756

Prediction of Ambulatory Performance Based on Motor Scores

Derived From Standards of the American Spinal Injury
Association
Robert L. Waters, MD, Rodney Adkins, PhD, Joy Yakura, RPT, Daniel Vigil, MD

ABSTRACT. Waters RL, Adkins R, Yakura J, Vigil D. Prediction of ambulatory performance based on
motor scores derived from standards of the American Spinal Injury Association. Arch Phys Med Rehabil
1994;75:756-60.
l Assessment of strength using motor scores derived from the standards of the American Spinal Injury Association
(ASIA) was compared with assessment using motor scores based on biomechanical aspects of walking in the
prediction of ambulatory performance. Measurements of strength, gait performance, and the energy expenditure
were performed in 36 spinal cord injured patients. The ASIA scoring system compared favorably with the
biomechanical scoring system. The ASIA score strongly correlated with the percent increase in the rate of 0,
consumption above normal (p < .0005), O2 cost per meter @ < .0006), peak axial load exerted by the arms on
crutches @ < .OOOl), velocity @ < .OOOl), and cadence @ < .OOOl). Patients with lower extremity ASIA scores
s 20 were limited ambulators with slower average velocities at higher heart rates, greater energy expenditure,
and greater peak axial load exerted on assistive devices than patients with lower extremity ASIA scores 2 30
who were community ambulators. We conclude the ASIA motor score is a simple clinical measure that strongly
correlates with walking ability.
0 1994 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and
Rehabilitation

Determining the potential for ambulation is one of the from a biomechanical perspective, provides near normal sub-
most important evaluations a clinician must make in caring stitution for ankle/foot paralysis with only a minimal in-
for a patient with spinal cord injury (XI). A patient’s level crease in energy expenditure.
of injury, conditioning. motivation, and severity of paralysis Recently the American Spinal Injury Association (ASIA)
all contribute to ambulatory potential.lm5 Previous investiga- motor scoring system” has become accepted as the intema-
tors have measured physiological parameters such as heart tional standard for reporting muscle strength and gauging
rate, respiratory rate, and oxygen consumption in SC1 pa- paralysis after SCI. The ASIA lower extremity muscle score
tients to describe their increased energy expenditure during (LEMS) is derived from manual muscle testing of five lower
walking.‘~5~q extremity muscle groups representing each neurological
In a 1989 study.“’ we found that patients with progres- level from L2 to S 1. However, only two muscle groups are
sively increasing degrees of lower extremity paralysis mea- common to both the AMI and LEMS. This study determines
sured by the Ambulatory Motor Index (AMI) increasingly the correlation of the LEMS to gait performance and energy
substituted for the loss of lower extremity muscle function expenditure and, in this regard, compare its effectiveness
by exerting greater upper extremity exertion on crutches and with the AMI. Although the AM1 has been shown to be an
slowing the gait velocity. The magnitude of upper extremity effective predictor of ambulatory performance, the impor-
exertion was proportional to the increase in physiological tance of using uniform standards is well recognized.” Vali-
energy expenditure in comparison to normal walking. dation of the use of ASIA standards in prognostication of
The AM1 is derived from manual muscle testing of five ambulation status is an important step in promoting the wider
lower extremity muscles about the hip and knee that are use and acceptance of these standards and fostering more
important in the gait cycle from a biomechanical perspec- reliable communication among investigators and clinicians.
tive.“’ Muscle groups about the ankle are not represented in
the AMI because prescription of an appropriate orthosis, SUBJECTS

From the Reg~unal Spinal Cord Injury Care System of Soulhun Culltornia. Rancho
Loa Amigor Mcd~cal Cener. Do%nq. CA. Subjects
Submmed for publxxmn November 73. lY93. Acccptcd in revised fkni Februnry
II. 1994. Thirty-six SCI patients (30 men, 6 women) were studied.
Funded in part by the Natiunnl In4tutr of Disablhty Rehahilitarmn Rcsurch. Field
The average age was 29.0 ? IO. 1 years, height 1.72
Initiated Rexarch gram, Go()XJZSO?8 and H I33G9OI IS
No commercial party having a direct hnancnl intewt in the rewIt\ 01 tho research -C O.lOm. and weight 66.1 5 11.3kg. The average interval
wpponing thk article ha\ or will confer 3 hrnetit upon Ihe author\ or upon any between SC1 and testing was 0.5 ? 0.7 years. Each patient
organirarion with whxh the author\ are aaociated.
Reprint request\ to Roben L. Water\, MD. Clinical Professor of Oflhoprdlc Sur-
had completed a gait training program and was able to walk
gery. UnlverGty of Southern Calitbrmn. Ranch0 Lo\ Amigos Medical Center HB- independently for 5 minutes.
I 17. 7601 E. Imperial Hlghwuy. Downey. CA 90242. Twelve patients were tetraparetic (incomplete cord le-
c 1994 by the American Conpre\\ r,f Rehabilitation Medicine and the American
Academy ol Phyknl Medicine and Rehnhilitation
sions) with sufficient preserved motor strength in the lower
0~X~3~9993/94/7507-00?5$2.oll) limbs to ambulate. Of the remaining 24 patients, four were

Arch Phys Med Rehabil Vol75, July 1994

 AMBULATION AND ASIA MOTOR SCORES, Waters 757

paraplegic (complete cord lesions), and 20 had motor incom- value for normal walking enables determination of gait efti-
plete paraplegia or were paraparetic (incomplete cord Ie- ciency or the O1 rate increase (percent). This is the percent
<ions). Each patient’s trunk extension strength was sufficient increase in the SC1 patient’s rate versus an able-bodied indi-
to allow independent sitting without the use of arms for vidual’s rate of V&/min at comparable speeds.
support. With regard to the 0, rate increase. the relationship be-
The orthotic prescription followed standard clinical prac- tween VO: and speed during normal walking can be defined
tices and was primarily based on the strength of the quadri- according to the equation
ceps. When quadriceps strength was less than 4/S, a knee-
ankle-foot orthosis (KAFO) was prescribed with the knee Vti, = 2.6 + (.129 > V)
locked in extension. Patients wore an ankle-foot orthosis
(AFO) if necessary. Indications for the use of an AFO in- where V@ equals the rate of oxygen consumption per min-
cluded weakness in the ankle plantarflexors or dorsiflexors. ute and V equals the walking speed. To control for differ-
Thirty-four patients used a reciprocating gait pattern, and ences in walking speeds among patients and able-bodied
two patients used a swing-through gait pattern. subjects, the velocity-adjusted value for the normal rate of
V@/min was subtracted from the patient’s rate of V& and
expressed as a percentage of the normal value according to
Manual Muscle Testing
the equation
Manual muscle testing was performed for all lower ex-
tremity muscle groups in the study previously reported in (Vi), Patient -- Vo, Normal)
0, rate increase = 100 X --
which the AMI was found to strongly correlate with gait
VC& Normal
performance and energy expenditure. It was possible to de-
rive the ASIA LEMS score from the original data and corre-
late it to gait performance and energy expenditure.
Peak Axial Load
Manual muscle testing using the standard six-grade scale:
absent = 0; trace = visible or palpable contraction = l/5; Preceding outdoor energy expenditure measurement, in-
poor = active movement through range of motion with grav- door gait analysis was performed along a 6-meter walkway
ity eliminated = 2/S; fair = active movement through range to determine the peak force exerted by the subject’s arms on
of motion against gravity = 3/5: good = active movement adjustable. instrumented upper extremity assistive devices.
through range of motion against gravity and resistance = 4/ Patients were tested with the type of device (cane, crutch,
5: and normal = 51.5. or walker) they normally used. For the swing-through gait
ASIA uses the six-grade grading for tive key muscles pattern, in which both assistive devices make floor contact
representing each neurological segment between L2 and S I simultaneously, the sum of the longitudinal forces exerted
(hip flexion = 1-7. knee extension = L3. ankle dorsiflexion on the right and left devices was calculated and averaged
= 1~4. great toe extension = L5, and ankle plantarflexion over the recorded cycles. For the reciprocating gait pattern.
.- S 1). The maximal LEMS obtained from the bilateral sum the sum of the longitudinal forces exerted on the right and
of muscle grades is SO points. left assistive devices was divided by two to calculate the
In contrast. the AM1 is calculated based on key muscles average maxima1 peak axial load (PAL,) on each assistive
about the hip and knee (hip flexion, hip extension, hip abduc- device during the gait cycle. Both values were expressed as
tion. knee extension, and knee flexion). Grades l/5 and 215 a percentage of the patient’s body weight.
as well as grade 4/S and 5/S are combined into single muscle
grades as follows: 0 = 0. l/S or 215 = I, 3/S = 2. 45 or 51 Data Analysis
5 = 3. The maximum bilateral AMI score is 30 points.
Regression analysis was used to determine the relation-
ships between LEMS and velocity (V) cadence, 0: rate
Energy Measurement increase, O7 cost, and the PAL.
Energy expenditure testing was conducted on a level An examination of the data distributions associated with
outdoor track and has previously been described in de- the regression analyses indicated that the means of various
tail,“.’ I I’ I” The subjects wore their customary shoes and or-
subgroups of the data, partitioned by the degree of neurologi-
(hoses and were instructed to walk until they were fatigued cal impairment, extended in logical patterns along the lines
or for a maximum of 20 minutes. Expired air was analyzed or curves which defined the relationships between both the
for physiological data using a lightweight ( I .Skg) air collec- LEMS and AMI. Furthermore, certain of these means ap-
tion system harnessed to the subject’s shoulders and heart peared to cluster either below an LE,MS score 120 (AMI
rate. respiratory rate. and cadence were monitored and trans- score below 5 12). or above an LEMS score of 30 (AMI
mitted using a portable FM radiotelemetry system.q score 2 18). with an area between a.n LEMS score of 20
Energy expenditure during walking was, expressed by and 30 showing different cluster patterns depending on the
three different parameters. The rate of VOJmin (mL/kg measure (eg. velocity vs O? rate increase). Because of this
< min) indicates the power requirement. The 0: cost per we decided to examine the data according to these apparent
meter (mL/kg i: m) is the amount of oxygen needed to thresholds: less than or equal to a LEMS of 20 points; LEMS
walk a unit distance and indicates physiological work. A of 21 to 29 points; and LEMS greater than or equal to 30
comparison of O! cost per meter of a patient to the average points.

Arch Phys Med Rehabil Vol75, July 1994

 758 AMBULATION AND ASIA MOTOR SCORES, Waters

500
r .
t
$
d- E
Y

Fig l-The relationship between the ASIA LEMS and gait
velocity is defined by the equation, V = 17.2 + LEMS (solid
line). This equation is similar to the data fit previously reported
using the AMI defined by the equation, V = 8.6 + 0.62 x AMI
(dotted line).”
RESULTS
Gait Characteristics
Both gait velocity and cadence were previously found to
vary directly with the AMI and the same relationship was
observed with the LEMS (figs 1 and 2). There was a strong
(r = 0.64. p < .OOOl) linear relationship described by the
equation
V = 17.2 + LEMS
where V is in meters per minute.
Similarly, a strong (I‘ = 0.75. p < .OOOl) linear relation-
ship between the LEMS and cadence was seen. as follows:
C = 35.2 + (1.47 x LEMS)
where cadence is in steps per minute.
The above equations are graphed in figs 1 and 2 along
with equations previously derived using the AMI.
Energy Expenditure
To control for V. the O? rate increase was determined
according to the methods discussed previously. There was
ASIA-LEMS
Fig 3-The relationship between the ASIA LEMS and 0, rate
increase is defined by the equation, O2 rate increase = 207
~ 4.4 x LEMS (solid line). This equation is similar to the data
fit previously reported using the AM1 defined by the equation
O2 rate increase = 257.5 - 2.82 x AM1 (dotted line).”
a significant (r = .56,/> < .OOOl ) linear relationship between
the LEMS and O2 rate increase characterized by the equation
0: rate increase = 207 ~ (4.4 x LEMS)
where 0, rate increase is the percent increase in the rate of
V@/min in comparison to the value for an able-bodied sub-
ject walking at the same speed (fig 3).
The 0: cost per meter was also significantly (r = .5S, p
< .0006) related to the LEMS (fig 4). This relationship was
best defined by the equation
0, cost = 0.99 - (0.02 x LEMS)
The previous equations regarding energy expenditure are
graphed in figures 3 and 4 along with equations previously
derived from the AMI.
Peak Axial Load
There was a strong (1. = .80, p < .OOOl) relationship
between the PAL and the AMI, which was defined as fol-
lows:
PAL = (7.27 - (0.14 x LEMS)]’

5 15 45 5b

Fig 2-The relationship between the ASIA LEMS and cadence
is defined by the equation, C = 35.2 + 1.47 x LEMS (solid
line). This equation is similar to the data fit previously reported
using the AM1 defined by the equation, C = 36.2 + 0.70
x AM1 (dotted line).”
AS,:-LEM:
Fig 4-The relationship between the ASIA LEMS and 0, cost
is defined by the equation, Oz cost = 0.99 ~ 0.02 x LEMS
(solid line). This equation is similar to the data fit previously
reported using the AM1 defined by the equation, 0, cost = 1.39
- (0.027 X AMI) + (0.00015 X AMI’) (dotted line).”

Arch Phys Med Rehabil Vol75, July 1994

 AMBULATION AND ASIA MOTOR SCORES, Waters 759

Table 1: Correlation Coefficient (r) LEMS Versus AMI

_____~__
I ,EMS r .4’M1*r
--____._~-- ___
VdWllj 0.6-l 0.75
Cadcncr 0.: 0.73
% 0, rate II~CIP;I~C 0.5(1 O.hX
0, Coat 0.55 0.77
Peak il\lill load 0.x0 0.73
-. _~_______
.: W;,,& 1

value of the correlation coefticient, I’. tabulated in table I.

Fig 5-The relationship between the LEMS and the PAL is
defined by the equation. PAL = 7.27 - 0.14 x LEMS’ (solid
line). This equation is similar to the data fit previously reported
using the AMI defined by the equation, PAL = 82.75 - (1.72
* AMI) i (0.009 “: AMI? (dotted line).”
Slightly higher values were obtained fl.>rthe LEMS than fol
previously reported values using the AMI for the equations
derived for the PAL (0.80 versus 0.73 I .md cadence (0.75
versus 0.73). Although the correlation cotbfticient was lower
for the LEMS than for the AMI regarding the equations
describing velocity (0.64 vs 0.75). %mO1rate increase (0.56
vs 0.68) and 0: cost (0.55 vs 0.77).

Mhere PAL is expressed as a percent of total body weight.
This equation, based on the LEMS. and the equation previously
reported derived from the LEMS. are depicted in figure 5.
LEMS Versus ,4MI
A comparison of the LEMS to the AMI showed an ex-
tremely robust (I’ = 0.93.11 < .OOOl) linear relationship (fig
6). This relationship is described by the equation
Ambulatory Motor Index Thresholds
Patients were divided in three group\ according to the
LEMS: group I. 520 points: group 2. 21-29 points; and
group 3. 230 points. There were highly statistically signiii-
cant differences in the parameters of gait performance and
energy expenditure between groups. The\e data are summa-
rized in table 2. All patients in group -3 achieved community
ambulation status. All patient\ in group I were household
ambulators.

LEMS = (0.96 x AMI) - 0.90 DISCUSSION

The ASIA L,EMS closely correlates with the parameters of
Hecauae ot‘ the extremely close relationship between the
gait performance (velocity and cadence 1. energy expenditure
LEMS and AMI. it is not surprising that the equations relat-
(rate of O2 uptake. heart rate. 0, cost). and the peak axial
ing muscle paralysis to gait velocity, cadence, rate of energq
load exerted by the arms on crutches. The quality of data fit
expenditure. oxygen cost. and peak axial load on crutches
using the LEMS as determined by the calculation of the
using either the LEMS or AMI as the dependent variable
correlation coefticient, j’, was approximately the same as
are quite similar. The equations based on the LEMS and the
using the AMI.
AMI are graphed in tigures I-S. A comparison of the quality
Although only two muscle groups arz <ommon to both
of data tit using the LEMS and AMI can be judged by the
the LEMS and AMI (hip flexion and knt*e extension) the
reason for the excellent correlation u\inp either the LEMS
and AMI as the dependent variable in relation to the parame-
ters of gait performance and energy expenditure is that both
systems represent similar neurological segnlents of the spinal
cord. The muscle groups unique to the .\Ml are the hip
extensors (S I. S7). hip abductors (S I, S:! 1. and hnee tlexorx

Table 2: ASIA Lower Extremity Motor Score

(Mean in 1 SD)

Fig h--The was a very strong linear relationship (r = 0.93. p

‘c .OOOl) between the LEMS and AMI. This relationship is
defined by the equation LEMS = (0.96 x AMI) ~~ 0.90 (solid
line).

Arch Phys Med Rehabil Vol 75, July 1994

760 AMBULATION AND ASIA MOTOR SCORES, Waters

(L4. L5, Sl). The ankle dorsiflexors (L4), great toe extensors capability for determining ambulatory potential in SC1 pa-
(L5), and ankle plantarflexors (S 1) are unique to the LEMS. tients.
Nevertheless, because the muscle groups exclusive to each References
motor scoring system are innervated by approximately the I. Chantraine A, Crielaard JM. Pimay F. Energy expenditure of ambula-
same neurological segments, there is an extremely strong tion in paraplegics: effects of long term bracing. Paraplegia
1984;22:173-81.
linear relationship (r = 0.93) between the two scoring sys-
2. Hussey RW. Stauffer ES. Spinal cord injury: requirements for ambula-
tems despite differences in their derivation. tion. Arch Phys Med Rehabil 1973;54:544-7.
There were differences in the physiological and gait per- 3. Lemer-Frankiel M. Vargas S, Brown M. Krusell L. Schoneberger W.
formance parameters between subgroups partitioned ac- Functional community ambulation: what are your criteria? Clin Manag
Phys Ther 1986; 6: 12-S.
cording to the LEMS as previously noted for the AMI.‘O As
4. Merkel KD. Miller NE, Merritt J. Energy expenditure in patients with
shown by Hussey and Stauffer,’ walking ability is directly low-. mid-, or high-thoracic paraplegia using Scott-Craig knee-ankle-
related to motor power. At least “fair” hip flexor strength foot otthoses. Mayo Clin Proc 1985;60:165-8.
and at least “fair” knee extensor strength unilaterally was Waters RL. Lunsford BR. Energy cost of paraplegic locomotion. J
Bone Joint Surg [Am] 1985:67: 1245-50.
required to enable a patient to achieve a reciprocal gait pat-
Clinkingbeard EE. Gersten JW. Hoehn D. Energy cost of ambulation
tern. Using the LEMS, patients who met these criteria had in the traumatic paraplegic. Am J Phys Med 1964:43: 157-65.
a LEMS 2 30, group 3. Their mean walking velocity was Gordon EE. Vanderwalde H. Energy requirements in paraplegic ambu-
57Sm/min, mean heart rate was 108bpm, average rate of lation. Arch Phys Med Rehabil 1956;37:276-85.
Huang C-T, Kuhlemeier KV. Moore NB. Fine PR. Energy cost of
V@ was 14.6mL/kg min, and the average peak axial load
ambulation in paraplegic patients using Craig-Scott braces. Arch Phys
exerted on assistive devices was 8.1% body weight. These Med Rehabil 1979;60:595-600.
values regarding gait performance and energy expenditure 9. Mcbeath AA, Bahrke M. Balke B. Efficiency of assisted ambulation
are reasonably close to comparable values for able-bodied determined by oxygen consumption measurement. J Bone Joint Surg
[Am] 1974:56:994-1000.
control subjects examined previously.lh The values for able- IO. Waters RL. Yakura JS, Adkins R. Barnes G. Determinants of gait
bodied control subjects were 80m/min, 99.6bpm. and performance following spinal cord injury. Arch Phys Med Rehabil
12.1 mL/kg/min, respectively. This similarity accounts for 1989:70:8 I l-8.
the fact that all group 3 patients were community ambulators 11. American Spinal Injury Association. Standards for neurological classi-
fication of spinal injury patients. Chicago: ASIA. 1992.
and did not rely on wheeling as a primary mode of transpor- 12. Stover S. Classifications standards for SCI, revised 1992 [editorial].
tation. In contrast, most patients in group 1 with a LEMS Arch Phys Med Rehabil 1992;79:783.
5 20 required two KAFOs and two crutches to ambulate. 13. Waters RL, Barnes G. Husserl T, Silver L. Liss R. Comparable energy
Their mean V was 30Sm/min, mean heart rate was 130bpm, expenditure after arthrodesis of the hip and ankle. J Bone Joint Surg
[Am] 1988;70:1032-7.
average rate of Voz was 15.2mL/kg/min and the average 14. Waters RL, Campbell J. Perry J. Energy cost of three-point crutch
peak axial load exerted on assistive devices was 36.4%. ambulation in fracture patients. J Orthop Trauma 1987: 1:170.3.
Consequently it is not surprising that wheeling was the pri- 15. Waters RL, Hislop HJ. Perry J. Antonelli D. Energetics: application to
the study and management of locomotor disabilities: energy cost of
mary means of mobility outside the home for these patients.
normal and pathologic gait. Orthop Clin North Am 1978:9:35 l-77.
We conclude that calculation of the ASIA LEMS provides 16. Waters RL. Lunsford BR, Perry J. Byrd R. Energy-speed relationships
the clinician with a simple method of estimating ambulatory of walking: standard tables. J Orthop Res 1988;6:215-22.

Arch Phys Med Rehabil Vol75, July 1994