
patients admitted for acute inpatient rehabilitation. Am J Phys Med
Rehabil 1998; 77: 388393.
J Rehabil Med 34
272 E. Duarte et al.