Clinical Gait Assessment in the Neurologically Impaired

Reliability and Meaningfulness

MAUREEN K. HOLDEN,
KATHLEEN M. GILL,
MARIE R. MAGLIOZZI,
JOHN NATHAN,
and LINDA PIEHL-BAKER

This study of subjects with multiple sclerosis or hemiparesis assessed the

interrater and test-retest reliability of temporal-distance (TD) gait measures and
examined the relationship of TD valus to functional ambulation ability. Sixty-one
subjects ambulated twice on a 9-m (30-ft) paper walkway and rested 15 minutes
between trials. An ink footprint record and ambulation time were used to calculate
velocity, cadence, step and stride lengths, stride length to lower extremity length
ratio, and step- and stride-time differentials. Subjects were rated on a scale that
assessed the amount of manual assistance required for ambulation. Interrater
and test-retest reliability were high for all TD measures except stride-time
differential for the total sample and within diagnostic categories. The TD values
were highly reliable in all functional categories except one. All TD measures
except stride- and step-time differential displayed a strong linear relationship to
the functional ambulation category. Implications for using TD measures in clinical
settings are discussed.
Key Words: Gait, Hemiplegia, Multiple sclerosis, Physical therapy.

Treating gait disorders is one of the
most common activities of physical
therapists. Most clinical approaches to
assessing treatment outcomes with these
patients, however, remain subjective
and nonstandardized.1-3 Although
standardized qualitative gait assess-
ments exist4. 5 (including the Gait Anal-
ysis Form, Physical Therapy Depart-
ment, Ranchos Los Amigos Hospital,
Downey, CA), they do not have wide-
spread clinical use as outcome indica-
tors. One reason for their infrequent use
may be that many gait abnormalities
assessed by these systems involve dura-
tions of only fractions of a second6. 7 and
thus require a high level of skill and
training to assess the results reliably. The
Ms. Holden is Assistant Professor, Massachusetts
General Hospital Institute of Health Professions
and Clinical Specialist in Neurology, Department
of Physical Therapy, Massachusetts General Hos-
pital, Boston, MA 02114 (USA).
Ms. Gill is Physical Therapy Supervisor, Reha-
bilitation Unit, Massachusetts General Hospital,
Boston, MA 02114.
Ms. Magliozzi, Mr. Nathan, and Ms. Piehl-Baker
are physical therapists at Massachusetts General
Hospital, Boston, MA 02114.
This study was supported by a grant from the
Foundation for Physical Therapy and was presented
at the Annual Conference of the American Physical
Therapy Association, Anaheim, CA, June 19-23,
1982.
This article was submitted February 17, 1983;
was with the authors six weeks for revision; and was
accepted July 1, 1983.
extensive length of time required to per-
form a detailed, standardized clinical
gait analysis is another constraint. Al-
though qualitative assessment of gait de-
viations is essential for planning a treat-
ment program, this method has limita-
tions in assessing treatment outcomes.
Clinically meaningful reductions in
each gait deviation are not quantifiable
with these approaches.4. 5 Comparing
the effect of a specific treatment in pa-
tients with different gait deviations is
also difficult. For example, would a de-
crease in hip circumduction be rated as
superior to a decrease in knee hyperex-
tension?
A multitude of sophisticated quanti-
tative systems to assess gait performance
have been developed,5, 8-10 but most are
too expensive in terms of time, technical
expertise, or equipment requirements to
be available in the average clinic. Fur-
thermore, these systems often provide
more information than is usually
needed to assess treatment outcomes in
most clinical settings.1-3
Temporal-distance (TD) measure-
ment is a clinically feasible, quantitative
approach to gait assessment that has
received recent attention in the litera-
ture.1-3, 6-7 This approach offers many
advantages. The system is inexpensive,
is easy to learn, takes little time to ad-
minister, and has been well studied in
unimpaired individuals.1-3, 7, 8, 11-14 Un-
like traditional qualitative gait assess-
ments, using measurements such as ve-
locity or stride length permits easy quan-
tification of change and comparisons of
outcomes across different subjects or
treatments.
Although many authors describe the
potential usefulness of this clinical
tool,1-3 few reports of TD values in dis-
abled subjects have been published.15-18
One possible reason is the lack of data
on the reliability of the scores, especially
in impaired individuals. Test-retest reli-
ability for selected TD measures has
been reported as high in unimpaired
women (range, r = .69 to r = .97),l as
adequate in unimpaired men,12 and as
high in subjects with hip disorders
(range, r = .96 to r = .99).15 Reliability
of TD measures in the neurologically
disabled has not been described,9. 16. 17
with the exception of one report on test-
retest reliability of timed ambulation (r
= .89) in subjects with mild neurological
deficits.18 The reliability of TD values
could differ in subjects with different
neurological diagnoses or different func-
tional abilities. A clinician might also
ask if TD measures truly reflect the pa-
tient's functional performance. To be
clinically meaningful as an outcome
measure, changes in the values of TD
gait measurements should correlate with

Volume 64 / Number 1, January 1984 35

Fig. 1. Measurement of leg length. Fig. 2. Timed ambulation and ink footprint record.

TABLE 1 individuals. Specifically, the study ad- formed consent forms before participat-
Background Characteristics of Subjects dressed the following questions: ing in the study. We selected 61 subjects
by Diagnosis 1. What is the interrater and test-retest from patients in the physical therapy
reliability of TD measures in subjects department and the neuromedical out-
Multiple Hemi-
with multiple sclerosis or hemipa- patient clinic of Massachusetts General
Characteristic Sclerosis (n paresis (n
=24)(%) = 37)(%) resis? Hospital. Intake criteria included 1) a
2. Does the reliability of TD measures diagnosis of multiple sclerosis (39%) or
Agea change across diagnostic categories or hemiparesis from any cause (61%), 2)
21-40 yr 63 32 functional levels? We hypothesized ability to ambulate at least 9 m (30 ft)
41-60 yr 37 30
that the reliability would be reduced three times using any type of assistive
61-80 yr 0 38
Sex
for patients with multiple sclerosis device (except parallel bars), 3) required
males 33 59 and for patients in the more depend- walking assistance from no more than
females 67 41 ent functional categories. one person, and 4) age greater than 18
Duration of 3. How do TD values correlate with years. Table 1 displays the age, sex, di-
disease function ability? We hypothesized agnosis, and duration of disease for all
1 yr or less 6 81 that the values of all seven TD meas- 61 subjects.
>1-5 yr 46 11 ures would bear a strong linear rela-
>5-10 yr 16 0 tionship to functional ambulation sta-
>10 yr 21 8 Measurements
tus.
a
For multiple sclerosis, = 39 yr, s = 9.8 Temporal distance. We obtained the
yr; for hemiparesis, = 5 1 yr; s = 17 yr. METHOD following TD measures by using an ink
footprint record, recording ambulation
Subjects time, and measuring leg lengths. The
significant changes in functional ambu- Before we began the data collection, protocol (unpublished, M. Holden,
lation status. This investigation was un- an institutional research review com- Temporal Distance Gait Measures and
dertaken to fill some of the gaps in our mittee reviewed and approved the proj- Functional Ambulation Classification
knowledge of TD gait values in impaired ect protocol. All subjects signed in- test protocol, Massachusetts General

36 PHYSICAL THERAPY
Hospital, Boston, MA) was a modifica- TABLE 2
tion of tests of Boenig1 and Tucker and Interrater and Test-Retest Reliabilitya of
Nelson (unpublished, J. Tucker and A. Temporal-Distance Measures by Diagnosis
Nelson, Functional Ambulation Per-
Interrater Test-Retest
formance Test II, Kessler Institute of
Rehabilitation, W. Orange, NJ). TD Measure Total Multiple Hemi Total Multiple Hemi-
1. Velocity: meters/second. Sample Sclerosis -paresis Sample Sclerosis paresis
2. Cadence: number of steps/minute. (n = 24) (n = 6) (n = 18) (n = 61) (n = 24) (n = 37)
3. Step length: Perpendicular distance Velocity 1.00 1.00 1.00 .97 .97 .97
in meters from the heel strike of one Cadence .98 .95 .99 .97 .98 .97
foot to the next heel strike of the Left step length 1.00 .90 1.00 .47 .92 .94
opposite foot. We used mean step Right step length 1.00 1.00 1.00 .96 .95 .96
length for each trial in the analysis. Left stride length 1.00 1.00 1.00 .96 .96 .95
Right stride length 1.00 1.00 1.00 .96 .97 .95
4. Stride length: Perpendicular distance
Left SL:LEL ratio .99 .99 1.00 .95 .95 .94
in meters from the heel strike of one Right SL:LEL ratio 1.00 1.00 1.00 .96 .97 .94
foot to the next heel strike of the same Step-time differential 1.00 1.00 1.00 .97 .94 .97
foot. We used mean stride length for Stride-time differential .35 .00 .35 .68 .92 .45
each trial in the analysis. a
5. Stride length to lower extremity Pearson correlation coefficients.
length ratio (SL:LEL): Stride length
divided by lower extremity length.
6. Step-time differential: Average step on their shoes, which left behind a foot- functional ambulation score. The Pear-
time in seconds wasfirstderived from print record. Ambulation time for 6.1 son r statistic measures the strength of
step-length data. Average step time m (20 ft) was recorded with a digital each linear relationship.20
equaled average step length divided stopwatch* (Fig. 2). The first and last
by velocity. Step-time differential was 1.5 m (5 ft) of the walk were not used
the absolute value of average step time because of changes in velocity that occur RESULTS
for one limb minus average step time when a person starts and stops walking.
Table 2 shows interrater and test-re-
for the opposite limb. Each subject had one practice run (no
test reliability for the total group and for
7. Stride-time differential: Average ink), followed by a five-minute rest and
each diagnostic category. Analysis of the
stride time in seconds wasfirstderived then two trials (with ink), separated by
raw data revealed that the main sources
from stride-length data. Average a 15-minute rest.
of unreliability were measurement of leg
stride time equaled average stride
length and of ambulation time. Despite
length divided by velocity. Stride-time
Data Analysis variability of these measurements, inter-
differential was the absolute value of
rater reliability for the group ranged
average stride time for one limb mi-
We performed two independent TD from .98 to 1.00 for all TD measures
nus average stride time for the oppo-
assessments and recorded the results of except stride-time differential. Test-re-
site limb.
the first 24 subjects to determine inter- test reliability was also high (range, .95-
Functional ambulation category. The
rater reliability. Test-retest reliability .97) for all TD measures except stride-
physical therapist who treated or tested
was performed on all 61 subjects. Seven time differential.
the subjects rated their functional am-
therapists participated in testing reliabil- No significant differences in reliability
bulation ability with a scale developed
ity. Training conssted of reading the of TD measures occurred when subjects
at Massachusetts General Hospital. The
written protocol and performing two were grouped by diagnostic category.
scale assesses the amount of human as-
practice sessions. We used the Pearson Interrater and test-retest reliability were
sistance rather than devices, needed for
correlation coefficient to assess interra- high (range, .90-1.00) for all TD meas-
ambulation (Appendix). This scale
ter and test-retest reliability for the ures (except stride-time differential) in
achieved a kappa interrater reliability of
group as a whole, for each diagnostic both the multiple sclerosis and hemipa-
.72 when tested by nine therapists on
category, and for each functional cate- resis subgroups (Tab. 2).
five patients before this investigation.19
gory. Test-retest reliability of TD measures
The kappa statistic measures how much
To assess clinical meaningfulness of for each functional ambulation category
agreement exists beyond the amount ex-
TD scores, we used a one-way analysis is shown in Table 3. (Because of the
pected by chance alone.
of variance (ANOVA) to determine if reduced sample size, interrater reliabil-
the value of each TD measure was re- ity testing for each functional category
Procedure lated significantly to the functional am- was not performed.) The reliability of
bulation scores. The nature and strength TD measures within Category 1, 2, 4,
Therapists collected TD data using a of the observed relationships were as- and 5 followed the same pattern dis-
combination of detailed written proto- sessed using the r2 and Pearson r statis- played by the group as a whole, that is,
cols previously described.Briefly, leg tics. The r2 statistic describes the pro- all TD measures except stride-time dif-
lengths of standing patients were meas- portion of the variance in TD measures ferential were reliable (range, .91-1.00).
ured as the distance from the superior explained by a linear relationship to Stride-time differential had poor relia-
border of the greater trochanter to the bility in all functional categories except
floor, bisecting the lateral malleolus Category 4. The TD measures of sub-
(Fig. 1). Subjects ambulated 9.2 m (30 * Siliconix Inc, Accusplit Group, 2290A Ridge-
jects in Category 3 were less reliable than
ft) on a paper walkway with ink patches wood Ave, San Jose, CA 95131. those of subjects in all other functional

Volume 64 / Number 1, January 1984 37

TABLE 3 DISCUSSION
Test-Retest Reliabilitya of Temporal-Distance Measures in Subjects
with Neurological Impairments by Functional Category These results confirm that five of the
seven TD measures examined are relia-
Functional Ambulation Categoryb ble and meaningful tools for the meas-
1 2 urement of neurologic treatment out-
3 4 comes. Unlike standardized qualitative
TD Measure Physical Physical 5
Super­ Independent gait assessments that require advanced
Assistance Assistance Independent
vision on Level
Level II Level I (n = 22) skill and training to perform,5, 6 TD val-
(n = 6) (n = 8)
(n = 10) (n = 15) ues can be measured reliably by testers
Velocity .97 .94 .94 .99 .95 who have received only a minimum of
Cadence .92 .94 1.00 1.00 .97 training. Because interrater reliability is
Left step length .95 .98 .53 .94 .93 high, TD values made by different ther-
Right step length .96 .95 .80 .99 .96 apists can be compared with confidence
Left stride length .98 .96 .66 .98 .96 that any observed differences truly re-
Right stride length .97 .96 .70 .99 .96 flect patient performance differences
Left SL:LEL .97 .95 .68 .98 .95 and not intertherapist measurement er-
Right SL:LEL .97 .96 .70 .99 .96
ror. This high interrater reliability in-
Step-time differential .91 .99
.72 .94 .98
creases the utility of TD measures in
Stride-time differential .18 .57 .23 .97 .19 busy clinical settings. Test-retest relia-
a
bility is also high and increases confi-
Pearson correlation coefficients. dence that changes in TD scores over
b
See Appendix for complete definitions. time are the result of real changes in the
TD values and not caused by random
categories (Tab. 3). This group also had ranged from .31 to .47 (except for step-
the smallest n. time differential). The strength of this TABLE 4
Because of the poor reliability dis- linear relationship ranged from .55 to Relationship of Temporal-Distance
played by stride-time differential, the .67 for all TD measures except step-time Measures to Functional Ambulation
relationship of this TD measure to func- differential (Tab. 4). We considered the Category
tional ambulation category was not as- strength of this relationship to be sub- Pearson
sessed. A one-way ANOVA revealed stantial, considering the multitude of TD Measure r2
r
that velocity, cadence, step length, stride factors that could contribute to the var- Velocity .45a .67
length, and SL:LEL were significantly iance in TD measures (eg, age, sex, du- Cadence .38a .62
related to functional ambulation cate- ration of disability, nature of clinical Left step length .39a .63
gory (p < .001, Tab. 4), but that step- symptoms, and type of walking aid or Right step length .31 a .55
time differential was not. The nature orthosis). The linear relationship of ve- Left stride length .47a .65
and strength of the relationship to func- locity, cadence, step and stride length, Right stride length .43a .66
tional status was assessed by the r2 and and SL:LEL to functional ambulation Left SL:LEL .39a .63
Pearson r statistics. Table 4 shows that category can also be seen by examining Right SL:LEL .42a .64
the proportion of total variance in TD the means and standard deviations for Step-time differential .12b .35
measures explained by a linear relation- each TD measure in the different func- a
p < .001, one-way ANOVA.
ship to functional ambulation category tional categories (Tab. 5). b
NS.

TABLE 5
Mean and Standard Deviation of Temporal-Distance Measures
in Neurologically Impaired Subjects by Functional Category
Functional Ambulation Categorya
1 2 4
■

3 5
Physical Physical Supervision Independent Independent
TD Measure Assistance Assistance (n = 6) on Level (n = 22)
Level II Level I (n = 8)
(n = 10) (n == 15)
(s) (s) (s) (s) (s)
Velocity (m/sec) 0.14 (0.08) 0.23 (0.14) 0.24 (0.17) 0.38 (0.23) 0.64 (0.28)
Cadence (stp/min) 34.00 (12) 40.00 (19) 39.00 (20) 48.00 (22) 69.00 (17)
Left step length (m) 0.22 (0.09) 0.32 (0.14) 0.33 (0.08) 0.39 (0.16) 0.48 (0.11)
Right step length (m) 0.23 (0.13) 0.30 (0.13) 0.31 (0-12) 0.40 (0.12) 0.46 (0.15)
Left stride length (m) 0.46 (0.18) 0.62 (0.21) 0.64 (0.18) 0.80 (0.23) 0.95 (0.25)
Right stride length (m) 0.45 (0.17) 0.62 (0.21) 0.64 (0.18) 0.81 (0.23) 0.95 (0.25)
Left SL:LEL 0.51 (0.25) 0.68 (0.22) 0.74 (0.21) 0.83 (0.21) 1.03 (0.28)
Right SL:LEL 0.50 (0.21) 0.68 (0.22) 0.74 (0.21) 0.85 (0.22) 1.03 (0.27)
Step-time differential (sec) 1.09 (1-05) 1.16 (1.82) 0.56 (0.46) 0.61 (0.75) 0.21 (0.37)
Stride-time differential (sec) 0.12 (0.11) 0.06 (0.07) 0.05 (0.04) 0.13 (0.31) 0.02 (0.02)
a
See Appendix for complete definitions.

38 PHYSICAL THERAPY
error in the measurement procedure.
This reliability reduces the need to per-
form consecutive trials to calculate a
more stable measure, further shortens
the testing time, and makes testing more
practical.
The low interrater and test-retest re-
liability of stride-time differential (Tab.
2) was unexpected and difficult to ex-
plain. Thirty-one subjects, however, had
a zero value for stride-time differential,
that is, right and left stride times were
equal. Among those with a stride-time
differential greater than zero, no con-
sistent pattern appeared. Differences in
stride time, right versus left, appeared in
subjects with both diagnoses and in all
functional categories. One notable fact
is that the range of raw scores for this
TD measure was quite restricted (.00-
.42) compared with the range of raw
scores for other TD measures. The re-
stricted range, plus the high incidence
of zero values, could account for a math-
ematical magnification of measurement
error and thus the lower reliability
scores.21 For these reasons, stride-time
differential appears to be a poor choice
as a treatment-outcome measurement.
The data do not support our hypoth-
esis that reliability would be lower in
subjects with multiple sclerosis. Despite
the fact that variability in symptoms is
considered to be a hallmark of the dis-
ease,22 Table 2 clearly shows that both
diagnostic groups displayed comparable
reliability. This reliability occurred de-
spite the smaller sample size in the mul-
tiple sclerosis group, which would tend
to magnify any measurement error.21
Temporal-distance values, therefore, are
reliable measures of treatment outcomes
in patients with either multiple sclerosis
or hemiparesis. Whether TD values
could be reliably measured in patients
with other neurological disorders re-
mains to be tested; our results suggest
that they could be.
We further hypothesized that TD
value reliability would be reduced in
subjects falling in lower functional cat-
egories. As is clear in Table 3, the data
do not support this hypothesis. Appar-
ently, the 15-minute rest period between
trials sufficiently eliminated fatigue that
could have reduced reliability in subjects
at lower functional levels.
Although the TD values for our sub-
jects differed from values of unimpaired
individuals (Tab. 5),8, 11-14 the values dis-
played by our subjects trial to trial were
consistent, despite the degree of neuro-
logical impairment (Tab. 3). This con-
sistency of performance lends support
Volume 64 / Number 1, January 1984
to the idea that locomotion is controlled
by automatic reflex centers whose tem-
poral patterning is relatively fixed.23, 24
One exception to consistent reliability
of TD measures across functional cate-
gories occurred with subjects who re-
quired ambulatory supervision (Appen-
dix, Category 3). This decreased relia-
bility may be a result of the small sample
size,21 or it may represent a real differ-
ence in reliability. Subjects in this cate-
gory required verbal assistance in am-
bulation mainly because of problems
with judgment or attention rather than
physical weakness. Temporal-distance
assessment of gait performance in such
individuals may not be a reliable way to
measure gait performance. More sub-
jects, however, need to be tested. In the
interim, TD values obtained from a
mean of two or more trials would prob-
ably be more representative of true per-
formance in patients requiring verbal
supervision to ambulate.
The significant relationship of veloc-
ity, cadence, step and stride length, and
SL:LEL to functional ambulation status
supports the validity of their use as out-
come measures (Tab. 4). The linear na-
ture and strength of this relationship are
substantial, considering the number of
other variables that could affect TD val-
ues, such as age,11, 13 height,8, 12 types of
brace or walking aid,7, 10 or type of clin-
ical symptoms, and considering the
amount of measurement error inherent
in our crude measure of functional am-
bulation ability. The substantial linear
relationship of TD measures to func-
tional status means that as abnormal
temporal patterning of gait begins to
approximate normal gait, so does func-
tional status. The abnormality of the TD
measures seen at lower functional
levels8, 11-14 supports the idea that ther-
apists should work on the underlying
abnormal motor pattern, especially tim-
ing, to be effective in changing func-
tional status, but does not serve to estab-
lish this view.
We were surprised that step-time dif-
ferential did not relate to functional sta-
tus (Tab. 4), as symmetry of gait is con-
sidered to be one of the most stable
characteristics of normal gait.4, 8, 17 The
lack of a clear relationship of step-time
differential to functional status was evi-
dent even within diagnostic groups. We
expected that hemiplegic subjects at
lower functional levels would have
greater asymmetry of step time than
those who were more independent. Al-
though we did see a trend toward this
distribution, step-time asymmetry was
not consistently related to the degree of
independence. For example, Category 5
hemiplegic subjects had a greater mean
step-time differential (0.4 sec) than did
Category 4 hemiplegic subjects (0.14
sec), and Category 2 hemiplegic subjects
had larger step-time asymmetries than
those in Category 1 (1.44 sec vs 1.19
sec). These data indicate that the sensi-
tivity of the functional rating scale may
have contributed to the poor correlation
of step-time differential to functional
category. The sensitivity of the scale,
however, did not affect the relationship
of the other TD factors to functional
category.
Perhaps step-time differential is a bet-
ter indicator of overall cosmetic appear-
ance or normality of the gait pattern
rather than functional ambulation sta-
tus. A preliminary examination of our
qualitative gait assessments of these pa-
tients revealed that subjects in the lowest
and highest functional categories were
the ones who displayed the greatest
number of gait deviations. Subjects at
higher functional levels may be using
patterns that are more automatic and
require little cognitive attention, but are
abnormal, to improve their independ-
ence in gait. Norton et al have noted
that velocity of gait in hemiplegic sub-
jects is not correlated with degree of
spasticity of the knee muscles (a factor
that may contribute to step-time asym-
metries).25 If the more independent sub-
jects are consistently using abnormal
patterns to achieve their improved func-
tional independence, an interesting
question is raised concerning how to
balance the goals of cosmetic appear-
ance of gait with function of gait in the
process of planning physical therapy
treatments for patients with gait disor-
ders. Further analysis of the qualitative
data we collected as part of this project
should provide some assistance in an-
swering this question.
CONCLUSIONS
1. Velocity, cadence, step length, stride
length, and SL:LEL appear to be ex-
cellent tools for assessing physical
therapy treatment outcomes in sub-
jects with multiple sclerosis or hem-
iparesis because they are highly reli-
able and relate significantly to func-
tional status. Testing of TD measures
in subjects who require verbal assist-
ance to walk because of poor judg-
ment or attention span may be un-
reliable, but further study is needed
to confirm this.
39
2. Stride-time differential is not rec-
ommended as a tool for measuring
physical therapy treatment outcomes
because it displayed low interrater
and test-retest reliability across all
diagnostic categories and all func-
tional levels.
3. Step-time differential may not be a
useful outcome indicator in physical
therapy assessments where func-
tional improvement is the main goal
because it did not relate significantly
to functional status.
Acknowledgements. We wish to ex-
press our appreciation to Alan Jette,
PhD, for his careful review of the man-
uscript and assistance with statistical
analysis; Julian Haynes, PhD, and
Nancy Watts, PhD, for their advice in
initial planning of the study; Robert Lef-
fert, MD, and James Lehrich, MD, for
their cooperation in providing access to
patients; and to Almas Dossa, MS, for
her assistance in conducting the study.
REFERENCES
1. Boenig D: Evaluation of a clinical method of gait
analysis. Phys Ther 57:795-798,1977
2. Shores M: Footprint analysis in gait documen-
tation. Phys Ther 60:1163-1167,1980
3. Robinson JL, Smidt GL: Quantitative gait eval-
uation in the clinic. Phys Ther 61:351-353,
1981
4. Saunders M, Inman V, Eberhert H, et al: The
major determinants in normal and pathophy-
siological gait. J Bone Joint Surg[Am] 35:543-
558,1953
5. Stanic U, Bajd T, Valencic V, et al: Standard-
ization of kinematic gait measurements and
automated pathological gait pattern diagnos-
tics. Gait analysis form: Ljubljana Rehabilitation
Institute, Ljubljana, Yugoslavia. Scand J Reh-
abil Med 9:95-105,1977
6. Gardner GM, Murray MP: Method of measuring
the duration of foot-floor contact during walk-
ing. Phys Ther 55:751-756,1975
7. Smidt GL, Mommens MA: System of reporting
and comparing influence of ambulatory aids on
gait. Phys Ther 60:551-558,1980
8. Murray M, Kory R, Sepic S: Walking patterns
of normal women. Arch Phys Med Rehabil
51:637-650,1970
9. Takebe K, Basmajian J: Gait analysis in stroke
patients to assess treatments of foot drop.
Arch Phys Med Rehabil 52:305-310,1976
10. Ely D, Smidt G: Effect of cane on variables of
gait for patients with hip disorders. Phys Ther
57:507-512, 1977
11. Finley F, Cody K, Finizie R: Locomotive pat-
terns in elderly women. Arch Phys Med Rehabil
50:140-146, 1969
12. Murray M, Drought A, Kory R, et al: Walking
patterns in normal men. J Bone Joint Surg[Am]
46:335-360, 1964
13. Murray M, Kory R, Cladisen B: Walking pat-
terns of healthy old men. J Gerontol 24:169-
178,1969
14. Finley R, Cody K: Locomotive characteristics
of urban pedestrians. Arch Phys Med Rehabil
51:423-426,1970
15. Wadsworth JB, Smidt GL, Johnston RC: Gait
characteristics of subjects with hip disease.
Phys Ther 52:829-837,1972
16. Knutsson E: An analysis of parkinsonian gait.
Brain 95:475-486,1972
17. Kljajik M, Boyd T, Stanic V: Quantitative gait
evaluation of hemiplegic patients using electri-
cal stimulation orthoses. IEEE Trans Biomed
Eng 22:438-444,1975
18. Nelson AJ: Functional ambulation profile. Phys
Ther 54:1059-1065,1974
19. Fleiss J: Statistical Methods for Rates and
Proportions. John Wiley & Sons Inc, New York,
NY, 1973, pp 140-154
20. Nie N, Hull C, Jenkins J, et al: Statistical Pack-
age for the Social Sciences, ed 2. McGraw-Hill
Inc, New York, NY, 1975, pp 260-261
21. Numally J: Psychometric Theory. McGraw-Hill
Inc, New York, NY, 1967, pp 221-226
22. Brown J, Beebe G, Kurtzke J, et al: The design
of clinical studies to assess therapeutic efficacy
in multiple sclerosis. Neurology 29:3-23,1979
23. Grillner S: Interaction between central and pe-
ripheral mechanisms in the control of locomo-
tion. Prog Brain Res 50:227-235,1979
24. Stein P: Motor systems, with specific reference
to the control of locomotion. Annu Rev Neu-
rosci 1:61-81,1978
25. Norton BJ, Bomze HA, Sahrmann SA, et al:
Correlation between gait speed and spasticity
at the knee. Phys Ther 55:355-364,1975

APPENDIX
Functional Ambulation Classification

Category Definition
0 Nonfunctional Ambulation Patient cannot ambulate, ambulates in parallel bars only, or requires supervision or physical assist­
ance from more than one person to ambulate safely outside of parallel bars.

1 Ambulator-Dependent for Patient requires manual contacts of no more than one person during ambulation on level surfaces to
Physical Assistance— prevent falling. Manual contacts are continuous and necessary to support body weight as well as
Level II maintain balance and/or assist coordination.

2 Ambulatory-Dependent for Patient requires manual contact of no more than one person during ambulation on level surfaces to
Physical Assistance— prevent falling. Manual contact consists of continuous or intermittent light touch to assist balance
Level I or coordination.

3 Ambulator-Dependent for Su­ Patient can physically ambulate on level surfaces without manual contact of another person but for
pervision safety requires standby guarding of no more than one person because of poor judgment, question­
able cardiac status, or the need for verbal cuing to complete the task.

4 Ambulator-Independent Level Patient can ambulate independently on level surfaces but requires supervision or physical assistance
Surfaces Only to negotiate any of the following: stairs, inclines, or nonlevel surfaces.

5 Ambulator-Independent Patient can ambulate independently on nonlevel and level surfaces, stairs, and inclines.

40 PHYSICAL THERAPY