© Med Sci Monit, 2006; 12(7): CR273-281 WWW. M ED S CI M ONIT.

COM
PMID: 16810132 Clinical Research

Received: 2005.09.21
Accepted: 2005.11.29
Published: 2006.07.01
Authors’ Contribution:
A Study Design
B Data Collection
C Statistical Analysis
D Data Interpretation
E Manuscript Preparation
F Literature Search
G Funds Collection
CR
Correlation between gait and balance in people with
and without Type 2 diabetes in normal and subdued
light
Jerrold Scott Petrofsky1 ABCDEF, Maria Cuneo1 ABCDEF, Scott Lee2 ABCDEF,
Eric Johnson1 ABCDEF, Everett Lohman1 ABCDEF
1
Departments of Physical Therapy, Loma Linda University, Loma Linda, CA, U.S.A.
2
Department of Internal Medicine, Loma Linda University, Loma Linda, CA, U.S.A.
Source of support: This work was supported in part by a grant from Galaxo Smith Kline 0313473100

Summary
Background: Balance and gait are both impaired in people with diabetes but no study has examined both in the
same subjects in people either with or even without diabetes or related these to room lighting.
Material/Methods: Twelve subjects with type 2 diabetes (D) and 15 age-matched controls (C) were examined under
conditions of full light, eyes closed (no light) and low light (5 candle power). Balance was assessed
during standing by a computerized posturography device. Gait was analyzed during the initiation
of movement, while walking at uniform speed and during turns of 0.66 meters diameter through
accelerometers, foot contact sensors and the electomyogram recorded from the gastrocnemius
and tibialus anterior muscles.
Results: Subjects with diabetes had poorer balance during standing in diminished light compared to full
light and no light conditions. When the room light was dimmed, sway during standing increased
by an average of 25% in D. Control subjects did not have different sway with diminished light com-
pared to the other lighting conditions. Gait was slower, circumduction greater and muscle use great-
er in D than C. There was a significant negative correlation between balance and gait; the worse
the balance, the slower and poorer the gait for both groups of subjects (p<0.05), impaired balance
accounting for 70% of the deviation in gait in D whereas it only accounted for 52% in C.
Conclusions: Balance and gait are related in people with and without diabetes. Diabetes causes balance and gait
to both be impaired compared to C.

key words: diabetes • balance • gait • motor control

Full-text PDF: http://www.medscimonit.com/abstract/index/idArt/452198

Word count: 5272
Tables: 5
Figures: 2
References: 38

Author’s address: Dr. Jerrold Scott Petrofsky, Professor and Director of Research, Department of Physical Therapy, Loma Linda
University, Loma Linda, CA 92350, U.S.A., e-mail: jerry-petrofsky@sahp.llu.edu

Current Contents/Clinical Medicine • SCI Expanded • ISI Alerting System • Index Medicus/MEDLINE • EMBASE/Excerpta Medica • Chemical Abstracts CR273
Clinical Research
Background
Walking and balance are both critical for independence in
activities of daily living [1]. Recent studies show that peo-
ple with diabetes are 15 times more likely to fall during
gait than is seen in the same age matched population of
people without diabetes [2]. Thus in addition to promot-
ing independence, a better understanding of gait and bal-
ance may represent an important aspect of diabetes man-
agement for safety.
The prevalence of diabetes is increasing each year. Today,
nearly 38% of people 65 years and older have diabetes [3].
Complications commonly associated with diabetes include
vascular disease [4] and microvascular complications of the
kidney, nervous system and vision [3]. However, diabetes can
also lead to other disturbances including a loss of balance
[5–7]. The loss of postural stability has been largely attrib-
uted to peripheral sensory impairment [8]. Balance impair-
ments have been shown to precede sensory loss in the foot
[9]. Some have hypothesized that defects such as impair-
ment in vision [10,11] or the vestibular system could be im-
portant contributing factors as well [6,7,12–17].
Gait is also impaired in people with diabetes [5–7,12–16].
Studies in people with both type 1 and type 2 diabetes show
gait to be slower and performed with circumduction dur-
ing the swing phase of gait. Circumduction is an outward
swing of the leg during forward progression of a step. There
is also increased tremor and movement of the joints com-
pared to age matched control subjects in subjects with dia-
betes (13–16,18–21].
While previous studies have shown that both balance and
gait are impaired in people with diabetes, no studies have
examined both in the same individuals to identify their in-
terrelationship. Resnick et al. [22], examined sensory nerve
function, balance and gait speed and found a positive cor-
relation between loss of nerve function and balance and
speed of gait. Resnick and colleagues however described
only gross characteristics of gait during continuous, unin-
terrupted, walking in a linear path. But more falls occur
during turns than walking in a linear direction [23,24].
Earlier studies on gait, suggest that analyzing gait during
turns provides a much more sensitive analysis of the ear-
ly onset of pathologies of gait than during constant speed
walking [5,25]. This is probably due to the inability of the
body to compensate for rapidly changing gravity vectors if
any vestibular or sensory pathology is present [26]. Finally,
in previous studies, we found a greater degree of gait pat-
tern abnormalities during the initial few steps when gait
is started (initiated) in patients with diabetes than was ob-
served during the phase that occurred once gait was start-
ed [13–16].
Therefore, in the present investigation, gait and balance
were examined during 3 conditions, with full room light,
diminished room light, and with the eyes closed. The hy-
pothesis to be tested were 1) Gait (speed and muscle use)
and balance (postural sway) are correlated in both control
and subjects with diabetes. 2) Gait and balance are poor-
er in dark compared to well lit conditions and 3) subjects
with diabetes will show poorer balance and gait in low or
no light conditions than control subjects.
CR274
Med Sci Monit, 2006; 12(7): CR273-281
Material and Methods
Material
Fifteen subjects with no history of diabetes (10 male and 5
female) and twelve subjects with Type 2 diabetes (8 male and
4 female) were examined in this study. The diagnosis criteria
for diabetes was a fasting glucose of greater than 126 mg% as
per the standards of the American Diabetes Association for
two consecutive days [28,29]. The general characteristics of
the subjects are shown in Table 1. In the group of subjects
with diabetes, the mean length of time since the diagnosis
was 5.9±1.5 years and mean HbA1c was 7.8%. The subjects
in the two groups were selected to be age-matched as close-
ly as possible because of the effect of aging on gait and bal-
ance [13]. Subjects first replied to an advertisement to the
University at large or at the Diabetes Treatment Center of
Loma Linda University. They were then screened by a phys-
ical therapist including manual muscle and sensory testing
[14,15]. Sensory testing was accomplished with Semmes-
Weinstein’s monofilaments at 8 locations on the sole of the
foot including the base of each ray, bilaterally mid foot, and
near the calcaneous. Sensory loss was established if any area
required more than 10 grams of pressure for the subject to
perceive sensation [30,31]. The average sensation thresh-
old for the subjects with diabetes was 2.3±1.5 grams where-
as the control subjects had a threshold of 0.59±1.1 grams.
Subjects with diabetes underwent a thorough review of their
records and final evaluation by the director of the Diabetes
Treatment Center. Inclusion criteria for both controls and
subjects with diabetes included that they were free of any
history of falls in the last year or known orthopedic inju-
ries such as impaired joint mobility. Subjects were exclud-
ed if there was any clinical deficit in strength (strength of
all muscle groups in the legs was 5/5), sensation in their
feet or range of motion on testing. Subjects with diabetes
were included if they had diabetes for more than 3 years
and their blood pressure was less than 140/90 mmHg. The
HbA1c was to be less than 9% for at least 9 months. Blood
pressure could be controlled with an ACE inhibitor but not
alpha or beta agonists or blockers. All subjects with diabetes
were managed with oral medication and/or insulin. All ex-
perimental procedures and methods were explained to each
subject, who then signed a statement of Informed Consent
approved by the committee on human experimentation of
Loma Linda University.
Methods
Assessing balance
Balance was assessed using a computerized dynamic posturog-
raphy device that was custom built [14,33]. The frame of the
platform was constructed from two sheets of plywood. The
1-inch plywood sheets were separated and suspended from
one another by four metal bars. Each bar was positioned at
90° with reference to the other bars at the four corners of
the platform. While a subject stood in the center of the plat-
form or leaned in any direction, the resulting weight shift
on the four corners of the platform was transduced through
strain gauges mounted on the metal bars to an electrical out-
put (Figure 1). The output of the strain gages was connected
to four strain gauge amplifiers (Biopac Incorporated, Santa
Barbara, California model ADC 100) and digitized with a six-
Med Sci Monit, 2006; 12(7): CR273-281
Table 1. General characteristics of subjects.
Diabetes
Number Height (cm) Weight (Kg) Age (years)
12 170.2±11.2 91.2±12.5 44.2±14.67
Controls 15 172.4±14.6 89.2±16.3 46.7±14.3
Figure 1. A subject standing on the balance platform.
teen-bit analogue to digital (A/D) converter (sample rate 1000
samples per second) (Biopac Incorporated, Santa Barbara,
California model MP150). The device was tested for validity
and reliability, the results of which are published elsewhere
[14,33]. To calculate balance, differences between the weight
measured on sensors pairs was used. The difference between
the front and back sensors was used for the y coordinate of the
placement of the body weight in the forward back direction
from the center of the platform and differences between the
left and right sensor pairs provided the x coordinate of the
position of the weight of the person on the platform. These
rectangular coordinates were then converted to polar nota-
tion to obtain the magnitude and angle of any movement
away from the center of the platform (assessed by weight
shift) during quiet standing or when reaching. The displace-
ment of weight was one measure used to assess how well sub-
jects could lean. A large displacement showed that the sub-
ject was leaning more away from the center of the platform.
Since body weight of the subjects was different, the displace-
ment weight was normalized as a% of total body weight. If
the displacement was great at the perceived limit of a lean,
then they were able to control balance better.
But when standing or leaning, there is also a variation in
weight which varies cyclically called sway. The variation in the
magnitude of the weight on the platform and the direction
of any movement in the weight of the person on the plat-
form was used to calculate sway. But sway has 2 components.
One is the extent of the movement on the x and y axis of the
platform and the second is the rate or frequency of the sway
(conversely the period of sway equals 1/frequency); in oth-
er words, what we are measuring is how far does someone
move (assessed by the movement of the center of gravity of
the person on the platform) back and forth and how fast
(frequency). Normally there is some sway when standing.
Since each person has a different body weight, a light per-
son with 5 kg of movement of their weight is large but for a
heavy person it may be insignificantly small. Therefore, for
the calculation of sway, the shift in the center of gravity was
Petrofsky JS et al – Correlation between gait and balance in people with…
normalized against each person’s body weight by dividing
the shift by body weight and multiplying by 100.
Accelerometry CR
In previous studies of accelerometry during gait, we have
found that measurement of acceleration at either the knee
or ankle (both flexion-extension and lateral movement)
provided a good index of movement of the legs during gait.
Measuring acceleration at the other joints (hip and shoul-
der) offered no more useful data [13–16]. Therefore, in the
present investigation, only acceleration at the ankle was meas-
ured. Three axis accelerometers weighing 3 grams (Analog
Devices Corporation 202xb) were placed bilaterally on the
ankle (1 cm above the malleolus) on the tibia. The acceler-
ometers measured movement in the side-to-side (lateral), for-
ward back (flexion extension), and vertical directions in the
range of 0–2 gs (1 g equals the force of gravity at the earth
surface). The output of the accelerometers was digitized at
1024 samples per sec and stored by an IBM computer by a
Biopac MP100 16 bit signal digitizer (Biopac Corporation,
Santa Barbara, California). Flexion extension accelerometer
data was used to measure the acceleration during forward
movement of the body. By using the integral of the accelera-
tion data from the forward back accelerometer, forward ve-
locity could be assessed on a step by step basis. The side move-
ment accelerometer data was used to assess lateral movement
of the legs during gait. The up down or Z axis accelerometer
showed impact force as each foot came down during gait.
Timing during gait
Shoe insert transducers measured the timing of foot contact
during gait. Six transducers were placed at various locations
under the foot. Each load cell was composed of two thin lay-
ers of brass (0.4 mm) separated by a layer of conductive rub-
ber (1.0 mm, Zoplex Corporation ZF60). Pressure caused
an electrical resistance change in the conductor, which was
then recorded on a BioPac MP100 recording system [32].
Optical sensors were triggered by the movement of the legs
to mark movement during walking in a straight path and
during the turns for measurement of intervals. Three inter-
vals were analyzed, 1) initiation of gait (first 2 meters), 2)
uninterrupted walking (next 4 meters) and during a turn.
By knowing the timing of crossing points on the floor, ve-
locity of gait could be assessed. Velocity was confirmed from
accelerometer data as described above. By measuring heel
and toe contact times, the length of the stance and swing
phases of gait and the stance/swing time ratios ( ration of
time during the stance phase and the swing phase of gait)
could be calculated. Using heel contact points, the number
of steps in each walking interval could be measured.
Electromyogram
EMG was recorded through three silver silver-chloride adhe-
sive electrodes above the active muscles. One of the electrodes
was placed at the belly of the muscle on the motor point and
the other active electrode was placed 2 cm distal. The final
electrode was the guard and was placed near the same area.
Electrode output was amplified with an EMG100 Biopac am-
plifier (Biopac Corporation, Santa Barbara, California) with
a gain of 2000 and the frequency response was flat from 1
Hz to 500 Hz. The signal was digitized at two thousand sam-
CR275
Clinical Research
ples per second by a 16 bit A/D converter and stored for lat-
er analysis. For each muscle studied, the EMG was normal-
ized by placing the body into a position that isolated each
muscle from contraction of other muscles [13–15]. Three,
3-second maximal efforts were accomplished for each mus-
cle; one minute was allowed between the contractions. EMG
during gait was then divided by the maximum EMG activi-
ty during the strength determination to normalize EMG as
a percent of the maximum muscle activity.
Data analysis
Statistical Analysis involved the calculation of means, stand-
ard deviations, ANOVA, Pearson Correlation Coefficients and
T tests. The level of significance was p< 0.05. In addition to
means and standard deviations of the individual groups of
data, an additional data analysis was accomplished.
During walking in a linear path, as described below under
Results, data from the accelerometers was digitized and an-
alyzed over three periods. One period was from the initi-
ation of gait for a distance of 2 meters. A second analysis
was started at the 2-meter point and continued for anoth-
er 4.5 meters. The first period, then, analyzed the ability of
the body to overcome inertia and start gait, while the sec-
ond period analyzed continuous movement. During turns,
data was analyzed over a 1 second period likewise centered
in the middle of the turn data. For each second of data that
was collected from each accelerometer, the total number of
data samples was 1,024 samples. As a means of analyzing the
variation in acceleration (that is the joint movement) from
each axis on each accelerometer, the mean g forces (one g
equals the force of gravity at the earth’s surface) were calcu-
lated from all digitized data. This was accomplished by first
taking an average of all digital data over each gait analysis.
The mean was then subtracted from each data point to re-
move any offset in the amplifiers so that the mean of all data
was zero. Next the data was full wave rectified by taking the
absolute value of the digital data points. A new mean was
then calculated through the data to provide the average ac-
celeration over the period. Over a given period, the stand-
ard deviation of the acceleration data was also calculated
for each subject. To normalize accelerometer data for each
subject, the coefficient of variation (CV) was then calculated
(standard deviation/mean) to be used as the percent vari-
ation in movement during gait in the flexion extension di-
rection (FE) and the lateral direction (Lat).
Normally, movement is in the forward back direction during
gait. Lateral movement, while seen to some extent, is minimal
in controls. Any sideways movement is called circumduction.
To normalize data to each person’s speed of movement, the
lateral accelerometer data (CV Lat) was divided by the for-
ward data (CV FE) to obtain a circumduction index. If there
was the same movement in the flexion-extension and lateral
directions, then the index would be 1. If most movement was
in the flexion extension direction it would be less than 1.
Procedures
Balance
Balance was assessed in two ways. First, subjects stood mo-
tionless on the balance platform for 30 seconds and postural
CR276
Med Sci Monit, 2006; 12(7): CR273-281
sway was assessed with their arms folded in front of them as
shown in Figure 1. This was repeated in dim light and with
the eyes closed. Next each subject stood in the middle of the
balance platform and remained motionless for ten seconds.
They then extended their trunk in the forward, backward,
left and right directions to the maximum perceived limit of
stability in that direction. The arms were either at the side
or extended fully forward during these tests. Subjects held
this maximum position for 10 seconds and sway and sway
frequency were determined as well as the weight shift from
baseline position on the platform. Subjects repeated all
measurements 3 times with the light full, light diminished
to 5-candle power, and with no light. Sway magnitude and
frequency was measured under all conditions.
Gait
Subjects rested for 10 minutes for acclimation. Accelerometers
were placed on the ankles, sensors inserted the shoes and
EMG electrodes attached to the right and left tibialis an-
terior and the right and left gastrocnemius muscles as de-
scribed under Methods. The subjects were asked to walk
for 6.5 meters then make a turn of 0.66 meters diameter
to the left and to the right side. The protocol was repeat-
ed with eyes closed, diminished light at 5 candlepower and
in a fully lit room. Since there is a history of instability in
gait in patients with diabetes, for all subjects, a gait belt
was worn during walking and a physical therapist walked
behind each person to catch them if they were to fall. In
practice, no one did.
Results
Balance analysis during quiet standing with the arms
down
When subjects stood for 30 seconds with full room light, di-
minished room light and with their eyes closed, there was
greater postural sway in the subjects with diabetes compared
to control subjects. For the control subjects, as shown in
Figure 2, average sway was 1.62% over the 30 second period
with full room light. There was no statistical difference over
the 30 second period (ANOVA) but in no room light, sway
was greater averaging 1.93%. For the subjects with diabe-
tes, average sway was 2.6% with full light. With eyes closed,
sway significantly increased to 2.8% (ANOVA p<0.01). In
dim light, sway was the greatest at 3.4%. Sway increased sig-
nificantly over the 30 second period (p<0.05).
For the control subjects, the greatest sway was when the eyes
were closed; the dim light and full light conditions were not
statistically different from each other (p>0.05). Finally, the
sway did not increase over time. In full light, control sub-
jects leaned slightly backwards and to the right (–34.2±11.3
degrees) with a sway in an elliptical direction at a period of
2.3±1.1 seconds. With their eyes closed, the sway frequen-
cy did not statistically change (2.2±1.4 seconds) but lean av-
eraged to the left forward direction at an angle of 22.3±1.3
degrees. With dim light, sway shifted backwards again at an
average angle of 38.55±15.2 degrees.
For the control subjects, the sway angle was significantly less
under all 3 conditions than seen for the subjects with diabetes
(p<0.01 ANOVA). For the subjects with diabetes, the sway was
Med Sci Monit, 2006; 12(7): CR273-281 Petrofsky JS et al – Correlation between gait and balance in people with…

A 4.0 B 2.5
3.5
3.0
2.5
2.0
CR
1.5

Sway
Sway

2.0
1.5 1.0 full light
full light
1.0 no light
no light 0.5
0.5 dim
dim
0.0 0.0
5 sec 15 sec 25 sec 5 sec 15 sec 25 sec
Time Time

Figure 2. Sway during standing for 30 seconds with full light, no light and diminished light (dim) in subjects with diabetes (A) and control subjects (B).

Table 2. Platform weight shift as a% of the body weight of each
subject in subjects with diabetes (A) and control subjects
(B) when leaning as far as they could without loosing
balance to the front, right, left and back directions with the
arms to the side and arms extended to forward in full light
and diminished lighting conditions. All data is shown as the
group mean ± the standard deviations.
Table 3. Sway during a 4 second period where subjects with diabetes
(A) and control subjects (B) were leaning as far as they
could without loosing balance to the front, right, left and
back directions with the arms to the side and arms extended
to forward in no light, full light and diminished lighting
conditions. All data is shown as the group mean ± the
standard deviations.

A Diabetes Front Back Left Right A Diabetes Front Back Left Right
Arm side 25.6±6.1 16.4±4.7 22.3±11.4 20.5±11.4 Arm side 6.3±1.3 2.1±0.5 3.4±1.2 3.2±2.0
Light Arm Light Arm
26.4±9.1 10.8±3.7 21.2±11.4 20.5±9.2 6.9±2.2 2.4±1.2 3.7±0.4 3.4±1.8
extended extended
Arm side 24.9±8.3 10.2±2.9 16.5±9.6 15.5±8.6 Arm side 7.7±2.9 3.2±1.7 4.9±2.1 4.6±2.2
Dim Arm Dim Arm
25.5±11.9 6.2±1.3 18.3±9.7 14.1±13.6 8.9±2.7 3.0±1.7 4.1±1.1 4.7±0.9
extended extended
Arm side 24.3±6.7 14.2±3.1 18.4±8.7 17.4±8.2 Arm side 7.0±1.5 2.5±0.6 4.2±1.1 4.1±1.8
No LightArm No light Arm
25.9±8.2 10.4±4.7 19.8±9.3 18.6±9.1 8.1±2.6 2.7±1.1 3.7±0.9 4.3±1.4
extended extended
B Controls Front Back Left Right B Controls Front Back Left Right
Arm side 29.4±11.9 21.3±8.1 25.3±8.3 22.6±9.3 Arm side 5.1±2.5 1.8±0.7 2.1±0.9 2.2±1.1
Light Arm Light Arm
31.4±7.9 17.2±9.4 24.9±8.6 23.6±11.9 5.8±2.6 1.9±0.8 2.2±0.4 2.5±1.5
extended extended
Arm side 27.5±12.7 23.5±15.1 23.8±9.8 24.2±8.9 Arm side 5.2±2.5 2.1±1.1 2.1±1.5 2.3±1.7
Dim Arm Dim Arm
29.3±14.2 18.3±7.3 24.9±11.2 25.0±9.9 5.9±2.2 2.3±1.0 2.3±0.4 2.7±1.1
extended extended
Arm side 30.1±7.9 21.4±7.9 23.4±9.3 23.4±9.6 Arm side 5.2±2.1 1.9±0.8 2.1±0.8 2.2±0.9
No No
Light Arm 22.7±8.1 22.7±8.1 24.1±11.2 24.3±8.7 Light Arm 5.7±2.2 2.1±0.8 2.4±0.8 2.6±0.9
extended extended

in a circular motion from 22.6 degrees (leaning in the left for-
ward direction) to 67.9 degrees (motion moving toward the
forward direction). The period of the sway was 3.2 seconds
or a frequency of 0.31 Hertz. In contrast, when the eyes were
closed, the period of the sway remained the same at 3.1 sec-
onds (p>0.05) while the magnitude of the angular displacement
decreased to 14.1±6.3 degrees. As cited above, the magnitude
of the lean away from the center the platform was statistically
higher but similar to the measurements with the eyes open. In
contrast, with dim light, the weight on the platform was shift-
ed backward to a negative angle of 14.2 degrees. There was
still a circular motion at a period of 2.4 seconds but the back-
ward lean caused a large shift in weight backward.
Maximum reach
The results of the experiments, where subjects leaned to the
forward and backward directions and to the sides to their
maximum perceived limits of stability are listed in Tables 2
and 3. When subjects leaned to any of the four directions

CR277
Clinical Research
from a neutral position, the shift in the platform weight
was calculated from the weight data of balance platform. As
shown in Table 2, the weight shift was not significantly differ-
ent in normal, no light or diminished light conditions com-
paring identical conditions in control subjects and subjects
with diabetes for each of the 4 conditions when compared
within a group and under the same condition (ANOVA
p>0.05). The only difference in both groups was that there
was significantly less change in the platform weight when
leaning backwards for both groups of subjects compared to
movement in the other directions (ANOVA p<0.01). Further,
for the subjects with diabetes, there was significantly less shift
in the weight on the platform under all conditions than was
seen for the control subjects. (p<0.05)
However, when looking at the sway at the perceived lim-
its of stability in Table 3, there was a significant difference
between the subjects with diabetes (panel A) and the con-
trol subjects (panel B) under all experimental conditions
(ANOVA, p<0.01). For example, in panel A of Table 3, when
leaning in the forward direction with the arms at the side,
the sway was 6.3±1.3%. Under the same conditions, for the
control subjects, the sway was 5.1±2.5%. Furthermore, sway
was worsened in patients with diabetes when the light was
diminished (ANOVA p<0.01) but was not significantly dif-
ferent for control subjects under the same experimental
conditions (p>0.05).
Gait
The results of the measurements of gait are shown in
Table 4. As can be seen in this table, the stance/swing ra-
tios were constant under any experimental condition ex-
amined here for both populations. However, there was no
statistical difference in the stance/swing ratio comparing
any of the 3 gait sequences (initiation of gait (start), con-
tinuous gait phase (steady) or turns (turn) when compar-
ing data in the same groups of subjects under these differ-
ing conditions even with diminished light (ANOVA p>0.05).
However, there was a greater stance swing ratio in subjects
with diabetes under any condition compared to control sub-
jects (ANOVA p<0.01).
For subjects with diabetes, the velocity of gait was slower for
each condition than that seen for control subjects (p<0.01).
For subjects with diabetes, gait was slowed by about 25% in
the dim light condition. This reduction in speed seen in sub-
jects with diabetes was significant (ANOVA p<0.01). With the
eyes closed, gait was only reduced in speed by 20.2%, these
differences between light, no light and dim light all being
different form each other (p<0.05). Not only was the veloc-
ity of gait less in subjects with diabetes, but the subjects used
more steps (increased cadence) for each gait condition (start,
steady, and turn) when comparing the gait in the subjects
with diabetes to the control subjects (p<0.05). Whereas the
number of steps were not significantly different in the con-
trol subjects when going from full to dim light (p>0.05), there
was significantly greater steps in subjects with diabetes in dim
light compared to gait with full room light (p<0.05).
The movement at the ankles was assessed by accelerome-
ters in the lateral and flexion-extension directions. The CV
showed that there was almost double the movement at the
ankle in subjects with diabetes compared to control subjects
CR278
Med Sci Monit, 2006; 12(7): CR273-281
for each gait condition examined here (p<0.01). Further,
the CV of movement for flexion-extension (FE) and later-
al (Lat) movement increased significantly in subjects with
diabetes in diminished light compared to normal room
light (p<0.01). The no light condition was between the two
and different from low light and full light (p<0.01). Finally,
there was significantly more circumduction in the subjects
with diabetes than seen in the control subjects (p<0.01).
Further, circumduction during gait, in the subjects with di-
abetes increased in diminished light (p<0.05) compared to
full light. By measuring peak acceleration in the z axis, or up
down direction on heel contact, the impact force could be
assessed for the leg. As shown in Table 3, the subjects with
diabetes had greater impact than that seen for control sub-
jects (p<0.01) for all three periods that gait was analyzed.
Impact was significantly increased in the subjects with dia-
betes in diminished light (p<0.05).
EMG data showed that muscle activity was greater in each
of the 3 phases of gait in subjects with diabetes compared
to control subjects (p<0.01). Muscle activity increased for
both muscles examined in diminished light for subjects with
diabetes compared to these same subjects in normal light
(p<0.01). Here again, no light data was better than dim light
data but worse then bright light data. (p<0.05)
Correlation between balance and gait
Table 5 shows the Pearson correlation coefficients for the
subjects with diabetes (panel A) relating the sway for each
of the subjects with diabetes for forward, backward and
side reach (the average of all conditions for each respec-
tive group in Table 3) in full light, diminished light and
no light to the velocity of gait, the accelerometer data, im-
pact force on the foot during gait and EMG from the tibia-
lis anterior and medial gastrocnemius muscles during gait.
Controls are shown in panel B of this figure. The correla-
tion coefficients are shown in this table. Comparisons were
made for full light conditions and diminished light for the
3 phases of gait measured here.
As shown in this table, the correlations for subjects with di-
abetes were significant for all relations examined here. Any
correlation above 0.86 was significant at p<0.001. All other
values were significant at p<0.05. As can be seen here, un-
der both conditions, the r squared values predict that be-
tween 64 and 80 percent of the variation in gait observed
here can be explained by the increased movement during
standing seen in the subjects with diabetes.
For the control subjects there was also significant correla-
tions between sway and gait characteristics. Here gait was
also slower in subjects with greater sway. Gait was with wid-
er steps (more circumduction) and with greater muscle
activity. But the main difference was the lower correlations
between balance and gait characteristics. Here, using the r
squared values raged from 18 to 65% of the variation in gait
observed here can be explained by the increased movement
during standing seen in the subjects with diabetes.
Discussion
Microangiopathy in diabetes can lead to neuropathy, distur-
bances in the central nervous systems including the vestibu-
Med Sci Monit, 2006; 12(7): CR273-281 Petrofsky JS et al – Correlation between gait and balance in people with…

Table 4. Analysis of gait in subjects with diabetes (A) and control subjects (B). No light and low light data as well as full light data are shown for
subjects with diabetes. There was no statistical difference between any light condition in the control subjects so only full light and low light
data are shown.
CR
A Full light Low light No light
Diabetes
Start Steady Turn Start Steady Turn Start Steady Turn
Stance/swing ratio 1.57±0.27 1.64±0.37 1.85±0.39 1.61±0.31 1.62±0.42 1.77±0.36 1.60±0.33 1.63±0.44 1.81±0.33
Velocity (meters/sec) 0.33±0.12 0.76±0.25 0.41±0.13 0.25±0.09 0.61±0.25 0.23±0.41 0.27±0.13 0.66±0.14 0.31±0.08
Steps (#) 1.55±0.32 3.3±0.24 2.3±1.5 1.83±0.41 3.92±0.33 2.55±1.32 1.75±0.32 3.75±0.31 2.45±0.93
CV FE 1.26±0.29 1.94±0.92 2.12±1.32 1.33±0.42 1.81±0.88 2.12±1.11 1.29±0.22 1.88±0.72 2.18±0.8
CV Lat 1.52±0.62 2.23±1.14 2.01±1.21 1.47±0.69 2.38±1.27 2.56±1.52 1.50±0.45 2.31±1.12 2.41±1.14
Circumduction Index 1.5±0.45 2.52±0.98 2.65±1.21 1.92±0.54 2.72±1.43 2.85±1.52 1.81±0.36 2.65±1.12 1.78±1.33
Impact Force(Gs) 0.28±0.09 0.54±0.12 0.38±0.17 0.42±0.08 0.62±0.16 0.51±0.11 0.36±0.11 0.60±0/12 0.43±0.14
EMG tibialis (%) 59.3±16.7 49.6±17.3 62.3±19.7 65.4±19.2 55.7±26.3 67.2±22.4 61±15.8 52.3±11.8 65.3±22.2
EMG gastrocnemius (%) 68.2±17.4 55.4±21.2 66.4±14.6 73.9±22.3 63.2±17.9 71.9±22.6 70.3±16.8 61.2±16.3 68.4±15.6
B Full light Low light No light
Controls
Start Steady Turn Start Steady Turn Start Steady Turn
Stance/swing ratio 1.41±0.36 1.42±0.45 1.54±0.32 1.42±0.39 1.35±0.29 1.39±0.25 1.39±0.38 1.41±0.42 1.44±0.44
Velocity (meters/sec) 0.81±0.22 1.19±0.32 0.81±0.19 0.72±0.24 1.21±0.33 0.79±0.21 0.79±0.23 1.08±0.42 0.82±0.33
Steps (#) 1.21±0.5 2.6±1.2 1.7±0.5 1.10±0.2 2.6±1.2 1.8±0.4 1.22±0.43 2.73±1.3 1.74±0.44
CV FE 0.88±0.21 1.14±0.39 1.21±0.28 0.92±0.23 1.12±0.31 1.19±0.33 0.91±0.24 1.13±0.33 1.21±0.37
CV Lat 0.93±0.24 1.02±0.41 1.19±0.31 0.79±0.29 0.99±0.34 1.23±0.42 0.78±0.31 1.01±0.36 1.19±0.39
Circumduction Index 1.22±0.24 1.1±0.23 0.99±0.23 1.13±0.25 1.07±0.22 1.03±0.43 1.15±0.32 1.10±0.27 1.13±0.41
Impact Force(Gs) 0.22±0.08 0.18±0.04 0.20±0.07 0.27±0.09 0.21±0.09 0.21±0.09 0.29±0.11 0.23±0.11 0.24±0.10
EMG tibialis (%) 31.4±11.8 29.1±16.2 34.7±12.8 29.5±18.2 31.4±6.7 31.2±13.5 28.6±19.1 31.5±6.9 31.3±11.4
EMG gastrocnemius (%) 29.6±8.9 25.5±8.3 31.7±19.2 31.4±11.3 28.2±11.1 32.9±18.2 32.5±10.2 27.3±12.4 32.6±16.1

lar system [5–7] and vision [10]. Visual impairment in peo-
ple with diabetes can lead to poor balance [34,35]. People
with diabetes experience falls ten to fifteen times more fre-
quently than is seen in age matched controls [2,18–20].
Curiously, few studies have examined gait and balance char-
acteristics in patient with diabetes and none have correlat-
ed the relationship between the two.
In the present investigation, subjects with diabetes displayed
significantly more sway than that seen in control subjects
while standing on a balance platform. Thus, the steadiness
of the subjects with diabetes was less than that of control
subjects. Further, subjects with diabetes swayed in an ellip-
tical path favoring one side of the body more than the oth-
er, showing poor symmetry. The cause of these impairments
may be related to either deficits in the visual, vestibular or
somatosensory systems [13], all of which are impaired in
diabetes [13].
Vision, as cited above, provides a significant component of
balance [35]. In previous studies, it has been show that the
impairment in vision in subjects with diabetes includes slow-
ing in saccades such that the eyes do not track well during
gait [10,13]. The impairment is not seen when people with
diabetes walk in a straight path but manifests itself during
turns where there is a constantly changing direction [13].
In the present investigation, the sway of the subjects with di-
abetes was worse when they stood motionless on a balance
platform. When the light was dim, sway became significantly
worse. During darkness, patients with diabetes experienced
significant deficits in gaze-holding, small changes in vesti-
bule-ocular reflex phase velocity and a decrease in optoki-
netic reflex slow phase velocity [27].
In a separate study, this same group showed that these vis-
ual deficits seen in the dark caused a greater reliance on
the vestibular system [10]. This same phenomenon is seen
with sensory loss in the foot, which also causes greater re-
liance on the vestibular system [36]. However, this great-
er reliance may be ineffective since the vestibular system is
also compromised in diabetes [37]. Thus in low light condi-
tions, with impaired peripheral neurological function and
impaired visual pathways, the vestibular system, which may
also be damaged, may receive false cues from both the visual

CR279
Clinical Research Med Sci Monit, 2006; 12(7): CR273-281

Table 5. Correlations between balance and gait in the subjects with diabetes ± the standard deviation under full light, diminished light and low
light conditions (A) and control subjects (B).

A Full light Low light No Light

Start Steady Turn Start Steady Turn Start Steady Turn
Velocity –0.87 –0.89 –0.87 –0.99 –0.99 –0.93 –0.88 –0.89 –0.93
CV FE 0.97 0.87 0.89 0.99 0.99 0.98 0.95 0.91 0.89
CV Lat 0.89 0.85 0.92 0.89 0.99 0.99 0.91 0.90 0.91
Impact 0.83 0.88 0.96 0.99 0.96 0.92 0.85 0.89 0.92
EMG tibialis 0.98 0.96 0.96 0.99 0.99 0.93 0.89 0.90 0.89
EMG gastrocnemius 0.99 0.93 0.89 0.99 0.96 0.98 0.92 0.95 0.93
B Full light Low light No Light
Start Steady Turn Start Steady Turn Start Steady Turn
Velocity 1 –0.43 –0.54 –0.59 –0.78 –0.78 –0.82 –0.73 –0.69 –0.82
CV FE 0.54 0.74 0.74 0.82 0.84 0.82 0.69 0.58 0.67
CV Lat 0.63 0.68 0.81 0.83 0.81 0.67 0.66 0.74 0.59
Impact 0.73 0.79 0.81 0.84 0.76 0.64 0.81 0.72 0.74
EMG tibialis 0.64 0.68 0.78 0.79 0.79 0.66 0.68 0.67 0.82
EMG gastrocnemius 0.72 0.54 0.75 0.78 0.82 0.91 0.71 0.66 0.74

and somatosensory system which, combined with impaired
vestibular function, overloads the system and causes great-
er sway under diminished light conditions.
The same relationship was seen for gait. Prior studies re-
veal that subjects with diabetes walked significantly slow-
er than control subjects, with a wider stance and increased
flexion-extension and lateral movement of the major joints
[13,21,38]. These abnormalities occurred with clinically di-
agnosed sensory loss in the foot (Semmes Weinstein mono-
filament testing) or with a normal monofilament test but
neurological impairment diagnosed through H reflex and
nerve conduction loss [13–16]. Here, as published previ-
ously, gait was slower accompanied by more adverse move-
ment in the legs with greater muscle activity to stabilize the
joints than was found for the control subjects. However, in
diminished light there was a further increase in muscle ac-
tivity and joint movement in the subjects with diabetes, ad-
verse movement increased and gait speed decreased. Using
the same argument above, gait involves multiple body sys-
tems including the visual, vestibular and somatosensory and
motor systems. With all systems damaged in diabetes, false
cues in a dimly lit room could overload the system and cause
more gait abnormalities as was seen for balance in the first
series of experiments. Our preliminary data would suggest
that the greatest degree of balance and gait disturbance oc-
curs in a dimly lit room. Larger prospective matched stud-
ies are warranted given the important clinical implications
on educational awareness and reinforcement on fall and
fall prevention specific for diabetes. Of particular interest,
from a clinician’s perspective, is the fact that gait and bal-
ance are both significantly impaired even with normal sen-
sory monofilament tests. The 10-gram monofilament is used
commonly to assess sensory loss in the foot [31,34]. And yet,
it is common to see reductions in nerve conduction veloc-
ity in sensory nerves and reduced H reflexes with normal
monofilament testing [14–16]. This is disturbing since there
are also significant gait and balance disturbance with nor-
mal monofilament tests. Thus the use of these tests must be
questioned in a clinical setting as predictors of complica-
tions in diabetes. However, clinically, the 10 gram monofil-
ament test has been used and accepted. Since people with-
out diabetes show almost 20 times that sensitivity, it may be
more appropriate to reduce the test to 2 grams to distin-
guish sensory loss in a more sensitive manner.
Two things stand out from these studies. First, the subjects
here had minimal sensory loss and yet had significant gait
and balance impairments. It would be interesting to repeat
the experiments in subjects more typical of the diabetic pop-
ulation with 5–10 times more sensory loss to better assess
their balance impairments. The same is true of body weight.
Here the controls and subjects with diabetes were weight
matched. The normal population of people with diabetes
is much heavier than age matched controls and therefore,
this population would be of interest to examine.
One thing that becomes apparent is that it is more dan-
gerous for people with diabetes to walk in a dimly lit room
than a room with no light. People with diabetes should be
counseled to be careful in very dim lit conditions due to
false sensory cues when walking to avoid injury. Perhaps a
simple test would be to have a person with diabetes stand
with the lights dim or off and assess balance. This may help
show the person their limitations and help show them the
dangers in poorly lit rooms.

CR280
Med Sci Monit, 2006; 12(7): CR273-281
Conclusions
Balance is inversely correlated with gait in people with di-
abetes. Postural sway is worse during standing and gait is
slower and with more unsteadiness in people with diabetes
compared to the control subjects. When balance and gait
were examined in dim light and no light conditions, sway
was not different in control subjects while gait was slower
and knee movement showed an increase in unsteadiness
in dim light or no light conditions. However, for the sub-
jects with diabetes, the greatest sway during standing and
slowest gait was in dim light and not in the no light condi-
tions. There was a significant correlation between balance
and gait in both groups of subjects.
References:
1. Prasansuk S, Siriyananda C, Nakorn AN et al: Balance disorders in the
elderly and the benefit of balance exercise. J Med Assoc Thai, 2004; 87:
1225–33
2. Dingwell JB, Cavanagh PR: Increased variability of continuous over-
ground walking in neuropathic people is only indirectly related to sen-
sory loss. Gait Posture, 2001; 14: 1–10
3. Resnick HE, Stansberry KB, Harris TB et al: Diabetes, peripheral neu-
ropathy, and old age disability. Muscle Nerve, 2002; 25: 43–50
4. American Diabetes Association Diabetes: Vital Statistics, 2001
5. Hiltunen LA: Does glucose tolerance affect elderly persons’ balance,
gait or muscle strength? Cent Eur J Public Health, 2001; 9: 22–25
6. Simoneau GG, Ulbrecht JS, Derr JA et al: Postural instability in patients
with diabetic sensory neuropathy. Diabetes Care, 1994; 17: 1411–21
7. Simoneau GG, Derr JA, Ulbrecht JS et al: Diabetic sensory neuropathy
effect on ankle joint movement perception. Arch Phys Med Rehabil,
1996; 77: 453–60
8. Uccioli L, Giacomini PG, Pasqualetti P et al: Contribution of central
neuropathy to postural instability in IDDM patients with peripheral
neuropathy. Diabetes Care, 1997; 20: 929–34
9. Di Nardo W, Ghirlanda G, Cercone S et al: The use of dynamic pos-
turography to detect neurosensorial disorder in IDDM without clini-
cal neuropathy. J Diabetes Complications, 1999; 13: 79–85
10. Darlington CL, Erasmus J, Nicholson M et al: Comparison of visual-ves-
tibular interaction in insulin-dependent and non-insulin-dependent di-
abetes mellitus. Neuroreport, 2000; 11(3): 487–90
11. Buchanan JJ, Horak FB: Vestibular loss disrupts control fo head and
trunk on a sinusoidally moving platform. J Vestib Res, 2001–2002; 11(6):
371–89
12. Petrofsky JS, Besonis C, Rivera D et al: Does local heating really help
diabetic patients increase circulation? J Orthop Neurol Surg. 2003; 21;
40–46
13. Petrofsky JS, Bweir S, Andal A et al: Joint acceleration in relation to age.
EJAP, 2004; 92; 254–62
14. Petrofsky JS, Lee S, Mcnidor M, Navarro E: Autonomic, Endothelial
Function and the analysis of gait in patients with Type 1 and Type 2
Diabetes. Acta Diabetol, 2005; 42(1): 7–15
15. Petrofsky J, Macnider BS, Navarro BS, Lee S: Motor Control and Gait
Characteristics in Patients with Type 1 and Type 2 Diabetes without
Sensory Impairment in the Foot. Basic and Applied Myology, 2005; 15;
75–86
16. Petrofsky JS, Lee S, Bweir S: Gait characteristics in patients with type 2
diabetes. Europ. J. Appl. Physiol, 2005; 93: 640–47
Petrofsky JS et al – Correlation between gait and balance in people with…
17. Petrofsky JS: A device for the Evaluation of reach balance in sitting and
standing. J Med Eng Tech, In Press
18. Mueller MJ, Minor SD, Sahrmann SA et al: Differences in the gait char-
acteristics of patients with diabetes and peripheral neuropathy com-
pared with age-matched controls. Phys Ther, 1994; 74: 299–308; dis-
cussion 309–13
CR
19. Mueller MJ, Sinacore DR, Hoogstrate S, Daly L: Hip and ankle walking
strategies: effect on peak plantar pressures and implications for neu-
ropathic ulceration. Arch Phys Med Rehabil, 1994; 75: 1196–200
20. Sacco IC, Amadio AC: A study of biomechanical parameters in gait anal-
ysis. Clinical Biomechanics, 2000; 15: 196–202
21. Menz HB, Lord SR, St George R, Fitzpatrick RC: Walking stability and
sensorimotor function in older people with diabetic peripheral neu-
ropathy. Arch Phys Med Rehabil, 2004; 85(2): 245–52
22. Resnick HE, Vinik AI, Schwartz AV: Independent effects of peripher-
al nerve dysfunction on lower-extremity physical function in old age:
the Women’s Health and Aging Study. Diabetes Care, 2000; 23(11):
1642–47
23. Fried AV, Cwikel J, Ring H, Galinsky D: ELGAM – extra-laboratory gait
assessment method: identification of risk factors for falls among the
elderly at home. Int Disabil Stud, 1990; 12: 161–64
24. Gawron W, Pospiech L, Orendorz-Fraczkowska K, Noczynska A: The in-
fluence of metabolic disturbances present in diabetes mellitus Type 1
on vestibulo-spinal reflexes in children and young adults. Otolaryngol
Pol, 2002; 56(4): 451–57
25. Hennekes R, Pillunat L: Asynchronism of saccadic eye movement in
young diabetics as related to HbAIc. Graefes Arch Clin Exp Ophthalmol,
1985; 223(1): 50–52
26. Kennedy RL, Henry J, Chapman AJ: Accidents in patients with insulin-
treated diabetes: increased risk of low-impact falls but not motor vehi-
cle crashes--a prospective register-based study. J Trauma, 2002; 52(4):
660–66
27. Nicholson M, King J, Smith PF, Darlington CL: Vestibulo-ocular, optoki-
netic and postural function in diabetes mellitus. Neuroreport, 2002;
13(1): 153–57
28. American Diabetes Association: Report of the Expert Committee on
the Diagnosis and Classification of Diabetes Mellitus. Diabetes Care,
1997; 20: 1183–97
29. American Diabetes Association: Diagnosis and classification of diabe-
tes mellitus. Diabetes Care. 2004; 27(1): S5–S10
30. Prabhu KG, Patil KM, Srinivasan S: Diabetic feet at risk: a new method
of analysis of walking foot pressure images at different levels of neurop-
athy for early detection of plantar ulcers. Med Biol Eng Comput, 2001;
39(3): 288–93
31. Booth J, Young MJ: Differences in the performance of commercially
available 10-g monofilaments. Diabetes Care, 2000; 23: 984–88
32. Petrofsky JS, Bweir S: Variable output foot sensors to provide pressure
distribution on the foot during gait. Saudi J. Rehab, 2002; 8: 137–42
33. Petrofsky JS, Lohman E: Magentically coupled balance platform. J Med
Eng and Tech, 2004; 28; 211–16
34. Richardson JK: The clinical identification of peripheral neuropathy
among older persons. Arch Phys Med Rehabil, 2002; 83: 1553–58
35. Nichols DS: Balance retraining after stroke using force platform bio-
feedback. Phys Ther, 1997; 77(5): 553–58
36. Horak FB, Hlavacka F: Somatosensory loss increases vestibulospinal sen-
sitivity. J Neurophysiol, 2001; 86: 575–85
37. Perez R, Ziv E, Freeman S et al: Vestibular end-organ impairment in
an animal model of type 2 diabetes mellitus. Laryngoscope, 2001; 111:
110–13
38. Mueller MJ, Salsich GB, Bastian AJ: Differences in the gait characteris-
tics of people with diabetes and transmetatarsal amputation compared
with age-matched controls. Gait Posture, 1998; 7: 200–6
CR281