FEATURE

Validity of the Original and Short Versions of the Dynamic

Gait Index in Predicting Falls in Stroke Survivors
SeungHeon An1, PT, PhD, YoungJu Jee2, RN, PhD, HyeonHui Shin3, OT, MPH &
GyuChang Lee4, PT, PhD
1 Department of Physical Therapy, National Rehabilitation Center, Seoul, Republic of Korea
2 Department of Nursing, Kyungnam University, Changwon-si, Korea
3 Department of Rehabilitation Science, Graduate School of Inje University, Gimhae-si, Korea
4 Department of Physical Therapy, Kyungnam University, Changwon-si, Korea

Keywords
Falls; stroke; injury prevention.
Correspondence
GyuChang Lee, Department of Physical
Therapy, Kyungnam University, 7
Kyungnamdaehak-ro, Masanhappo-gu,
Changwon-si, Gyeongsangnam-do 631-701,
Korea.
E-mail: leegc76@kyungnam.ac.kr
Accepted March 8, 2016.
doi: 10.1002/rnj.280
Abstract
Purpose: This study aimed to investigate the validity of the original version
and short version of the Dynamic Gait Index (DGI-8 and DGI-4) in predicting
falls in stroke survivors.
Design: This is a retrospective, cross-sectional study.
Method: This study collected data for 57 chronic stroke survivors, and evalu-
ated the validity of the DGI-8 and DGI-4. To test functional ability, the Sit-to-
Stand Test, gait subscale of the Performance-Oriented Mobility Assessment, the
10-meter Walk Test, the Fugl-Meyer assessment, and the Trunk Impairment
Scale were used.
Findings: For the DGI-8, the cut-off value for the prediction of a fall was
shown to be 16.5, with an area under the curve (AUC) of 0.78. The cut-off
value of the DGI-4 was shown to be 9.5, with an AUC of 0.77.
Conclusions: The study results show that the DGI-8 and DGI-4 have discrimi-
nation in the prediction of fall in stroke survivors.
Clinical Relevance: DGI-8 and DGI-4 can be useful for predicting falls of
stroke patients, allowing better quality of care.

walking over uneven surfaces, turning, going up and

Introduction
Patients with neurological problems such as stroke find it
difficult to maintain balance during activities and loco-
motion. Impairment of balance may result in persistent
disabilities such as a decrease in mobility, restriction of
physical activities, inability to live independently, and
restrictions on participating in social activities (Jonsdottir
& Cattaneo, 2007). Ultimately, the most important goal
for stroke survivors is the restoration of balance and
locomotor function.
An accurate evaluation of balance is a highly important
component in achieving this goal. Reliable and valid
examination tools in a clinical setting are fundamental for
planning care and evaluating outcomes. Most stroke sur-
vivors experience falls during ambulatory activities (e.g.,
down stairs, navigating past obstacles; Cho, Yu, & Rhee,
2015; Hollands, Agnihotri, & Tyson, 2014; Weerdesteyn,
de Niet, van Duijnhoven, & Geurts, 2008). Balance defi-
cits of stroke survivors also can cause reduced postural
stability during quiet standing, and delayed and less coor-
dinated responses to both self-induced and external per-
turbation balance. Gait deficits include reduced
propulsion at push-off, decreased lower extremity flexion
during the swing phase, and reduced stability during the
stance phase (Weerdesteyn et al., 2008).
Therefore, an evaluation of dynamic balance and
gait ability is particularly important (Cakar, Durmus,
Tekin, Dincer, & Kiralp, 2010). Although the balance
performance of stroke survivors has been studied with
respect to quiet and dynamic standing posture

© 2016 Association of Rehabilitation Nurses

Rehabilitation Nursing 2016, 0, 1–10 1
Validity of the 4- and 8-Item Dynamic Gait Index in Stroke Survivors
responses following diverse altered sensory inputs
(Lim, Jung, Kim, & Paik, 2012), none of the devel-
oped tools can be used for evaluating dynamic balance
during locomotion.
The Dynamic Gait Index (DGI) was developed to
evaluate gait dysfunctions related to the risk of falls and
the ability to maintain stability during ambulation in
the elderly (Shumway-Cook & Woollacott, 1995). The
eight items of the DGI are: walking on a level surface,
walking with speed changes, horizontal head turns, ver-
tical head turns, walking with pivot turns, stepping over
an obstacle, stepping around an obstacle, and climbing
steps. Most stroke survivors have deficiencies in sensory
and neuromotor functions with a substantially lower
ability to control momentum during activities (Ng
et al., 2015). The DGI has been shown to have high
reliability in evaluating elderly people (Shumway-Cook,
Gruber, Baldwin, & Liao, 1997), those with vestibular
dysfunction (Wrisley, Walker, Echternach, & Strasnick,
2003), those with multiple sclerosis (Cattaneo, Regola,
& Meotti, 2006), and stroke survivors (intraclass corre-
lation coefficient [ICC] = .96–.98). Studies on the pre-
diction of falls showed that the cut-off points for
elderly people (Shumway-Cook et al., 1997), those with
multiple sclerosis (Cattaneo et al., 2006), and those with
Parkinson’s disease (Landers et al., 2008) ranged from
12 to 19 points. For stroke survivors, it was reported
that the DGI shows significant correlations with the
Berg Balance Scale (r = .83), Activities-specific Balance
Confidence (r = .68) (Jonsdottir & Cattaneo 2007), and
10-meter Walk Test (10 mWT) (r = .68–.83) (Lin, Hsu,
Hsu, Wu, & Hsieh, 2010). Thus, it has been shown that
the DGI is a useful tool in the prediction and discrimi-
nation of falls in stroke survivors with balance deficits
and gait dysfunction.
A four-item version of the DGI (DGI-4; short version)
designed by Marchetti and Whitney (2006) was devel-
oped by modifying the eight-item version of the DGI
(DGI-8; original version). The DGI-4 is an examination
of psychometric measures similar to the original version
and shows reliability in the prediction of falls and the
evaluation of the level of gait dysfunction in patients
with balance deficits and vestibular dysfunction. How-
ever, no study has been conducted on the accuracy of
the DGI-8 in predicting falls in stroke survivors. Further,
the revised DGI-4 has not been investigated to determine
whether it has retained the characteristics of the original
version and reflects the functional characteristics of
stroke survivors.
2 Rehabilitation Nursing 2016, 0, 1–10
S.-H. An et al.
Therefore, this study aimed to investigate the validity
of the DGI-8 and DGI-4 in predicting falls and reflecting
functional abilities of stroke survivors.
Methods
Participants
In this study, stroke survivors were recruited from an
inpatient rehabilitation center through advertisements. A
total of 78 stroke survivors were recruited, and they were
screened against the following inclusion criteria: the abil-
ity to walk more than 10 m without a walking aid; the
absence of neurological or musculoskeletal problems in
the lower extremities; and a score of >24 points on the
Mini-mental state examination. Stroke survivors with a
pacemaker or uncontrolled diabetes were excluded from
the study. Out of 78 recruited stroke survivors, seven
stroke survivors were excluded based on the criteria;
eight dropped out because of an inability to attend the
evaluation, discharge from the rehabilitation center, or
poor health; and six were excluded after they were found
to have incomplete data. Therefore, data were collected
from 57 stroke survivors. The purpose and procedure of
the study was explained to all participants, and they
signed informed consent forms. The study was approved
by the Kyungnam University Institutional Review Board.
Procedures
This study was a retrospective cross sectional study. The
general and medical characteristics (gender, age, etiology,
affected side, disease duration, and use of a walking aid)
of the subjects were collected from medical charts or brief
interviews. The number of falls within the last 6 months
was recorded from interviews with the patients or care-
givers. The number of falls of the stroke survivors who
had a stroke incident less than 6 months ago was col-
lected from the time of stroke to the present. Falls were
defined as incidents in which the subject unintentionally
came to rest on a level below knee height (Lamb, Jorstad-
Stein, Hauer, & Becker, 2005).
After investigating the general and medical characteris-
tics and fall experiences, the subjects were examined using
the DGI-8 and DGI-4. In addition, to examine functional
ability, examination tools including the Sit-to-Stand Test
(STS), the Performance-Oriented Mobility Assessment
(POMA-Gait subscale), the 10-meter Walk Test
(10mWT), the Fugl Meyer-Lower Extremity (FM-LE),
© 2016 Association of Rehabilitation Nurses
S.-H. An et al.
and the Trunk Impairment Scale (TIS) were applied. The
STS was used to evaluate muscle strength in the lower
extremity of the affected side, the POMA-Gait subscale
for examining gait ability, the 10mWT for examining gait
speed, the FM-LE assessment for examining motor recov-
ery in the lower extremity of the affected side, and the
TIS for examining trunk control.
General and medical characteristics and the number of
falls of stroke survivors were collected by research assis-
tants on the first day. All examinations including DGI,
STS, POMA-Gait subscale, 10mWT, FM-LE, and TIS
were performed over a period of 2 days by two physical
therapists who had more than 15 years of experience
caring for neurological patients. The assessors had read
the protocols and understood the examination tools
sufficiently. The participants were allowed to rest for 2–
5 minutes after each examination followed by a verbal or
physical demon-
stration to help the participants understand the next
examination.
Outcome Measurements
Dynamic Gait Index
The DGI is composed of eight items: (1) walking on a
level surface, (2) walking with speed changes, (3) hori-
zontal head turns, (4) vertical head turns, (5) walking
with pivot turns, (6) stepping over an obstacle, (7) step-
ping around an obstacle, and (8) climbing steps. The
individual items are scored on an ordinal 4 point scale,
from 0 (severe dysfunction) to 3 (intact function); a max-
imum of 24 points can be obtained for the DGI-8. The
inter-rater reliability of the DGI for stroke survivors is
ICC = .96 (Jonsdottir et al., 2007). The DGI-4 is a short
version of the DGI-8 and consists of items 1–4 with a
total score of 12 (Marchetti & Whitney, 2006). The test-
retest reliability of DGI-4 for stroke survivors is
ICC = .92; the convergent validity with 10mWT was
reported to be r = .68–.73, whereas the convergent valid-
ity with the Postural Assessment Scale for Stroke was
reported as r = .85–.83 (Lin et al., 2010).
Gait Subscale in Performance-Oriented Mobility Assessment
(POMA-Gait Subscale)
The gait subscale in POMA was developed to evaluate the
mobility and fall risk in the elderly (Tinetti, 1986). This
examination tool consists of balance items (9 items, 16
points) and gait items (8 items, 12 points), for 28 points
in total. In this study, only the gait items were used. The
© 2016 Association of Rehabilitation Nurses
Rehabilitation Nursing 2016, 0, 1–10
Validity of the 4- and 8-Item Dynamic Gait Index in Stroke Survivors
eight gait items are as follows: gait initiation, stepping
length from the left and right side, foot clearance, sym-
metrical gait, step continuity, path deviation, postural
stability, and gait stance. The items are scored on a
2–3-point scale, with 12 total scores. In stroke patients,
the inter-rater reliability of the POMA-Gait subscale was
reported as ICC = .85–.94 (Sterke, Huisman, van Beeck,
Looman, & Van der Cammen, 2010).
Sit-to-Stand Test
The STS test evaluates muscle strength in the lower
extremities by assessing the time taken to repeat the sit-
to-stand task five times. In this study, the participants
sat on a chair with a back and without armrests and
then crossed their arms on their chests and sat and
stood five times without the aid of the upper extremities.
The measurement started the moment the participant’s
back left the chair’s back and stopped when the partici-
pant’s back touched the chair’s back again (Mong, Teo,
& Ng, 2010). The criteria for standing up was achieved
when both knee and hip joints were fully extended with
the trunk erect. The intra-rater/inter-rater reliability of
this test was reported to be ICC = .99–.97 (Mong et al.,
2010).
10-meter Walk Test
The 10mWT was used to assess gait ability and is mea-
sured as the time taken to walk a 10-m distance following
subtraction of a 2-m acceleration zone and 2-m decelera-
tion zone for a total 14-m distance. In this study, we
marked a zero point (beginning line) and an endpoint on
the flat ground and had the participants walk at a com-
fortable walking pace when an assessor provided the start
signal. As soon as the participants arrived at the 10 m
endpoint, the stopwatch recorded the time. The test-retest
reliability of the 10mWT has been reported to be high
(ICC = .88–.97) (Flansbjer, Holmback, Downham, Patten,
& Lexell, 2005).
Fugl Meyer-Lower Extremity (FM-LE) Assessment
The FM-LE test was used to examine the motor func-
tion of the affected side (Fugl-Meyer, Jaasko, Leyman,
Olsson, & Steglind, 1975). It is a tool used to quantita-
tively examine the degree of motor recovery in stroke
patients. In particular, the FM-LE test for the lower
extremity is subdivided into hip, knee, ankle, and coor-
dination, with a total score of 34. The maximum score
for each item is 3, with 0 = “not able to perform,”
1 = “partial performance,” and 2 = “complete
3
Validity of the 4- and 8-Item Dynamic Gait Index in Stroke Survivors
performance.” The reported inter-rater reliability of the
FM-LE was ICC = .96 (Sanford, Moreland, Swanson,
Stratford, & Gowland, 1993).
Trunk Impairment Scale
The TIS was used to examine trunk control and consists
of three sections totaling 23 points. The items for static
sitting balance examined the ability to maintain posture
with the nonaffected side crossed on the affected limbs
and both feet touching the ground (7 points). The items
for dynamic sitting balance examine the separate move-
ment of both the upper trunk and lower trunk through
trunk lateral flexion (10 points). The items for coordi-
nation examined the rotating movement in the horizon-
tal plane of both the upper trunk of the shoulder girdle
and the pelvic girdle (6 points). The reliability of the
TIS for stroke patients is ICC = .96 (Verheyden et al.,
2004).
Statistical Analysis
Statistical analysis was performed using SPSS, version
16.0 (IBM, Armonk, NY). Descriptive statistics were
used to analyze the general characteristics of the partici-
pants. The independent t test or chi-squared test was
used to compare differences in the general characteristics
and functional level between the two groups. The cut-off
values of DGI-8 and DGI-4 for predicting falls was
determined by using receiver operating characteristic
(ROC) curves. To investigate the relationship between
the cut-off values of DGI-8 and DGI-4 depending on
the fall experience, the chi-square test was used. In addi-
tion, positive predictive values (PPVs) and negative pre-
dictive values (NPVs) were calculated to evaluate the
accuracy for diagnosis. The accuracy of prediction was
measured by the area under the ROC curve (AUC),
which was classified into less informative (AUC = 0.5),
less accurate (0.5 < AUC ≤ 0.7), middle level of accuracy
(0.7 < AUC ≤ 0.9), very accurate (0.9 < AUC < 1), and
perfect accuracy (AUC = 1) (Greiner, Pfeiffer, & Smith,
2000). The subjects were divided into two groups if the
fall prediction values calculated using the cut-off values
for the DGI-8 and DGI-4 were significant. The indepen-
dent t-test was conducted to check for differences in the
functional ability (POMA-Gait subscale, 10mWT, STS,
FM-LE, TIS). Moreover, logistic regression analysis was
used for the validity of fall prediction. Odds ratios
(ORs) were calculated for forward walking. The statisti-
cal significance level was a = .05.
4 Rehabilitation Nursing 2016, 0, 1–10
S.-H. An et al.
Results
Subject Characteristics by Falling Status
The participants were divided into two groups, a group
with a history of falls (faller group) and a group without
a history of falls (nonfaller group). Among general and
medical characteristics, gender, age, diagnosis, type, dis-
ease duration, and use of a gait aid showed no significant
differences between the two groups. However, the faller
group had a significantly lower score than the nonfaller
group on the DGI-8 (p < .001), DGI-4 (p < .001),
POMA-Gait subscale (p < .001), 10mWT (p < .05), STS
(p < .01), FM-LE (p < .05), and TIS (p < .01) (Table 1).
Cut-off Value of the DGI-8 and DGI-4 for Predicting
Falls
Based on the ROC curves, the cut-off value for predicting
falls using the DGI-8 was 6.5 (sensitivity, 60%; specificity,
72%), with an AUC of 0.78 (95% CI: 0.67–0.90,
p < .001). The cut-off value for DGI-4 was 9.5 (sensitiv-
ity, 68%; specificity, 59%), with an AUC of 0.77 (95%
CI: 0.65–0.89, p < .001). The PPVs were 63% and 57%
for DGI-8 and DGI-4 in regard to a positive diagnosis
(≤16.5 and ≤9.5, respectively), whereas the NPVs were
70% and 70% with respect to a negative diagnosis (>16.5
and >9.5, respectively) (Table 2).
Comparison of Functional Ability Depending on the
Cut-off Values of the DGI-8 and DGI-4
The POMA-Gait subscale, 10mWT, STS, FM-LE, and TIS
showed significant differences between the two groups in
the comparison of functional ability depending on the
cut-off values of DGI-8 and DGI-4 for falls (p < .01)
(Table 3).
Analysis of Factors Affecting the Experience of Falls
In Model A, the DGI-8 showed the most significant effect
on the fall experience (OR = 0.74). In Model B, the
POMA-Gait subscale had the most significant effect
(OR = 0.50). Moreover, stroke survivors with a DGI-8
score of ≤16.5 had a 3.8-fold higher probability of falling
compared to those with a DGI-8 score of >16.5. In addi-
tion, stroke survivors with a DGI-4 score of ≤9.5 had a
3.1-fold higher probability of falling relative to those with
a DGI-4 score of >9.5 (Table 4).
© 2016 Association of Rehabilitation Nurses
S.-H. An et al. Validity of the 4- and 8-Item Dynamic Gait Index in Stroke Survivors

Table 1 Subject characteristics by faller status

All Subject (N = 57) Fallers (N = 25) NonFallers (N = 32) t/v2 p

Gender, male/female (n) 32 (56.1%)/25 (43.9%) 13 (52%)/12 (48%) 19 (59.38%)/13 (40.62%) .310 .602
Age (years) 52  15.48 53.72  16.18 50.66  15.03 .739 .463
Etiology, infarction/hemorrhage (n) 39 (68.4%)/8 (31.6%) 20 (80%)/5 (20%) 19 (59.38%)/13 (40.62%) 2.763 .151
Affected side, left/right (n) 30 (52.6%)/27 (47.4%) 15 (60%)/10 (40%) 15 (46.88%)/17 (53.12%) .970 .425
Stroke duration (month) 9.04  3.02 8.44  2.89 9.50  3.08 1.325 .191
Walking aids type, 25 (43.9%)/15 8 (32%)/7 17 (53.12%)/8 3.022 .221
independent/one-point (26.3%)/17 (29.8%) (28%)/10 (40%) (25%)/7 (21.88%)
cane/four-point cane (n)
DGI-8-item (score) 15.56  4.69 12.76  4.88 17.75  3.18 4.662 .001***
DGI-4-item (score) 8.79  2.53 7.28  2.84 9.97  1.45 4.651 .001***
POMA-Gait subscale (score) 8.81  1.89 7.76  1.48 9.63  1.79 4.202 .001***
10mWT (m/s) .77  .32 .65  .26 .86  .34 2.515 .015*
STS (sec) 15.79  7.40 18.69  7.29 13.53  6.75 2.768 .008**
FM-LE (score) 22.00  6.75 19.96  5.88 23.59  7.03 2.076 .043*
TIS (score) 16.65  4.23 14.76  4.18 18.13  3.71 3.218 .002**

The values are presented as mean (SD) or frequency (%).

DGI, Dynamic Gait Index; POMA-Gait scale, Gait subscale in Performance-Oriented Mobility Assessment; 10mWT, 10-meter Walk
Test; STS, Sit-to-Stand Test; FM-LE, Fugl Meyer-Lower Extremity; TIS, Trunk Impairment Scale.
Significant differences between two groups were presented as *p < .05, **p < .01, ***p < .001.

Table 2 Cut-off value of the DGI-8 item and DGI-4 item as the factor for predicting falls

Sensitivity PPV
Variable Cut-off value AUC 95% CI Specificity NPV p

DGI-8-item ≤16.5 score .78 .67–.90 15/25 (60%) 15/24 (63%) .001*
23/32 (72%) 23/33 (70%)
DGI-4-item ≤9.5 score .77 .65–.89 17/25 (68%) 17/30 (57%) .001*
19/32 (59%) 19/27 (70%)

AUC, area under the curve; CI, confidential interval; PPV, positive predictive value; NPV, negative predictive value.
Significant differences were presented as *p < .001.

4. For both DGI-8 and DGI-4, the subjects in the faller

Discussion
Falls risk prediction should be performed using a valid tool
on admission to hospital. A plan of care should be initi-
ated for all those identified as at risk of falls. Examples of
suggested falls risk screening tools used in the hospital set-
ting include the STRATIFY, a 5-item tool (Coker & Oliver,
2003); the Conley scale, a 6-item scale (Conley, Schultz, &
Selvin, 1999); and the Morse Scale, a 6-item scale (Morse,
Morse, & Tylko, 1989). Each of these has been shown to
have intermediate to high accuracy in the identification of
patients or residents at risk of falls. However, in some
instances, these tools have been shown to classify the
majority of patients or residents as being at high risk, thus
limiting the usefulness of the screening process. Hence, in
this study, we attempted to investigate the validity of pre-
dicting falls in stroke survivors using the DGI-8 and DGI-
group (scores of 12.76 and 7.28, respectively) had lower
scores compared to those in the nonfaller group (scores of
17.75 and 9.97, respectively). However, in a setting where
the majority of people were considered to be at increased
risk of falling (e.g., in some residential care settings), a
screening process might be redundant.
Analysis of Cut-Off Value of the DGI-8 and DGI-4 for
Predicting Falls
In one previous study, the faller group had a score of
15.3, as assessed with the DGI-8, while the nonfaller
group had a score of 18.6 (Cattaneo et al., 2006). More-
over, Forsberg, Andreasson, and Nilsag
ard (2013)
reported there were significant differences in the DGI
scores between the faller (score of 13.3) and nonfaller

© 2016 Association of Rehabilitation Nurses

Rehabilitation Nursing 2016, 0, 1–10 5
Validity of the 4- and 8-Item Dynamic Gait Index in Stroke Survivors S.-H. An et al.

Table 3 Comparison of functional ability depending on the cut-off values of the DGI-8-item and DGI-4-item

Type

Variables DGI-8-item (≤16.5 score, n = 24) DGI-8-item (>16.5 score, n = 33) t p

POMA-Gait subscale (score) 7.33  1.20 9.88  1.56 6.684 .001**

10mWT (m/s) 0.62  0.27 0.88  0.32 3.140 .003*
STS (sec) 19.53  7.33 13.07  6.25 3.581 .001**
FM-LE (score) 18.42  4.61 24.61  6.91 3.809 .001**
TIS (score) 14.13  3.83 18.48  3.55 4.434 .001**
DGI-4-item (≤9.5 score, n = 30) DGI-4-item (>9.5 score, n = 27)
POMA-Gait subscale (score) 7.90  1.61 9.81  1.69 4.389 .001**
10mWT (m/s) 0.65  0.28 0.91  0.32 3.251 .002*
STS (sec) 18.25  7.67 13.06  6.12 2.806 .007*
FM-LE (score) 19.00  5.65 25.33  6.37 3.979 .001**
TIS-total (score) 14.77  4.26 18.74  3.12 3.982 .001**

POMA-Gait subscale, Gait subscale in Performance-Oriented Mobility Assessment; DGI, Dynamic Gait Index; 10mWT, 10-meter
Walk Test; STS, Sit-to-Stand Test; FM-LE: Fugl Meyer-Lower Extremity; TIS, Trunk Impairment Scale.
Significant differences were presented as *p < .01, **p < .001.

Table 4 Analysis of factors affecting the experience of falls

Type Independent Variables Regression Coefficient Standard Error Wald p Odd Ratio (95% CI)

Model A DGI-8-item .301 .088 11.70 .001** .740 (.623–.880)

Model B POMA-Gait subscale .698 .215 10.50 .001** .498 (.326–.759)
DGI-8-item 1.344 .567 5.62 .018* 3.833 (1.262–11.642)
DGI-4-item 1.133 .560 4.10 .043* 3.106 (1.037–9.304)

Model A: Independent variables: gender, age, etiology, affected side, stoke duration, walking aids type, POMA-Gait subscale,
10mWT, STS, FM-LE, TIS, DGI-8-item, DGI-4-item. Dependent variable: fall (nonfallers = 0, fallers = 1).
Model B: Dichotomous model in DGI-8-item and DGI-4-item; DGI (≤16.5 score = 0, >16.5 score = 1), DGI-4-item (≤9.5
score = 0, >9.5 score = 1). Independent variables: gender, age, etiology, affected side, stoke duration, walking aids type,
POMA-Gait subscale, 10mWT, STS, FM-LE, TIS, DGI-8-item, DGI-4-item. Dependent variable: fall (nonfallers = 0, fallers = 1).
DGI, Dynamic Gait Index; POMA-Gait subscale, Gait subscale in Performance-Oriented Mobility Assessment.
Significant differences were presented as *p < .05, **p < .001.

(score of 15.3) groups, as assessed with the DGI-4. In the
present study, the subjects showed significant differences
in the DGI-4 score based on their fall experience. The
results demonstrate the DGI-4 reflects functional ability
and can serve as a useful assessment tool for the predic-
tion of falls. In addition, based on previous studies on
the prediction of falls using DGI, patients with multiple
sclerosis and Parkinson’s disease were found to have
mean scores of 12 and 18.5, respectively (Cattaneo et al.,
2006; Landers et al., 2008). Healthy elderly people had a
score of 19 (Shumway-Cook et al., 1997).
A discriminative, valid, and reliable assessment tool
that can identify patients at risk of falling is very impor-
tant in setting goals and planning a care program for
stroke survivors. Thus, rehabilitation professionals such as
physical therapists, occupational therapists, and nurses
can play an important role in enhancing the
understanding of stroke survivors and their families
regarding the possibilities of falling down while perform-
ing gait activities by using the DGI cut-off value.
Marchetti et al. (2006) conducted a study on predict-
ing falls in patients with balance deficits and vestibular
dysfunction using the DGI-8 and DGI-4. The results con-
firmed these two assessment tools have similar character-
istics, with an AUC of 0.67. The cut-off values for the
DGI-8 and DGI-4 in predicting falls were reported to be
19 (sensitivity, 68%; specificity, 60%) and 9 (sensitivity,
68%; specificity, 59%), respectively, with the AUC at the
acceptable level of 0.77–0.78. The average gait speed of
the subjects was 0.77 m/s, whereas the gait speeds of
patients with limited-community ambulation ranged from
0.4 to 0.8 m/s (Perry, Garrett, Gronley, & Mulroy, 1995).
Hence, the average speed was sufficient for performing
the task of walking on a level surface. Among the four

© 2016 Association of Rehabilitation Nurses

6 Rehabilitation Nursing 2016, 0, 1–10
S.-H. An et al.
items of the DGI-4, the change in gait speed is a dual
task in which patients need to cope with the gait speed
demanded by the assessor. This effort to maintain the
expected gait speed may cause a bottleneck phenomenon
and interfere with the use of the same nerve pathway
(Pashler, 1994), resulting in staggering phenomena such
as foot drag. Performing simultaneous balance mainte-
nance during gait activity (e.g., preventing the foot from
dragging while maintaining gait speed) makes it more dif-
ficult for stroke survivors to walk (Ng, Ng, Lee, Ng, &
Tong, 2012).
Analysis of Functional Ability Depending on the Cut-Off
Values of the DGI-8 and DGI-4
The comparison of functional ability for the different
DGI-8 and DGI-4 cut-off values showed significant differ-
ences in the functional ability (POMA-Gait subscale,
10mWT, STS, FM-LE, and TIS). When the POMA-Gait
subscales of the subjects with a DGI-8 score of ≤16.5 and
those with a DGI-4 score of ≤9.5 were compared, the
scores were 7.33 and 7.9, respectively. However, the com-
parison of the two other groups (i.e., those with a DGI-8
score of >16.5 and those with a DGI-4 score of >9.5)
revealed scores of 9.88 and 9.81, respectively. The cut-off
values of the POMA-Gait subscale for predicting falls is
known to be less than 10.5 (faller group, average of 5.8;
nonfaller group, average of 10.3). The above results indi-
cate the group with a DGI-8 score of ≤16.5 and those
with a DGI-4 score of ≤9.5 have a higher risk of falls
compared to the other groups. Moreover, the results of
the 10mWT for the group with a DGI-8 score of ≤16.5
and those with a DGI-4 score of ≤9.5 were 0.62 m/s and
0.65 m/s, respectively; these groups are classified as hav-
ing limited-community ambulation (0.4–0.8 m/s). The
converted gait speeds are 1.3 mph (2.23 km/h) and 1.45
mph (2.34 km/h), respectively, when the intensities of
physical activity are 2.76 MET and 2.86 MET, respectively
(Ainsworth et al., 2011).
According to standard levels for healthy adults, a light
walk in the garden corresponds to 2.3 MET; activities of
daily living (e.g., preparing meals or cleaning the house),
to 2.5 MET. Based on this standard, the stroke survivors
with a DGI-8 score of ≤16.5 or DGI-4 score of ≤9.5
belong to the limited-community ambulation group, and
even a light walk in the garden is impossible. These stroke
survivors are heavily restricted in their physical activity,
which results in a higher possibility of being exposed to
the risk of falling. The STS results for the subjects with a
© 2016 Association of Rehabilitation Nurses
Rehabilitation Nursing 2016, 0, 1–10
Validity of the 4- and 8-Item Dynamic Gait Index in Stroke Survivors
DGI-8 score of ≤16.5 and those with a DGI-4 score of
≤9.5 were 19.53 s and 18.25 s, respectively. Meanwhile,
the STS results for the subjects with a DGI score of >16.5
and those with a DGI-4 score of >9.5 were 13.07 s and
13.06 s, respectively. These results imply the group with a
DGI-8 score of ≤16.5 and those with a DGI-4 score of
≤9.5 have a higher risk of falls compared to the other
groups.
The average FM-LE score in this previous study was
18.7 and differed from the average score of 22 in this
study. Belgen, Beninato, Sullivan, and Narielwalla (2006)
reported the FM-LE score of a chronic stroke survivor to
be 23.8, suggesting that this assessment tool has a low
discrimination for predicting falls. The nonfaller group
had a score of 24.3, while the faller group (those with
one or more falls) had a score of 22.9; this difference was
not statistically significant. However, the FM-LE scores of
the subjects with a DGI-8 score of ≤16.5 and those with a
DGI-4 score of ≤9.5 were 18.42 and 19, respectively.
Moreover, the FM-LE scores of the subjects with a DGI-8
score of >16.5 and those with a DGI-4 score of >9.5 were
24.61 and 25.33, respectively; this difference was statisti-
cally significant. According to Verheyden et al. (2005) the
TIS score of <20 implies the person is incapable of per-
forming independent daily living activities and controlling
the trunk. Most of the subjects in this study had a TIS
score of <20. In particular, the subjects with a DGI-8
score of ≤16.5 and those with a DGI-4 score of ≤9.5 had
TIS scores of 14.13 and 14.17, respectively. Furthermore,
the subjects with a DGI-8 score of >16.5 and those with a
DGI-4 score of >9.5 had TIS scores of 18.43 and 18.74,
respectively, suggesting a higher exposure to the risk of
falls not only during gait activity but also during bed
mobility and transfers. Collectively, these results suggest
DGI-8 and DGI-4 can reflect functional ability in stroke
survivors and serve as useful examination tools.
Limitations and Summary
Although nursing staff routinely monitor the general
health status of patients, it is also important to know
how much function patients have recovered (e.g., level of
independence, and level of professional assistance
needed), and whether patients are at risk for falls. In par-
ticular, patients with neurological problems such as
stroke, have a relatively high risk of falling. Although
caregivers try to avoid patient falls while maneuvering the
patient in and out of bed, transferring to the toilet, etc.,
falls often occur. Therefore, it is important to have fall
7
Validity of the 4- and 8-Item Dynamic Gait Index in Stroke Survivors
prevention programs in place, and for nursing staff to
identify predictors of fall risk.
DGI can be used to predict a fall that may occur
while walking. The present results support the use of
the DGI for fall prediction of stroke survivors. In addi-
tion, the results for the validity of fall prediction using
the cut-off values of the DGI-8 and DGI-4 showed that
the POMA-Gait subscale had the highest effect. The
probability that stroke survivors with a DGI score of
≤16.5 will experience a fall is 3.8 times higher than that
of stroke survivors with a DGI score of >16.5. Further,
the probability that stroke survivors with a DGI-4 score
of ≤9.5 will experience a fall is 3.1 times higher than
that of stroke survivors with a DGI-4 score of >9.5. Gait
variability has a significant correlation with an increased
risk of falling and with a balance deficit when walking
(Verma, Arya, Sharma, & Garg, 2012). The POMA-Gait
subscale can identify gait asymmetry in stroke survivors,
and can assess step length in relation to gait speed, as
well as balance ability in relation to postural sway, gait
initiation, and base of support when walking (Canbek,
Fulk, Nof, & Echternach, 2013). The two examination
tools were demonstrated to be valid for predicting falls
in stroke survivors.
In particular, the DGI-4 includes four items such as
gait on level surface, gait with speed changes, horizontal
head turns while walking, and vertical head turns while
walking. This tool can be used without special equipment
or device, and it does not include tasks that may increase
risk of falls (walking around obstacles or climbing stairs).
In addition, the time for evaluation using this tool is less
than five minutes. These points show that DGI-4 can be
used to predict or identify fall risk for patients under a
nurse’s care, and this is supported in the results of the
study. In other words, at nursing care centers, rehabilita-
tion nurses, physical therapists, and occupational thera-
pists can predict or identify the risk of falling in stroke
patients, and through that, qualified care of patients
would be possible.
A limitation of this study is that it retrospectively eval-
uated the use of DGI-8 and DGI-4 in predicting falls in
stroke survivors. In addition, this might lead to the use of
a convenience sample, which may affect the generalizabil-
ity of the findings. However, this study has significance in
suggesting that these two examination tools can be used
to predict the risk of falls. The subjects were patients able
to perform functional ambulation. Patients with consider-
ably low gait ability, an age of >50, severe disability, or
the inability to follow instructions were excluded, since
8 Rehabilitation Nursing 2016, 0, 1–10
S.-H. An et al.
Key Practice Points
 The DGI-8 and DGI-4 items may be valid for predicting
falls in stroke survivors.
 The DGI-8 and DGI-4 items may reflect functional ability.
 The rehabilitation professionals can play an important
role in enhancing the understanding of stroke survivors
and their families regarding the possibilities of falling
down while performing gait activities by using the DGI
cut-off value.
 Future prospective studies are needed to confirm the
validity of the DGI-8 and DGI-4 items in predicting falls
they might have shown a floor effect (i.e., poor discrimi-
native ability). Therefore, the results of this study can
only be generalized within similar age and functional
groups. Further, the number of falls was recorded
through the patient’s subjective reporting, which may
deviate from the standard for defining falls. However, the
results of this study indicate that both DGI-8 and DGI-4
were able to identify dynamic balance and functional abil-
ity in stroke survivors during walking and have validity in
predicting falls.
Conclusion
In this study, we confirm that the DGI-8 and DGI-4 were
valid for predicting falls and also reflected functional
ability. In particular, DGI-4 can be used without any spe-
cial equipment or device, and includes tasks that may
not increase the risk of falls. In addition, the time for
evaluation using the tool is less than five minutes. Thus,
professionals such as rehabilitation nurses, physical thera-
pists, and occupational therapists can use these tools to
better predict the risk of falling for stroke patients, allow-
ing better quality of care of patients. In the future, fur-
ther prospective study for prediction of falling will be
needed.
Conflict of Interest
The authors declare no conflict of interest.
References
Ainsworth, B.E., Haskell, W.L., Herrmann, S.D., Meckes, N.,
Bassett, D.R., Jr, Tudor Locke, C., . . . Leon, A.S. (2011).
2011 compendium of physical activities: A second update of
© 2016 Association of Rehabilitation Nurses
S.-H. An et al.
codes and MET values. Medicine and Science in Sports and
Exercise, 3(8), 1575–1581.
Belgen, B., Beninato, M., Sullivan, P.E., & Narielwalla, K.
(2006). The association of balance capacity and falls
self-efficacy with history of falling in community-dwelling
people with chronic stroke. Archives of Physical Medicine
and Rehabilitation, 87(4), 554–561.
Cakar, E., Durmus, O., Tekin, L., Dincer, U., & Kiralp, M.Z.
(2010). The ankle-foot orthosis improves balance and reduces
fall risk of chronic spastic hemiparetic patients. European
Journal of Physical and Rehabilitation Medicine, 46(3), 363–368.
Canbek, J., Fulk, G., Nof, L., & Echternach, J. (2013). Test-
retest reliability and construct validity of the Tinetti
performance-oriented mobility assessment in people with
stroke. Journal of Neurologic Physical Therapy, 37(1), 14–19.
Cattaneo, D., Regola, A., & Meotti, M. (2006). Validity of six
balance disorders scales in persons with multiple sclerosis.
Disability and Rehabilitation, 28(12), 789–795.
Cho, K., Yu, J., & Rhee, H. (2015). Risk factors related to
falling in stroke patients: A cross-sectional study. Journal of
Physical Therapy Science, 27(6), 1751–1753.
Coker, E., & Oliver, D. (2003). Evaluation of the STRATIFY
falls prediction tool on a geriatric unit. Outcomes
Management, 7(1), 8–14.
Conley, D., Schultz, A., & Selvin, R. (1999). The challenge of
predicting patients at risk of falling: Development of the
Conley Scale. Medsurg Nursing, 8(6), 348–354.
Flansbjer, U.B., Holmback, A.M., Downham, D., Patten, C., &
Lexell, J. (2005). Reliability of gait performance tests in men
and women with hemiparesis after stroke. Journal of
Rehabilitation Medicine, 37(2), 75–82.
Forsberg, A., Andreasson, M., & Nilsag
ard, Y.E. (2013).
Validity of the dynamic gait index in people with multiple
sclerosis. Physical Therapy, 93(10), 1369–1376.
Fugl-Meyer, A.R., Jaasko, L., Leyman, I., Olsson, S., &
Steglind, S. (1975). The post-stroke hemiplegic patient. 1. A
method for evaluation of physical performance.
Scandinavian Journal of Rehabilitation Medicine, 7(1), 13–31.
Greiner, M., Pfeiffer, D., & Smith, R.D. (2000). Principles and
practical application of the receiver-operating characteristic
analysis for diagnostic tests. Preventive Veterinary Medicine,
45(1-2), 23–41.
Hollands, K.L., Agnihotri, D., & Tyson, S.F. (2014). Effects of
dual task on turning ability in stroke survivors and older
adults. Gait & Posture, 40(4), 564–569.
Jonsdottir, J., & Cattaneo, D. (2007). Reliability and validity of
the Dynamic Gait Index in persons with chronic stroke.
Archives of Physical Medicine and Rehabilitation, 88(11),
1410–1415.
Lamb, S.E., Jorstad-Stein, E.C., Hauer, K., & Becker, C.
(2005). Development of a common outcome data set for fall
© 2016 Association of Rehabilitation Nurses
Rehabilitation Nursing 2016, 0, 1–10
Validity of the 4- and 8-Item Dynamic Gait Index in Stroke Survivors
injury prevention trials: The Prevention of Falls Network
Europe consensus. Journal of the American Geriatrics Society,
53(9), 1618–1622.
Landers, M.R., Backlund, A., Davenport, J., Fortune, J.,
Schuerman, S., & Altenburger, P. (2008). Postural instability
in idiopathic Parkinson’s disease: Discriminating fallers
from nonfallers based on standardized clinical measures.
Journal of Neurologic Physical Therapy, 32(2), 56–61.
Lim, J.Y., Jung, S.H., Kim, W.S., & Paik, N.J. (2012).
Incidence and risk factors of poststroke falls after discharge
from inpatient rehabilitation. PM & R, 4(12), 945–953.
Lin, J.H., Hsu, M.J., Hsu, H.W., Wu, H.C., & Hsieh, C.L.
(2010). Psychometric comparisons of 3 functional
ambulation measures for patients with stroke. Stroke, 41(9),
2021–2025.
Marchetti, G.F., & Whitney, S.L. (2006). Construction and
validation of the 4-item dynamic gait index. Physical
Therapy, 86(12), 1651–1660.
Mong, Y., Teo, T.W., & Ng, S.S. (2010). 5-repetition sit-to-
stand test in subjects with chronic stroke: Reliability and
validity. Archives of Physical Medicine and Rehabilitation, 91
(3), 407–413.
Morse, J., Morse, R., & Tylko, S. (1989). Development of a
scale to identify the fall-prone patient. Canadian Journal on
Aging, 8, 866–876.
Ng, S.S., Chan, L.H., Chan, C.S., Lai, S.H., Wu, W.W., Tse,
M.M., & Fong, S.S. (2015). Parallel walk test: Its correlation
with balance and motor functions in people with chronic
stroke. Archives of Physical Medicine and Rehabilitation, 96
(5), 877–884.
Ng, S.S., Ng, P.C., Lee, C.Y., Ng, E.S., & Tong, M.H. (2012).
Walkway lengths for measuring walking speed in stroke
rehabilitation. Journal of Rehabilitation Medicine, 44(1), 43–46.
Pashler, H. (1994). Dual-task interference in simple tasks: Data
and theory. Psychological Bulletin, 116(2), 220–244.
Perry, J., Garrett, M., Gronley, J.K., & Mulroy, S.J. (1995).
Classification of walking handicap in the stroke population.
Stroke, 26(6), 982–989.
Sanford, J., Moreland, J., Swanson, L.R., Stratford, P.W., &
Gowland, C. (1993). Reliability of the Fugl Meyer
assessment for testing motor performance in patients
following stroke. Physical Therapy, 73(7), 447–454.
Shumway-Cook, A., Gruber, W., Baldwin, M., & Liao, S.
(1997). The effect of multidimensional exercises on balance,
mobility, and fall risk in community-dwelling older adults.
Physical Therapy, 77(1), 46–57.
Shumway-Cook, A., & Woollacott, M. (1995). Motor control:
Theory and practical applications. Baltimore: Williams &
Wilkins.
Sterke, C.S., Huisman, S.L., van Beeck, E.F., Looman, C.W., &
Van der Cammen, T.J. (2010). Is the Tinetti Performance
9
Validity of the 4- and 8-Item Dynamic Gait Index in Stroke Survivors
Oriented Mobility Assessment (POMA) a feasible and valid
predictor of short-term fall risk in nursing home residents
with dementia? International Psychogeriatrics, 22(2), 254–
263.
Tinetti, M.E. (1986). Performance-oriented assessment of
mobility problems in elderly patients. Journal of the
American Geriatrics Society, 34(2), 119–126.
Verheyden, G., Nieuwboer, A., Feys, H., Thijs, V., Vaes, K., &
De Weerdt, W. (2005). Discriminant ability of the Trunk
Impairment Scale: A comparison between stroke patients
and healthy individuals. Disability and Rehabilitation, 27
(17), 1023–1028.
Verheyden, G., Nieuwboer, A., Mertin, J., Preger, R., Kiekens,
C., & De Weerdt, W. (2004). The Trunk Impairment Scale:
10
S.-H. An et al.
A new tool to measure motor impairment of the trunk after
stroke. Clinical Rehabilitation, 18(3), 326–334.
Verma, R., Arya, K., Sharma, P., & Garg, R.K. (2012).
Understanding gait control in post-stroke: Implications for
management. Journal of Bodywork and Movement Therapies,
16(1), 14–21.
Weerdesteyn, V., de Niet, M., van Duijnhoven, H.J., &
Geurts, A.C. (2008). Falls in individuals with stroke.
Journal of Rehabilitation Research and Development, 45(8),
1195–1213.
Wrisley, D. M., Walker, M. L., Echtemach, J. L., & Strasnick,
B. (2003). Reliability of the dynamic gait index in people
with vestibular disorders. Archives of Physical Medicine and
Rehabilitation, 84(10), 1528–1533.
© 2016 Association of Rehabilitation Nurses
Rehabilitation Nursing 2016, 0, 1–10